Data reports

Report guide

What is in each Clearilytics report download, what every column means, how to use it, the methodology behind it and the questions we are asked most.

Describes the Q3 2026 release

Is my segment covered? · All reports

Developer Market Report

Asking prices, how fast stock clears, and what each extra bedroom earns

Latest release and free sample

What is in the download

Each release downloads as a zip file holding the Excel workbook and a one-page Executive Summary (PDF). The workbook has 9 sheets, in tab order.

Overview: The release on one page: the headline figures, how much to trust them, the main findings and coverage by state.
Market: Every segment: price band, supply flow, how fast it clears, and its price and speed position.
Insights: Price and speed position (price against speed), what the marginal bedroom earns and the request mix.
Demand: What people ask to rent and to buy, at what budget, against what is advertised.
Land: Asking land rates per sqm and per plot, published rows only.
Scheme test: Where your price sits in a segment's asking band, and what holding the stock costs at your finance rate.
Bulk scheme test: The Scheme test for up to 50 schemes at once, with the same three inputs on each row.
Changes: What moved since Q2 2026: published segments, comparables, segments in and out.
Guide: Appraisal steps, column meanings, method, limits and checks.

How to use it

The appraisal, in five steps

Find what you could build. Filter Market to your state and area, then to the types and bedroom counts your site can take. Prefer neighbourhood rows; LGA rows are a separate, wider population.
Read the median, then the band. P25–P75 is the middle half of the market. A scheme that only works above P75 relies on being the most expensive stock in the segment.
Days to clear and holding cost. Days to clear sets the holding cost. Use Scheme test to price the carry at your own price and finance rate, or Bulk scheme test for up to 50 schemes at once.
Size units off the bedroom step. Insights shows what each extra bedroom earns and costs in days. Check the comparison count.
Check what is missing. A segment absent from Market did not reach the minimum comparables. Building against it means building without price evidence.

Columns and what they mean

Headers are exactly as printed in the workbook. 50 columns across 3 grids, and 12 hidden helper columns (the workbook's lookup sheets read them; unhide them in Excel to see them).

Market: price band, supply flow and clearance speed per segment

Segment

Grain: Neighbourhood = that neighbourhood's own comparables. LGA = the whole LGA's comparables. Separate populations, not a roll-up: nothing has been substituted for anything.
State: Where the segment is, from broad to narrow. Area is the local district people use; All neighbourhoods means the row pools every neighbourhood of the area.
Area: Where the segment is, from broad to narrow. Area is the local district people use; All neighbourhoods means the row pools every neighbourhood of the area.
Neighbourhood: Where the segment is, from broad to narrow. Area is the local district people use; All neighbourhoods means the row pools every neighbourhood of the area.
Property type: The kind of building, its bedroom count, and new-build or existing stock. Blocks of flats and self-contained rooms are not compiled on Market.
Beds (bedrooms): The kind of building, its bedroom count, and new-build or existing stock. Blocks of flats and self-contained rooms are not compiled on Market.
Vintage: The kind of building, its bedroom count, and new-build or existing stock. Blocks of flats and self-contained rooms are not compiled on Market.

Asking price band

Band P25 (naira): The middle asking price for similar properties here and the middle half of asking prices, after outliers are removed. Measured over the price quarter.
Median asking (naira): The middle asking price for similar properties here and the middle half of asking prices, after outliers are removed. Measured over the price quarter.
Band P75 (naira): The middle asking price for similar properties here and the middle half of asking prices, after outliers are removed. Measured over the price quarter.
Band width (percent): (P75 minus P25) divided by the median. Low means the market broadly agrees on price; high means it does not.

Supply and speed

Comparables (qtr) (listings): The asking comparables first advertised in the segment during the quarter: supply flow, not stock standing.
Days to clear (median) (days): Typical days from a listing appearing to it leaving the portals; listings still up are counted properly (Kaplan-Meier), not dropped. Leaving the portals is not proof of a sale. Reads under 8 days or 180+ at the limits of what is measurable.
Cleared by 180 days (percent): The share of the stock that had gone within the watch window: the fairest like-for-like speed comparison, because the population and the window are both fixed.
Speed grade (grade letter): How much to trust the speed figures, A to D, by the tracked listings behind them (A is 100 or more).

Price and speed position

Position: Higher or lower price, faster or slower, against the state's own medians over the full release. Blank where no speed is measured, the type is vague, or the state has too few segments to split.

Flags and gaps

UNIFORM: Yes or No, each its own filter. UNIFORM: the band is narrower than 5% of the median, so the comparables ask almost the same price. WIDE: the band is wider than 150% of the median, or P75 is at least 4 times P25: prices disagree.
WIDE: Yes or No, each its own filter. UNIFORM: the band is narrower than 5% of the median, so the comparables ask almost the same price. WIDE: the band is wider than 150% of the median, or P75 is at least 4 times P25: prices disagree.
VAGUE (yes or no): Yes where the listings did not say what kind of building it is, so the segment's slowness reflects listing quality, not liquidity. Kept on Market; left out of Position and Insights.
CLEAN: Yes when none of UNIFORM, WIDE or VAGUE applies. Blue is used for Yes here; orange marks the other three. The word is always in the cell.
Why blank: The reason for a blank speed or position: too few tracked listings, a curve that did not reach halfway, a vague listing type, or a state with too few segments to split. Empty when nothing is blank.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Scheme test picks by. One per row, and the key the Scheme test reads the grid with.
LGA: A helper kept out of the grid to keep it readable. The Scheme test reads it.
Fast quarter (days): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Slow quarter (days): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Tracked listings (listings): A helper kept out of the grid to keep it readable. The Scheme test reads it.

Demand: what people ask for to rent and to buy, against what is advertised

Where and what

Grain: How wide a net the row casts. Area + bedrooms: one property type at one bedroom count in one area, the strongest form. Area, all sizes and State, all areas: pooled rows, shown only where the finer split had too few requests to publish.
State: What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.
Area: What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.
Market: What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.
Property type: What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.
Beds (bedrooms): What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.

What is wanted

Requests (requests): How many requests were posted, and how many named a budget this pack can use (inside the market's own price band). A request is a stated intention posted on a portal, not a transaction.
With a budget (requests): How many requests were posted, and how many named a budget this pack can use (inside the market's own price band). A request is a stated intention posted on a portal, not a transaction.
Money basis: The unit of the money columns on the row, in words: Annual rent on a Rent row, Purchase price on a Sale row.
Budget P25 (naira): The stated budgets behind the median. The median needs 5 usable budgets and the quartiles 8; blank below that.
Median budget (naira): The stated budgets behind the median. The median needs 5 usable budgets and the quartiles 8; blank below that.
Budget P75 (naira): The stated budgets behind the median. The median needs 5 usable budgets and the quartiles 8; blank below that.

What is advertised

Median asking (naira): The middle advertised figure for the same area, type and size, and the asking comparables behind it. Blank means no advertised segment reached the minimum comparables, not that nothing is available.
Comparables (listings): The middle advertised figure for the same area, type and size, and the asking comparables behind it. Blank means no advertised segment reached the minimum comparables, not that nothing is available.

The gap

Budget vs asking (percent): Median budget against median asking, within one market. A gap between two medians over two different populations, not a discount anyone was offered. Marked at -25% or worse.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Scheme test picks by. One per row, and the key the Scheme test reads the grid with.

Land: asking rate per square metre by area, use class and plot size

Where and what

Grain: Where the rate applies. Area and Neighbourhood are separate populations, not a hierarchy; commercial and mixed-use land appear at area level only.
State: Where the rate applies. Area and Neighbourhood are separate populations, not a hierarchy; commercial and mixed-use land appear at area level only.
Area: Where the rate applies. Area and Neighbourhood are separate populations, not a hierarchy; commercial and mixed-use land appear at area level only.
Neighbourhood: Where the rate applies. Area and Neighbourhood are separate populations, not a hierarchy; commercial and mixed-use land appear at area level only.
Use class: Residential, mixed use or commercial land as the source classified it. Never pooled.
Plot size: Standard is under 2,000 square metres; Large is 2,000 and above. Larger plots trade at a lower rate per square metre, so the two are kept apart.

Rate

Band P25 / sqm (naira per square metre): The middle asking rate for land in the segment and the middle half of rates. Computed per listing and then taken as a median, so it does not move with the mix of plot sizes advertised.
Rate per sqm (naira per square metre): The middle asking rate for land in the segment and the middle half of rates. Computed per listing and then taken as a median, so it does not move with the mix of plot sizes advertised.
Band P75 / sqm (naira per square metre): The middle asking rate for land in the segment and the middle half of rates. Computed per listing and then taken as a median, so it does not move with the mix of plot sizes advertised.

Per plot

Plot standard (sqm) (square metres): The plot size this state actually trades in: the most common area among its asking comparables that describe themselves as one plot, and how many comparables sit at exactly that area. There is no national plot size. Each state's standard this release is in the table below.
Plot standard comparables: The plot size this state actually trades in: the most common area among its asking comparables that describe themselves as one plot, and how many comparables sit at exactly that area. There is no national plot size. Each state's standard this release is in the table below.
Value per plot (naira): The rate per square metre multiplied by that state's plot standard, because that is how the market quotes land. It carries no evidence of its own beyond the rate. Blank where the state has no measured standard — deliberately, rather than assuming a national figure.

Trust

Grade (grade letter): Graded on the comparables count and band width, then capped at B because title is unobserved. Land comparables is how many asking land comparables the rate rests on.
Land comparables (listings): Graded on the comparables count and band width, then capped at B because title is unobserved. Land comparables is how many asking land comparables the rate rests on.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Scheme test picks by. One per row, and the key the Scheme test reads the grid with.
Band P10 / sqm (naira per square metre): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Band P90 / sqm (naira per square metre): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Band width (percent): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Meets evidence rule (yes or no): A helper kept out of the grid to keep it readable. The Scheme test reads it.
Withheld because (Land): A helper kept out of the grid to keep it readable. The Scheme test reads it.

Frequently asked questions

The questions asked most

How do I use this for a residual land valuation?

The product supplies the revenue side: what similar stock is asking, how tightly priced the segment is and how long it takes to clear, plus what a site costs on Land. Build cost, finance, professional fees and your required profit are yours; the residual falls out of the combination. We do not compile construction costs.

Does "Days to clear" mean the property sold?

No. It measures how long the listing was visible, from the first time we saw it on a portal to the point it was no longer there. A listing leaving the portals is not a sale. The source records no sold status, so a departure is indistinguishable from a sale, a withdrawal, an expiry or a re-post, and every figure here is an UPPER BOUND on how fast stock actually sells.

What does the bedroom step tell me?

What one extra bedroom earns, and what it costs in extra days to clear, measured like for like: same neighbourhood, same property type, same vintage.

What does Position mean, and why is price not enough?

Position places a segment as higher or lower priced and faster or slower clearing against its own state's medians.

Is "Budget vs asking" a discount I can negotiate?

No. It compares two medians over two different populations: the people asking and the properties advertised are not the same set, and nobody was ever offered the gap. It is read within one market, a rent budget against an advertised rent and a purchase budget against an advertised price.

Why does Land list so few rows?

Because a rate needs 20 comparables and a normal band, and title is unobserved, which caps every grade.

What we do not answer

Certain questions are outside what a data supplier should answer. The consistent response is: here is what the data shows and why — what you do with it is your policy decision. This applies to:

Which valuation to accept
What LTV or tolerance to set
Whether to lend, build, or buy
Whether a specific scheme is viable
Which segment to invest in
Whether a particular valuer is right

Declining these is not unhelpfulness. It is the correct scope for a data provider, it avoids straying into advice we are not licensed to give, and it protects the client's own accountability for their decision.

Methodology

Source: Residential sale listings from Nigerian property portals, de-duplicated within each portal (a property on two portals can appear twice). Asking prices: Nigeria has no transaction register, so achieved prices are not observable. The Demand sheet adds a second feed, requirements posted by people looking to rent or buy, set against the rent or sale index.
Price: Tukey fence at 1.5 x IQR within each segment and period, bands computed after fencing. Minimum comparables: 20 in the quarter. The index itself publishes a stratum from 5 comparables; this product lists a segment from 20.
Speed: Kaplan-Meier survival over tracked listings entering in the 12 months to 2026-04-01, each watched 180 days; at least 20 tracked listings and a curve that reached halfway are needed.
Rank correlation: Spearman, computed at build time from the published rows that carry a measured speed and a specific property type, for LGA and neighbourhood rows separately. It is not a typed constant.
Scheme test and Bulk scheme test: Carry is your price x units x your finance rate x days / 365, worked for the fast quarter, the median clearance and the slow quarter of the segment. It is simple interest, a floor on the true holding cost; where a quarter reads 180+ days the true time is longer, so that carry is a floor too. Nothing is assumed: a result waits for the inputs it needs. The answer comes from the neighbourhood, then the LGA of the same area, never the state.
Land: A rate needs 20 comparables; grades are capped at B because title is unobserved. Only published rows are listed.

Limitations

Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales.
A listing leaving the portals is not a sale. The source records no sold status, so a departure is indistinguishable from a sale, a withdrawal, an expiry or a re-post, and every speed figure is an UPPER BOUND on how fast stock actually sells.
Price is Q3 2026; speed uses tracked listings entering up to 2026-04-01. The two periods on a row are not the same, and speed necessarily lags price.
Position and the bedroom step are presentation aggregations over the published rows of the full release, reproducible from Market. Position splits each state at its own medians over segments with a measured speed and a specific property type; a state with fewer than 4 such segments is not split.
Not here: build cost, absorption or months of supply (no sold status exists), planning or title risk. Not a valuation.
Percentages are stored as fractions (0.0698 = 6.98%) in the workbook.

Glossary

Band P25 / Median asking / Band P75: The middle asking price for similar properties here and the middle half of asking prices, after outliers are removed. Measured over the price quarter.
Band P25 / Rate per sqm / Band P75 (per sqm): The middle asking rate for land in the segment and the middle half of rates. Computed per listing and then taken as a median, so it does not move with the mix of plot sizes advertised.
Band width: (P75 minus P25) divided by the median. Low means the market broadly agrees on price; high means it does not.
Budget P25 / Median budget / Budget P75: The stated budgets behind the median. The median needs 5 usable budgets and the quartiles 8; blank below that.
Budget vs asking: Median budget against median asking, within one market. A gap between two medians over two different populations, not a discount anyone was offered. Marked at -25% or worse.
CLEAN: Yes when none of UNIFORM, WIDE or VAGUE applies. Blue is used for Yes here; orange marks the other three. The word is always in the cell.
Cleared by 180 days: The share of the stock that had gone within the watch window: the fairest like-for-like speed comparison, because the population and the window are both fixed.
Comparables (qtr): The asking comparables first advertised in the segment during the quarter: supply flow, not stock standing.
Days to clear (median): Typical days from a listing appearing to it leaving the portals; listings still up are counted properly (Kaplan-Meier), not dropped. Leaving the portals is not proof of a sale. Reads under 8 days or 180+ at the limits of what is measurable.
Grade / Land comparables: Graded on the comparables count and band width, then capped at B because title is unobserved. Land comparables is how many asking land comparables the rate rests on.
Grain: Neighbourhood = that neighbourhood's own comparables. LGA = the whole LGA's comparables. Separate populations, not a roll-up: nothing has been substituted for anything.
Grain: How wide a net the row casts. Area + bedrooms: one property type at one bedroom count in one area, the strongest form. Area, all sizes and State, all areas: pooled rows, shown only where the finer split had too few requests to publish.
Grain / State / Area / Neighbourhood: Where the rate applies. Area and Neighbourhood are separate populations, not a hierarchy; commercial and mixed-use land appear at area level only.
Median asking / Comparables: The middle advertised figure for the same area, type and size, and the asking comparables behind it. Blank means no advertised segment reached the minimum comparables, not that nothing is available.
Money basis: The unit of the money columns on the row, in words: Annual rent on a Rent row, Purchase price on a Sale row.
Plot size: Standard is under 2,000 square metres; Large is 2,000 and above. Larger plots trade at a lower rate per square metre, so the two are kept apart.
Plot standard (sqm) / Plot standard comparables: The plot size this state actually trades in: the most common area among its asking comparables that describe themselves as one plot, and how many comparables sit at exactly that area. There is no national plot size. Each state's standard this release is in the table below.
Position: Higher or lower price, faster or slower, against the state's own medians over the full release. Blank where no speed is measured, the type is vague, or the state has too few segments to split.
Property type / Beds / Vintage: The kind of building, its bedroom count, and new-build or existing stock. Blocks of flats and self-contained rooms are not compiled on Market.
Requests / With a budget: How many requests were posted, and how many named a budget this pack can use (inside the market's own price band). A request is a stated intention posted on a portal, not a transaction.
Speed grade: How much to trust the speed figures, A to D, by the tracked listings behind them (A is 100 or more).
State / Area / Market / Property type / Beds: What is asked for and where. Market is Rent or Sale and sets the unit of every money figure. All sizes and All areas mark a pooled row. Self-contained rooms appear at 0 beds.
State / Area / Neighbourhood: Where the segment is, from broad to narrow. Area is the local district people use; All neighbourhoods means the row pools every neighbourhood of the area.
UNIFORM / WIDE: Yes or No, each its own filter. UNIFORM: the band is narrower than 5% of the median, so the comparables ask almost the same price. WIDE: the band is wider than 150% of the median, or P75 is at least 4 times P25: prices disagree.
Use class: Residential, mixed use or commercial land as the source classified it. Never pooled.
VAGUE: Yes where the listings did not say what kind of building it is, so the segment's slowness reflects listing quality, not liquidity. Kept on Market; left out of Position and Insights.
Value per plot: The rate per square metre multiplied by that state's plot standard, because that is how the market quotes land. It carries no evidence of its own beyond the rate. Blank where the state has no measured standard — deliberately, rather than assuming a national figure.
Why blank: The reason for a blank speed or position: too few tracked listings, a curve that did not reach halfway, a vague listing type, or a state with too few segments to split. Empty when nothing is blank.

Investor Yield Report

See what a segment really returns, and whether the number can be trusted

Latest release and free sample

What is in the download

Each release downloads as a zip file holding the Excel workbook and a one-page Executive Summary (PDF). The workbook has 6 sheets, in tab order.

Overview: The release on one page: the headline figures, how much to trust them, the main findings and coverage by state.
Yield: Every segment: value, rent, gross yield, 12-month return, evidence and exit speed.
Lookup: Pick one segment and read its full record, flags, peer medians and a net-yield estimate.
Batch lookup: Up to 200 segments at once: the same record and net yield for every row.
Changes: What moved since Q2 2026: asking comparables, segments in and out, grade moves.
Guide: How to screen, what each column means, method, limits and checks.

How to use it

How to screen, in five steps

Filter before you sort. Set Meets evidence rule = Yes, then sort by yield. Meets evidence rule means grade A or B, normal band, measured at its own grain. Rows are ranked by yield within each grade, so filter on it first.
Read the Grain. Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area. The comparable counts on a row describe the area its Grain names.
Check both legs. A yield is one median divided by another. Both counts are shown: a thin rent count moves the yield as hard as a thin sale count, and it is the one people forget to check.
Price the exit. Days to clear is how long similar stock stays listed. A longer exit delays rent and can outweigh a small yield advantage. The slow-quarter figure is on the Lookup. A figure that reads 180+ is a lower bound on days to clear.
Compare with the alternatives. Gross yield is before agency fees, voids, management, tax and maintenance, and the Overview shows it against the T-bill yield. The 12-month Gross return vs T-bill column is gross too: before costs and tax. The Lookup's net-yield estimate and the Batch lookup take your own assumptions.

Columns and what they mean

Headers are exactly as printed in the workbook. 23 columns across 1 grid, and 9 hidden helper columns (the workbook's lookup sheets read them; unhide them in Excel to see them).

Yield: value, rent, gross yield, evidence and exit speed per segment

Segment

State: Where the segment is, from broad to narrow. Area is the local district people use.
Area: Where the segment is, from broad to narrow. Area is the local district people use.
Neighbourhood: Where the segment is, from broad to narrow. Area is the local district people use.
Property type: What kind of stock, how many bedrooms, and new-build or existing. These define the segment rather than describing one property.
Beds (bedrooms): What kind of stock, how many bedrooms, and new-build or existing. These define the segment rather than describing one property.
Vintage: What kind of stock, how many bedrooms, and new-build or existing. These define the segment rather than describing one property.

How much to trust it

Grade (grade letter): How much to trust the row, A to D: comparables count and price agreement. A: Many comparables, prices agree. B: Good comparables. C: Directional only. D: Wide spread: prices disagree. Read it before the yield.
Meets evidence rule (yes or no): Yes = grade A or B, normal band, measured at its own grain. The field that separates the best-evidenced rows from the rest.
Grain: Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area.

The return

Median asking (naira): Middle asking price and middle asking rent for the segment after outliers. Two medians over two different populations.
Median annual rent (naira): Middle asking price and middle asking rent for the segment after outliers. Two medians over two different populations.
Gross yield (percent): Annual rent divided by value. Before agency fees, voids, management, tax and maintenance.
Capital growth 12mo (percent): Capital growth is this quarter's median asking value against the same segment's median a year earlier, as the benchmark measures that quarter in this release. Total return is capital growth plus rental income on the base value: this quarter's median annual asking rent divided by the median asking value a year earlier (the price paid when the 12 months began). Gross: before agency fees, voids, costs and tax, with asking rent assumed received for the whole year. Median to median, not chain-linked. A dash means Not comparable.
Total return 12mo (percent): Capital growth is this quarter's median asking value against the same segment's median a year earlier, as the benchmark measures that quarter in this release. Total return is capital growth plus rental income on the base value: this quarter's median annual asking rent divided by the median asking value a year earlier (the price paid when the 12 months began). Gross: before agency fees, voids, costs and tax, with asking rent assumed received for the whole year. Median to median, not chain-linked. A dash means Not comparable.
Gross return vs T-bill: Beat T-bill (gross total return at or above the year-ago quarter's average 364-day T-bill yield), Beat inflation only (above inflation over the same 12 months, below the T-bill), Below both, or Not comparable (no reliable like-for-like 12-month comparison: not the same as no history; the Why blank column gives the reason). Gross means before agency fees, voids, costs and tax; the bill pays tax-free and cost-free.

Evidence and exit

Sale comparables (listings): Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales.
Rent comparables (listings): Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales.
Days to clear (median) (days): Median days from a listing appearing to it leaving the portals (a departure is not a sale, so this is an upper bound on time to sell). Shown as <8 or 180+ at the limits of what is measurable. Where the figure is 180+, it is a lower bound on days to clear.

Flags (Yes / No)

UNIFORM: UNIFORM: Yes where the prices barely vary: the sale P25–P75 band is narrower than 5% of the median, or P25 equals P75. The median then rests on a few repeated prices, not a spread. WIDE: Yes where the sale P25–P75 band is wider than 90% of the median (the limit for grade C), so asking prices disagree. VAGUE: Yes where the source adverts named no specific property type. THIN RENT LEG: Yes where the yield rests on fewer than 30 rent comparables. A yield inherits the weaker of its two counts. Filter any of them to Yes or No. An orange cell always also reads Yes.
WIDE: UNIFORM: Yes where the prices barely vary: the sale P25–P75 band is narrower than 5% of the median, or P25 equals P75. The median then rests on a few repeated prices, not a spread. WIDE: Yes where the sale P25–P75 band is wider than 90% of the median (the limit for grade C), so asking prices disagree. VAGUE: Yes where the source adverts named no specific property type. THIN RENT LEG: Yes where the yield rests on fewer than 30 rent comparables. A yield inherits the weaker of its two counts. Filter any of them to Yes or No. An orange cell always also reads Yes.
VAGUE: UNIFORM: Yes where the prices barely vary: the sale P25–P75 band is narrower than 5% of the median, or P25 equals P75. The median then rests on a few repeated prices, not a spread. WIDE: Yes where the sale P25–P75 band is wider than 90% of the median (the limit for grade C), so asking prices disagree. VAGUE: Yes where the source adverts named no specific property type. THIN RENT LEG: Yes where the yield rests on fewer than 30 rent comparables. A yield inherits the weaker of its two counts. Filter any of them to Yes or No. An orange cell always also reads Yes.
THIN RENT LEG: UNIFORM: Yes where the prices barely vary: the sale P25–P75 band is narrower than 5% of the median, or P25 equals P75. The median then rests on a few repeated prices, not a spread. WIDE: Yes where the sale P25–P75 band is wider than 90% of the median (the limit for grade C), so asking prices disagree. VAGUE: Yes where the source adverts named no specific property type. THIN RENT LEG: Yes where the yield rests on fewer than 30 rent comparables. A yield inherits the weaker of its two counts. Filter any of them to Yes or No. An orange cell always also reads Yes.
Why blank: The reason a 12-month return or a days-to-clear figure is blank on that row. Empty when nothing is blank.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Lookup picks by. One per row, and the key the Lookup reads the grid with.
Answered by: A helper kept out of the grid to keep it readable. The Lookup reads it.
Band width (percent): A helper kept out of the grid to keep it readable. The Lookup reads it.
Slow quarter (days): A helper kept out of the grid to keep it readable. The Lookup reads it.
Price-to-rent yrs (years): A helper kept out of the grid to keep it readable. The Lookup reads it.
Capital growth 12mo, real (percent): A helper kept out of the grid to keep it readable. The Lookup reads it.
Total return 12mo, real (percent): A helper kept out of the grid to keep it readable. The Lookup reads it.
Vs year-ago T-bill (percentage points): A helper kept out of the grid to keep it readable. The Lookup reads it.
Vs inflation (percentage points): A helper kept out of the grid to keep it readable. The Lookup reads it.

Frequently asked questions

The questions asked most

Is the yield gross or net?

Gross. Median annual asking rent divided by median asking value, before agency fees, void periods, management, maintenance and tax, all of which sit outside this data and all of which reduce the return. The Lookup has a net-yield estimator that takes your own assumptions.

Why are the yields low compared to the T-bill?

Because they are, and the workbook shows them rather than hiding them. The Overview sets gross residential yield beside the 364-day Treasury bill yield, the return an investor can actually earn, and inflation. The case for residential in this market rests on capital growth, not yield.

Why is the file sorted by grade before yield?

Because sorting by yield alone puts the least reliable rows at the top. Thin segments produce extreme ratios in both directions, and the highest yields in the file are often the least well evidenced.

What does Meets evidence rule mean?

That the row is grade A or B, normal band, measured at its own grain: grade A or B and a normal price band. It is the field that separates the best-evidenced rows from the rest.

What does Grain mean?

Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area.

Why are sale comparables and rent comparables counted separately?

Because a yield is one median divided by another, over two different populations. A segment with many sale comparables and few rent comparables gives a confident price and a guessed rent, and the yield inherits the weaker of the two.

Why is the verdict tested against the T-bill and not the policy rate?

Because the policy rate is not a return an investor can earn. MPR is the rate at which the CBN lends to commercial banks; the 364-day Treasury bill is federal-government-backed and investable. Gross total return over 12 months is compared with what a 364-day bill paid when those 12 months began: the average true yield of the same quarter a year earlier, over every auction in it. A bill is sold at a discount, so each auction's stop rate is converted to the yield an investor earns, rate / (1 - rate x 364 / 365), before the average is taken.

What we do not answer

Certain questions are outside what a data supplier should answer. The consistent response is: here is what the data shows and why — what you do with it is your policy decision. This applies to:

Which valuation to accept
What LTV or tolerance to set
Whether to lend, build, or buy
Whether a specific scheme is viable
Which segment to invest in
Whether a particular valuer is right

Methodology

Source: Residential sale and rent listings from Nigerian property portals, de-duplicated within each portal (a property on two portals can appear twice). Asking prices and asking rents: Nigeria has no transaction register, so achieved prices are not observable.
Yield: Median annual asking rent divided by median asking value, same segment, grain and quarter, after a Tukey fence at 1.5 x IQR applied to each leg separately. Computed once in the Collateral Benchmark and read here unchanged. Segments under 20 comparables are withheld, not estimated. Every figure is measured at its own grain. A neighbourhood below the minimum comparables is withheld with the reason shown; it is not given the figure of the area around it.
Exit speed: Kaplan-Meier survival over tracked listings entering up to 2026-04-01, each watched 180 days. An older window than the yield on the same row, and a listing leaving the portals is not proof of a sale, so days to clear is an upper bound on time to sell. Where a figure reads 180+, it is a lower bound on days to clear.
12-month return: Capital growth is the median value against the same segment's median a year earlier, as the benchmark measures that quarter in this release, which can differ from what the year-ago release printed; median to median, not chain-linked. Total return is capital growth plus rental income on the base value: this quarter's median annual asking rent divided by the median asking value a year earlier, gross, before agency fees, voids, costs and tax, with asking rent assumed received for the whole year. It is tested against the average 364-day T-bill yield of the same quarter a year earlier, never the MPR: the true (investment) yield converted from each auction's stop rate (the stop-rate average for Q3 2025, 16.65% over 6 auctions, is shown beside it as the discount-basis figure). Shown as Not comparable where there is no year-ago row, either period has under 30 comparables, the comparable count moved enough to describe a different population, or the swing itself exceeded a doubling or halving. Not comparable is not 'no history': most such rows have history that fails one of those tests.

Limitations

The rent index now includes self-contained single rooms, so aggregate rent series are not comparable with earlier releases; individual segment rows and every yield here are unaffected.
The T-bill figure is the true (investment) yield: each auction's stop rate converted with rate / (1 - rate x 364 / 365) and averaged over the quarter's auctions, because a bill is sold at a discount and the stop rate understates what it pays. The stop-rate average is shown beside it, labelled 'stop rate (discount basis)'.
Not a valuation, not a forecast and not a net return. Coverage is concentrated in Lagos and Abuja, and the thinner states carry wider bands and lower grades.
Percentages are stored as fractions (0.0698 = 6.98%) in the workbook.

Glossary

Capital growth / Total return 12mo: Capital growth is this quarter's median asking value against the same segment's median a year earlier, as the benchmark measures that quarter in this release. Total return is capital growth plus rental income on the base value: this quarter's median annual asking rent divided by the median asking value a year earlier (the price paid when the 12 months began). Gross: before agency fees, voids, costs and tax, with asking rent assumed received for the whole year. Median to median, not chain-linked. A dash means Not comparable.
Days to clear (median): Median days from a listing appearing to it leaving the portals (a departure is not a sale, so this is an upper bound on time to sell). Shown as <8 or 180+ at the limits of what is measurable. Where the figure is 180+, it is a lower bound on days to clear.
Grade: How much to trust the row, A to D: comparables count and price agreement. A: Many comparables, prices agree. B: Good comparables. C: Directional only. D: Wide spread: prices disagree. Read it before the yield.
Grain: Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area.
Gross return vs T-bill: Beat T-bill (gross total return at or above the year-ago quarter's average 364-day T-bill yield), Beat inflation only (above inflation over the same 12 months, below the T-bill), Below both, or Not comparable (no reliable like-for-like 12-month comparison: not the same as no history; the Why blank column gives the reason). Gross means before agency fees, voids, costs and tax; the bill pays tax-free and cost-free.
Gross yield: Annual rent divided by value. Before agency fees, voids, management, tax and maintenance.
Median asking / annual rent: Middle asking price and middle asking rent for the segment after outliers. Two medians over two different populations.
Meets evidence rule: Yes = grade A or B, normal band, measured at its own grain. The field that separates the best-evidenced rows from the rest.
Property type / Beds / Vintage: What kind of stock, how many bedrooms, and new-build or existing. These define the segment rather than describing one property.
Sale / Rent comparables: Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales.
State / Area / Neighbourhood: Where the segment is, from broad to narrow. Area is the local district people use.
UNIFORM / WIDE / VAGUE / THIN RENT LEG: UNIFORM: Yes where the prices barely vary: the sale P25–P75 band is narrower than 5% of the median, or P25 equals P75. The median then rests on a few repeated prices, not a spread. WIDE: Yes where the sale P25–P75 band is wider than 90% of the median (the limit for grade C), so asking prices disagree. VAGUE: Yes where the source adverts named no specific property type. THIN RENT LEG: Yes where the yield rests on fewer than 30 rent comparables. A yield inherits the weaker of its two counts. Filter any of them to Yes or No. An orange cell always also reads Yes.
Why blank: The reason a 12-month return or a days-to-clear figure is blank on that row. Empty when nothing is blank.

Collateral Benchmark Report

Test submitted valuations across an entire loan book

About this report and requesting access

What is in the download

Each release downloads as a zip file holding the Excel workbook and a one-page Executive Summary (PDF). The workbook has 7 sheets, in tab order.

Overview: The release on one page: the headline figures, how much to trust them, the main findings and coverage by state.
Benchmark: Every segment's asking-price band (P10–P90), grade, the band the evidence supports and evidence.
Land: Asking land rates per sqm and per plot, published rows only.
Valuation check: Test one valuation against its segment's asking-price band at the tolerance you set, and apply your discount.
Batch check: Up to 500 valuations at once: who answered, grade, your tolerance, position, outcome and asking less your discount.
Changes: What moved since Q2 2026: segments in and out, comparables, grade moves.
Guide: The four-step check, column meanings, method, limits, withheld segments and checks.

How to use it

The check, in four steps

Match the segment. On Valuation check pick the property's state, area, neighbourhood, type, beds and vintage, or paste up to 500 rows into Batch check. Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area. If the neighbourhood has no figure the LGA answers and the sheet says so; a state figure is never used. The Comparables column counts the asking comparables at that Grain.
Read the grade, then set your policy. Grade says how much to trust the row. Every figure is measured at its own grain. A neighbourhood below the minimum comparables is withheld with the reason shown; it is not given the figure of the area around it. The tolerance table on Valuation check holds one treatment per grade and one for a flagged band, and it arrives blank: the tolerance is your policy, not ours, and until you type it the outcome reads 'Enter your tolerance'. Beside each blank the table prints the band the evidence supports for that grade, which is not a recommendation. By construction, half of a segment's own comparables sit outside P25–P75, so a P25–P75 tolerance flags about half of any segment's own comparables as outside. Batch check reads the same table.
Compare the submitted value. The outcome is WITHIN or OUTSIDE the band your policy names (P25–P75 or P10–P90, edges included), REFER or NOT AUTOMATED where your policy for that grade is not to test the value, and NO BENCHMARK where neither the neighbourhood nor the LGA has a figure. The band shows how much the market itself varies; it is not a margin of error. Nothing is judged until a valuation is entered.
Record the index release. Log the index release with the decision. It lets anyone reproduce the same check months later, which an auditor will ask for.

Columns and what they mean

Headers are exactly as printed in the workbook. 39 columns across 2 grids, and 10 hidden helper columns (the workbook's lookup sheets read them; unhide them in Excel to see them).

Benchmark: median value, band and evidence grade per segment

Segment

State: Where the segment is, from broad to narrow. LGA is the official administrative area; Area is the local district people actually use, which is finer. Both appear because one LGA can contain several districts. LGA shows only where it differs from Area. On an LGA row, Neighbourhood reads All neighbourhoods.
LGA: Where the segment is, from broad to narrow. LGA is the official administrative area; Area is the local district people actually use, which is finer. Both appear because one LGA can contain several districts. LGA shows only where it differs from Area. On an LGA row, Neighbourhood reads All neighbourhoods.
Area: Where the segment is, from broad to narrow. LGA is the official administrative area; Area is the local district people actually use, which is finer. Both appear because one LGA can contain several districts. LGA shows only where it differs from Area. On an LGA row, Neighbourhood reads All neighbourhoods.
Neighbourhood: Where the segment is, from broad to narrow. LGA is the official administrative area; Area is the local district people actually use, which is finer. Both appear because one LGA can contain several districts. LGA shows only where it differs from Area. On an LGA row, Neighbourhood reads All neighbourhoods.
Property type: The kind of building, as described in the source listing. Blocks of flats and single rooms are excluded from the index, and so from this file. The number of bedrooms. Segments outside the bedroom range the index covers are not published. Whether the stock is new-build or existing. The two are kept separate because they price differently, and mixing them would blur both.
Beds (bedrooms): The kind of building, as described in the source listing. Blocks of flats and single rooms are excluded from the index, and so from this file. The number of bedrooms. Segments outside the bedroom range the index covers are not published. Whether the stock is new-build or existing. The two are kept separate because they price differently, and mixing them would blur both.
Vintage: The kind of building, as described in the source listing. Blocks of flats and single rooms are excluded from the index, and so from this file. The number of bedrooms. Segments outside the bedroom range the index covers are not published. Whether the stock is new-build or existing. The two are kept separate because they price differently, and mixing them would blur both.

How much to trust it

Grade (grade letter): How much to trust this row overall, A to D. Unlike Confidence, it also accounts for how spread the prices are. A: many comparables, prices agree. B: good comparables. C: directional only. D: wide spread: prices disagree. It sets which of your tolerance policies applies on the tools.
Meets evidence rule (yes or no): Meets the evidence rule for an automated check. Yes means published, band status Normal, and Grade A or B, measured at its own grain. It is a statement about the evidence, not an instruction to automate.
Grain: Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area.
Band the evidence supports: The band the evidence supports for the segment's grade, read from the grade's own definition: grade A: P25–P75; grade B: P10–P90; grade C: P10–P90; grade D: REFER; a flagged band: No usable band. It is not a recommended tolerance: your tolerance is your policy, not ours, and you set it in the Your policy column on Valuation check, which arrives blank. By construction, half of a segment's own comparables sit outside P25–P75, so a P25–P75 tolerance flags about half of any segment's own comparables as outside. The P10–P90 edges are the least stable figures on a thin segment: at 20 comparables each edge is interpolated between two neighbouring listings, so one extra or one missing comparable can move it, and a WITHIN or OUTSIDE result at P10–P90 can flip on a single listing.

Asking price band

Band P10 (naira): Asking-price percentiles for the segment after outlier removal: P25–P75 is the middle half of asking prices, P10–P90 the outer edges ignoring the cheapest and dearest tenth. The band is market variation, not a margin of error.
Band P25 (naira): Asking-price percentiles for the segment after outlier removal: P25–P75 is the middle half of asking prices, P10–P90 the outer edges ignoring the cheapest and dearest tenth. The band is market variation, not a margin of error.
Median asking (naira): The middle asking price for similar properties here: half ask more, half ask less. Outliers removed. A median is used rather than an average so one unusual property cannot distort it.
Band P75 (naira): Asking-price percentiles for the segment after outlier removal: P25–P75 is the middle half of asking prices, P10–P90 the outer edges ignoring the cheapest and dearest tenth. The band is market variation, not a margin of error.
Band P90 (naira): Asking-price percentiles for the segment after outlier removal: P25–P75 is the middle half of asking prices, P10–P90 the outer edges ignoring the cheapest and dearest tenth. The band is market variation, not a margin of error.
Band width (percent): How much prices vary within this segment. Low means the market broadly agrees on price here; high means it does not. Calculated as (P75 − P25) ÷ median.

Evidence and trend

Comparables (listings): Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales. The count is taken at the row's own Grain: a Neighbourhood row counts that neighbourhood's comparables, an LGA row counts the whole local government area's.
Capital growth 12mo (percent): This row's median asking price against the same segment's median twelve months earlier. Median to median, not chain-linked, and computed from asking prices alone: a segment with no rent figure still has one. A dash means there is no reliable like-for-like comparison.
Notes: Very wide price band: the segment's P25–P75 band is wider than 90% of its median, the limit for grade C. Band too narrow to use (the comparables rest on very few distinct prices) or too wide to use (it tells you nothing). Neither band can be used as an automated threshold.

Flags and gaps

UNIFORM: Yes/No flags you can filter on; a withheld row reads a dash. UNIFORM: the band is narrower than 5% of the median (band status Degenerate) or P25 equals P75, so a few repeated asking prices set the whole band; the comparables rest on very few distinct prices. WIDE: the band is implausible, or wider than 90% of its median, the limit for grade C. CLEAN: published, normal band, neither UNIFORM nor WIDE.
WIDE: Yes/No flags you can filter on; a withheld row reads a dash. UNIFORM: the band is narrower than 5% of the median (band status Degenerate) or P25 equals P75, so a few repeated asking prices set the whole band; the comparables rest on very few distinct prices. WIDE: the band is implausible, or wider than 90% of its median, the limit for grade C. CLEAN: published, normal band, neither UNIFORM nor WIDE.
CLEAN: Yes/No flags you can filter on; a withheld row reads a dash. UNIFORM: the band is narrower than 5% of the median (band status Degenerate) or P25 equals P75, so a few repeated asking prices set the whole band; the comparables rest on very few distinct prices. WIDE: the band is implausible, or wider than 90% of its median, the limit for grade C. CLEAN: published, normal band, neither UNIFORM nor WIDE.
Why blank: Why a figure on the row is blank. A withheld segment is blank throughout and the cell gives its reason. A published segment can still lack Capital growth 12mo, and the cell names the first test it failed.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Valuation check picks by. One per row, and the key the Valuation check reads the grid with.
Published key: A helper kept out of the grid to keep it readable. The Valuation check reads it.
Tolerance code: A helper kept out of the grid to keep it readable. The Valuation check reads it.
Band status: A helper kept out of the grid to keep it readable. The Valuation check reads it.
Confidence: A helper kept out of the grid to keep it readable. The Valuation check reads it.
Pre-fence (listings): A helper kept out of the grid to keep it readable. The Valuation check reads it.

Land: asking rate per square metre, title unobserved

Where and what

Grain: LGA (the area's own rate) or Neighbourhood. These are separate populations, not a hierarchy: a neighbourhood row is not part of the area row beside it, and neither is substituted for the other. Commercial and mixed-use land appear at area level only, because at neighbourhood level there is too little of either to publish.
State: Where the land is. (whole area) means the row is the area's own rate, not one neighbourhood in it.
Area: Where the land is. (whole area) means the row is the area's own rate, not one neighbourhood in it.
Neighbourhood: Where the land is. (whole area) means the row is the area's own rate, not one neighbourhood in it.
Use class: Residential, mixed use or commercial land, as the source classified it. They are not pooled: commercial land trades well above residential land in the same neighbourhood, so averaging them would describe neither.
Plot size: Standard is under 2,000 square metres; Large is 2,000 and above. Larger plots trade at a lower rate per square metre, so the two are kept apart.

Rate per sqm

Band P10 / sqm (naira per square metre): The same bands as the property grid, per square metre: P25–P75 the middle half of asking rates, P10–P90 the outer edges.
Band P25 / sqm (naira per square metre): The same bands as the property grid, per square metre: P25–P75 the middle half of asking rates, P10–P90 the outer edges.
Rate per sqm (naira per square metre): The middle asking rate for land in this segment — half ask more per square metre, half ask less. Computed per listing and then taken as a median, not as one price divided by one area, so it does not move with whatever mix of plot sizes happened to be advertised.
Band P75 / sqm (naira per square metre): The same bands as the property grid, per square metre: P25–P75 the middle half of asking rates, P10–P90 the outer edges.
Band P90 / sqm (naira per square metre): The same bands as the property grid, per square metre: P25–P75 the middle half of asking rates, P10–P90 the outer edges.

Per plot

Plot standard (sqm) (square metres): The plot size this state actually trades in, measured from land comparables that describe themselves as one plot. There is no single national plot. Read from this release, by state: Abia 464 sqm (16 comparables), Abuja 500 sqm (76 comparables), Akwa Ibom 900 sqm (6 comparables), Anambra 465 sqm (18 comparables), Delta 464 sqm (119 comparables), Enugu 500 sqm (14 comparables), Imo 900 sqm (136 comparables), Kwara 450 sqm (32 comparables), Lagos 500 sqm (696 comparables), Ogun 600 sqm (348 comparables), Ondo 600 sqm (14 comparables), Osun 450 sqm (27 comparables), Oyo 500 sqm (189 comparables), Rivers 465 sqm (237 comparables). Blank where too few comparables quote a plot to measure it.
Value per plot (naira): The rate per square metre multiplied by that state's plot standard, because that is how the market quotes land. It carries no evidence of its own beyond the rate. Blank where the state has no measured standard — deliberately, rather than assuming a national figure.

Trust

Grade (grade letter): Graded on comparables count and band width like the property grades, then capped: no land row may grade better than B, because title is unobserved on all of them. A land B and a property B are NOT the same claim.
Land comparables (listings): Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales. On Land, each comparable is an asking rate per square metre.

Helper columns (hidden)

Segment key: The segment's combined label (State, Area, Neighbourhood, Property type, Beds, Vintage) that the Valuation check picks by. One per row, and the key the Valuation check reads the grid with.
Band width (percent): A helper kept out of the grid to keep it readable. The Valuation check reads it.
Band status: A helper kept out of the grid to keep it readable. The Valuation check reads it.
Pre-fence (listings): A helper kept out of the grid to keep it readable. The Valuation check reads it.

Frequently asked questions

The questions asked most

What does "Meets evidence rule: Yes" actually mean?

That the row is published, its band status is Normal, and its grade is A or B, measured at its own grain. It is a statement about the evidence, not an instruction to automate.

What tolerance should we set for flagging valuations?

That is a policy decision for your credit committee. The tolerance table on Valuation check holds one treatment per grade, because segments vary enormously in how tightly prices cluster. It arrives blank, so the tolerance is yours, not ours, and the table shows only the band the evidence supports for each grade. Be aware that, by construction, a P25–P75 tolerance flags about half of a segment's own comparables as outside.

Which figure do we lend against: the submitted valuation or your benchmark?

Yours to decide. We provide market evidence; the lending policy is yours. The benchmark is a segment median of asking prices and a valuation is a professional judgement about one property. They measure different things and neither is automatically right.

What happens when a segment we lend in is withheld?

The row stays on Benchmark with its values blank and the reason in Why blank. Valuation check and Batch check ask the neighbourhood first and then the LGA: where the LGA has a figure it answers and the sheet says so, and where neither has one the outcome reads NO BENCHMARK. Withheld, not substituted: a neighbourhood with too few comparables is not given the figure of the area around it, and a state figure is never used.

What does Grain mean?

What if the property has features you cannot see?

Then the figure will be wrong for that property, and the band is where it shows. Corner plots, extensions, exceptional finish, a C of O against unregistered title: none of these are visible in listing data.

Can you value a specific property for us?

No. We publish segment-level market evidence, not property-level valuations. A specific property needs an inspection, a title check and professional judgement, none of which we do.

What we do not answer

Certain questions are outside what a data supplier should answer. The consistent response is: here is what the data shows and why — what you do with it is your policy decision. This applies to:

Which valuation to accept
What LTV or tolerance to set
Whether to lend, build, or buy
Whether a specific scheme is viable
Which segment to invest in
Whether a particular valuer is right

Methodology

Source: Residential listings from Nigerian property portals, de-duplicated within each portal (a property on two portals can appear twice). Asking prices: Nigeria has no transaction register, so achieved prices are not directly observable.
Stratification: State, area and neighbourhood, crossed with property sub-type, bedroom count and new-build or existing vintage, identical to the index, so a property matched here lands in the same stratum the index reports. The area is also published once as its own LGA row, using every comparable in the area.
Bands: A Tukey fence at 1.5 x IQR is applied within each segment and period, and the median and every band are computed over exactly the rows that survive it.
Minimum comparables: A segment with fewer than 20 comparables in the period is withheld, not estimated.
Grain: Every figure is measured at its own grain. A neighbourhood below the minimum comparables is withheld with the reason shown; it is not given the figure of the area around it.
Refresh: Quarterly, following the index release. The index release is on the Overview.

Limitations

Each comparable is built from listings (adverts) for the same area, type and bedrooms. These are asking prices, not sold transactions.
Your tolerance is your policy, not ours. By construction, half of a segment's own comparables sit outside P25–P75, so a P25–P75 tolerance flags about half of any segment's own comparables as outside. The P10–P90 edges are the least stable figures on a thin segment: at 20 comparables each edge is interpolated between two neighbouring listings, so one extra or one missing comparable can move it, and a WITHIN or OUTSIDE result at P10–P90 can flip on a single listing.
Capital growth is the change in the median asking price against the same segment a year earlier, shown where both quarters have at least 30 comparables and the sets of comparables are comparable: the comparables count did not move enough to describe a different population, the swing is within a doubling or halving, and the band moved with the median. It uses asking prices only, so a segment with no rent figure still has one (the calculation is shared with Investor Yield).
Rent, yield, total return and the return-against-benchmarks verdicts need a rent figure and are not on this workbook's grids: they are Investor Yield's.
Not a valuation: nothing here inspects, measures or appraises a specific property, and nothing addresses title, planning consent or approval risk, the constraints that most often bite in Nigeria. Floor area is unavailable for most built stock.
Percentages are stored as fractions (0.0698 = 6.98%) in the workbook.

Glossary

Band P10 / P25 / P75 / P90: Asking-price percentiles for the segment after outlier removal: P25–P75 is the middle half of asking prices, P10–P90 the outer edges ignoring the cheapest and dearest tenth. The band is market variation, not a margin of error.
Band P10 / P25 / P75 / P90 per sqm: The same bands as the property grid, per square metre: P25–P75 the middle half of asking rates, P10–P90 the outer edges.
Band the evidence supports: The band the evidence supports for the segment's grade, read from the grade's own definition: grade A: P25–P75; grade B: P10–P90; grade C: P10–P90; grade D: REFER; a flagged band: No usable band. It is not a recommended tolerance: your tolerance is your policy, not ours, and you set it in the Your policy column on Valuation check, which arrives blank. By construction, half of a segment's own comparables sit outside P25–P75, so a P25–P75 tolerance flags about half of any segment's own comparables as outside. The P10–P90 edges are the least stable figures on a thin segment: at 20 comparables each edge is interpolated between two neighbouring listings, so one extra or one missing comparable can move it, and a WITHIN or OUTSIDE result at P10–P90 can flip on a single listing.
Band width: How much prices vary within this segment. Low means the market broadly agrees on price here; high means it does not. Calculated as (P75 − P25) ÷ median.
Capital growth 12mo: This row's median asking price against the same segment's median twelve months earlier. Median to median, not chain-linked, and computed from asking prices alone: a segment with no rent figure still has one. A dash means there is no reliable like-for-like comparison.
Comparables: Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales. The count is taken at the row's own Grain: a Neighbourhood row counts that neighbourhood's comparables, an LGA row counts the whole local government area's.
Grade: How much to trust this row overall, A to D. Unlike Confidence, it also accounts for how spread the prices are. A: many comparables, prices agree. B: good comparables. C: directional only. D: wide spread: prices disagree. It sets which of your tolerance policies applies on the tools.
Grade: Graded on comparables count and band width like the property grades, then capped: no land row may grade better than B, because title is unobserved on all of them. A land B and a property B are NOT the same claim.
Grain: Neighbourhood: the figure uses only that neighbourhood's comparables. LGA: it uses every comparable in the local government area. Nothing is substituted: a neighbourhood with too few comparables is withheld, not filled in from a wider area.
Grain: LGA (the area's own rate) or Neighbourhood. These are separate populations, not a hierarchy: a neighbourhood row is not part of the area row beside it, and neither is substituted for the other. Commercial and mixed-use land appear at area level only, because at neighbourhood level there is too little of either to publish.
Land comparables: Comparables are asking prices (or rents, or land rates) for the same segment, first advertised in the quarter, at their latest asking price, de-duplicated within each portal and with outliers removed. A property advertised on more than one portal can therefore appear more than once. They are not sales. On Land, each comparable is an asking rate per square metre.
Median asking: The middle asking price for similar properties here: half ask more, half ask less. Outliers removed. A median is used rather than an average so one unusual property cannot distort it.
Meets evidence rule: Meets the evidence rule for an automated check. Yes means published, band status Normal, and Grade A or B, measured at its own grain. It is a statement about the evidence, not an instruction to automate.
Notes: Very wide price band: the segment's P25–P75 band is wider than 90% of its median, the limit for grade C. Band too narrow to use (the comparables rest on very few distinct prices) or too wide to use (it tells you nothing). Neither band can be used as an automated threshold.
Plot size: Standard is under 2,000 square metres; Large is 2,000 and above. Larger plots trade at a lower rate per square metre, so the two are kept apart.
Plot standard (sqm): The plot size this state actually trades in, measured from land comparables that describe themselves as one plot. There is no single national plot. Read from this release, by state: Abia 464 sqm (16 comparables), Abuja 500 sqm (76 comparables), Akwa Ibom 900 sqm (6 comparables), Anambra 465 sqm (18 comparables), Delta 464 sqm (119 comparables), Enugu 500 sqm (14 comparables), Imo 900 sqm (136 comparables), Kwara 450 sqm (32 comparables), Lagos 500 sqm (696 comparables), Ogun 600 sqm (348 comparables), Ondo 600 sqm (14 comparables), Osun 450 sqm (27 comparables), Oyo 500 sqm (189 comparables), Rivers 465 sqm (237 comparables). Blank where too few comparables quote a plot to measure it.
Property type / Beds / Vintage: The kind of building, as described in the source listing. Blocks of flats and single rooms are excluded from the index, and so from this file. The number of bedrooms. Segments outside the bedroom range the index covers are not published. Whether the stock is new-build or existing. The two are kept separate because they price differently, and mixing them would blur both.
Rate per sqm: The middle asking rate for land in this segment — half ask more per square metre, half ask less. Computed per listing and then taken as a median, not as one price divided by one area, so it does not move with whatever mix of plot sizes happened to be advertised.
State / Area / Neighbourhood: Where the land is. (whole area) means the row is the area's own rate, not one neighbourhood in it.
State / LGA / Area / Neighbourhood: Where the segment is, from broad to narrow. LGA is the official administrative area; Area is the local district people actually use, which is finer. Both appear because one LGA can contain several districts. LGA shows only where it differs from Area. On an LGA row, Neighbourhood reads All neighbourhoods.
UNIFORM / WIDE / CLEAN: Yes/No flags you can filter on; a withheld row reads a dash. UNIFORM: the band is narrower than 5% of the median (band status Degenerate) or P25 equals P75, so a few repeated asking prices set the whole band; the comparables rest on very few distinct prices. WIDE: the band is implausible, or wider than 90% of its median, the limit for grade C. CLEAN: published, normal band, neither UNIFORM nor WIDE.
Use class: Residential, mixed use or commercial land, as the source classified it. They are not pooled: commercial land trades well above residential land in the same neighbourhood, so averaging them would describe neither.
Value per plot: The rate per square metre multiplied by that state's plot standard, because that is how the market quotes land. It carries no evidence of its own beyond the rate. Blank where the state has no measured standard — deliberately, rather than assuming a national figure.
Why blank: Why a figure on the row is blank. A withheld segment is blank throughout and the cell gives its reason. A published segment can still lack Capital growth 12mo, and the cell names the first test it failed.