Rehani AnalyticsLive models

Kenya, Uganda, Tanzania

Read the numbers.Before youmake the offer.

Cap rate, a 10-year IRR, vacation-rental occupancy, and a due-diligence check for any listing or neighborhood across 20 countries and 58,454 neighborhoods in Africa.

Try , ,

The data behind RehAnalytics

Data points
585,536
Machine-learning models on price, rent, and risk
119
Markets with neighbourhood coverage
20

Six ways to use RehAnalytics

How it works

  1. 01Name the placeA neighbourhood, a project, or a listing URL.
  2. 02Pick the readingFull return, vacation rental, due diligence, or fraud.
  3. 03Read it, or send itOn screen, by email, or as a PDF.

Some feature neighborhoods

NeighbourhoodCityData points
WestlandsNairobi27,546
LekkiLagos17,447
KilimaniNairobi14,441
LavingtonNairobi11,104
KileleshwaNairobi10,390
BoleAddis Ababa6,799
East LegonAccra5,608
KarenNairobi4,111
AjahLagos4,004
NyaliMombasa3,600
PrampramAccra3,488
TemaAccra3,033
NairobiWestNairobi2,675
IkoyiLagos2,590
AdentaAccra2,161
Airport Residential AreaAccra2,034
Victoria IslandLagos2,033
Kampala CentralKampala1,998

What is an AVM?

An automated valuation model is a computer estimate of what a home would sell for, what it would rent for, or what a night in it would earn. It is the same kind of tool the big property sites use when they put a value on a house that has not been visited by a valuer. It is not an appraisal, and it is not a promise about the next sale.

How it is made

A separate model is trained for each kind of home, in each place. It studies homes that were actually sold or let, and learns which details move the price: the neighbourhood, the size, the type of home. The asking price on a listing is left out. When the sales on record for that place change, the estimate moves.

Every lease and offer signed through Hani adds a confirmed price to the model, so estimates sharpen as more deals close.

What you use it for

A first read, before you offer. It tells you whether a price is in line with what similar homes in that neighbourhood have done. Buyers, investors, and lenders use the same kind of model for that first pass. A bank valuation, or your own inspection, still comes after.

How we measure it

We test each model on real sales and lets it never saw, then report the typical miss: how far a usual estimate lands from the price actually paid. On sales in Nachu, Kiambu, the typical miss is 12.9%. On rents in Kampala, the typical miss is 18.7%. We also show the fit score, which tells you how closely the model follows past prices in that area.

We only show estimates where we have enough recent deals to test them. A model that does well in one city may not do as well in the next, so every neighbourhood is measured on its own.

How this compares

In the US and Europe, home-value tools typically land within about 2% of the sale price when a home is already listed, because they can see the asking price. For homes that aren't listed, they're closer to 7%.

Ours never use the asking price. In most of East Africa, the asking price is where negotiation starts, not where it ends, so we learn only from what homes actually sold or let for. Our typical miss on sales is 12.9%, and on rents 18.7%. Treat the estimate as a range, not a single price.

Some sample models

Sale
12.9%

Nachu, Kiambu

  • Fit0.92
Rent
18.7%

Kampala

  • Fit0.90
Vacation rental
Not enough recent deals here yet

Run it on a real listing.

Paste a URL or name a neighbourhood.

RehAnalytics is decision support. Figures reflect available data and may change. They are not an offer of financial, legal, or tax advice. Past performance does not predict future results. Do your own due diligence before you invest. Rehani Soko · 2026