Market research › Methodology

Methodology

How every AAE.NG inventory snapshot is produced. Rules version v1.0.0, applied identically to the edition published on 2026-10-17.

What an edition is

An edition is a single frozen capture of AAE.NG's own listed inventory, taken at one instant and never subsequently edited. It is not a national price guide, not a valuation of any vehicle, and not a time series. Each edition names its own geographic coverage; the marketplace's national reach does not extend the data's coverage.

Eligibility rules

Rules applied in order. Denominators are carried through every published figure.
Rule	Definition
Public listing	Listing status is `active`, `reserved` or `sold`. Draft, pending, rejected and archived listings are never captured.
Active set	Status `active`. Reported only as context.
Used, eligible	Active listings whose condition is `foreign_used` or `locally_used`. `brand_new` listings are excluded from the used analysis.
Used, priced	Eligible listings with a finite numeric price greater than zero. This is the analysis denominator for all price statistics.
Price on request	Listings with no numeric price. Excluded from price statistics and reported separately.

Price bands

Band boundaries in Naira. Bands are half-open: the lower bound is included, the upper bound is not.
Band key	Range
<5m	< ₦5m
5-<10m	₦5m – <₦10m
10-<15m	₦10m – <₦15m
15-<20m	₦15m – <₦20m
20-<30m	₦20m – <₦30m
30m+	₦30m and above

Band counts must sum exactly to the priced denominator. Any difference is an error, not a rounding effect.

Quantile method

Quantiles are computed by linear interpolation on the ascending sorted series (rules v1.0.0). Position i = (n − 1) × p; the value is interpolated between the adjacent order statistics. The method is fixed so that repeated runs over an unchanged snapshot return identical values. The published median is a mixed-listing distribution midpoint across different makes, models, years and conditions — not a valuation, not a “typical car price”, and not a national benchmark.

Suppression

Model × year × condition cohorts are published only when they contain at least 10 listings. Smaller cohorts are withheld because a median computed from so few listings is not a stable benchmark: it would shift materially when a single listing is added or removed, and it would be read as a market price for that model, year and condition when it is not one. When no cohort reaches the threshold, no cohort median is published and the edition says so explicitly.

This threshold is a sample-stability rule, not a privacy or disclosure-control measure. It does not exist to prevent inference of individual asking prices: those are public marketplace listings, published per listing in every downloadable snapshot. Suppression withholds an unstable aggregate, not the underlying data. The dataset's actual privacy controls are the field allowlist described below, which govern internal identifiers, dealer and owner references, buyer data, VINs and contact details.

Seller concentration

Concentration is computed and reported separately for the eligible and the priced used sets, and is presented alongside the price findings rather than buried in an appendix. Seller identity is never published — only an anonymous, stable token derived by hashing the internal seller reference.

Uniqueness and duplicates

Counts are listings. No VIN or photograph adjudication is performed, so listings are never described as “unique vehicles”. Duplicate-image flags, where they exist, record image reuse only; a dismissed flag is not evidence that two listings are the same vehicle, and no de-duplication is claimed on that basis.

Currency

The listing schema has no currency field. Prices are interpreted as Nigerian Naira (NGN) by marketplace display format and are not currency-validated. Every edition states this rather than asserting that amounts are denominated in NGN.

Privacy allowlist

Snapshots are built field-by-field from an explicit allowlist. Published per-listing fields are limited to: listing URL, slug, status, make, model, model year, condition, body type, transmission, fuel type, asking price, price-on-request flag, mileage and unit, city, and an anonymous seller token.

Never published: internal entity identifiers, dealer or owner identifiers, buyer data, VINs, contact details, or any private field. No field is added to the snapshot implicitly — new fields are private by default.

Snapshot integrity

Each edition publishes a snapshot file, a row digest, and a verification manifest. The capture timestamp is recorded in both UTC and WAT and reflects the instant the snapshot was taken.

The snapshot file is a pretty-printed JSON object that wraps the public rows array alongside its own metadata. Two different SHA-256 digests are published, and they cover different bytes:

Whole-file SHA-256 — the digest of a downloaded file exactly as delivered, byte for byte. The manifest carries one entry per published file, recording that digest and the file's byte length. This is the simple integrity check.
Row digest — the sha256 recorded inside the snapshot and repeated on the edition page. It covers the rows array alone, never the file: specifically the UTF-8 bytes of the compact JSON serialization of rows — no indentation and no line breaks, each row's keys in the fixed order stored on that row, and the rows in their stored order (ascending slug). Because the wrapper object is excluded, the digest can be published inside the very file it describes without circularity.

To verify an edition:

File integrity — hash the downloaded snapshot file as-is and compare it with that file's entry in the manifest. A match confirms the file arrived unaltered.
Source rows — parse the file, take the rows array, re-serialize it compactly, hash those bytes, and compare with the row digest shown on the edition page. A match confirms the rows are the frozen ones the figures were computed from.

The two digests are never equal, because the file digest also covers the wrapper and its indentation. Hashing the downloaded file and comparing the result with the row digest will not match, and is not a valid check.

Snapshots are frozen: an edition is never rewritten after publication. Corrections are issued as a new edition.

What these editions do not do

They do not publish an index, a trend line, or month-on-month movement. There is no retained historical series to support one.
They do not extrapolate beyond the geography named in the edition.
They do not present asking prices as sale prices or as valuations.

Independent inspection

AAE.NG recommends an independent pre-purchase inspection before any payment. See how to inspect a used car in Asaba and how to verify a car before buying.