Observe
Prices are captured from public retail surfaces within defined access boundaries, with stop conditions written down before collection rather than after.
Basket BeeTalk to usHow it works
Measurement is a chain of guarantees. Each stage below exists to make the next one trustworthy, and the whole chain exists so that a number leaving the system means exactly what it appears to mean.
Prices are captured from public retail surfaces within defined access boundaries, with stop conditions written down before collection rather than after.
Every retailer's own format becomes one common structure, so a price from one chain means the same thing as a price from another.
Discounts, bundles and conditional pricing are recorded as what they are, so a headline number is never mistaken for what a shopper actually pays.
Listings from different retailers are grouped into one product wherever they genuinely refer to the same thing, and deliberately not grouped where they do not.
Uncertain cases are escalated rather than guessed. A decision that cannot be justified does not enter the catalog.
Approved data lands under automated checks, and every figure keeps a trail back to the capture it came from.
Collecting prices is the easy half. The difficult half is deciding that a product sold by one chain and a product sold by its competitor are the same thing — when the two descriptions share no common identifier, mix Arabic and English, disagree about pack size, and were written by different people for different purposes.
Everything commercially valuable rests on that judgement being right. A comparison built on a wrong match is worse than no comparison, because it is confidently wrong.
Match quality is scored against barcode evidence withheld from the matcher, so the check cannot be influenced by the decision it is checking. Only a minority of matches have clean barcodes on both sides and can be checked at all; the rest have no ground truth, and we state that instead of folding it into a single flattering average. No rate is published until it has been measured against the catalog as it stands today.
Six responsibilities, each of which has to hold for the licensed product to be worth paying for.
Measurement spans the major grocery chains operating in Egypt plus the marketplace and delivery platforms that sell alongside them. Where a retailer sells both in store and for delivery, both are read separately rather than averaged.
The layer that decides when two listings are the same product. Everything commercially useful depends on this being right.
Ambiguity is routed to review instead of being resolved silently. The record shows who decided what.
Prices differ by area, so they are resolved by zone and never flattened into one national figure that is true nowhere.
Data reaches the licensed product through automated integrity checks, not by hand.
One internal console runs and monitors the whole cycle, which is what lets a small team operate infrastructure this size.
The line between what runs today and what is still being built is stated plainly. Trust in a measurement business is the product, and it does not survive a roadmap written in the present tense.
Broad multi-chain collection, cross-retailer product resolution, offer modelling, uncertainty review, controlled publication, and the operator console that runs all of it.
Taking the client data model, scoping and packaging from a controlled environment to a customer-facing intelligence portal, and moving collection from operator-run to scheduled.
Land the first licensed accounts, then widen category depth and the analytics sold on top of the same measurement.
The measurement layer is built. The work in front of us is turning it into a scheduled, customer-facing service: moving collection from operator-run to automatic, and delivering scoped access to the brands and agencies who license it.