Sero Lab Start
Lab productMarketing · DataEngine built

Muashirمؤشر

Marketing efficiency for stores that spend across several ad platforms and cannot trust the sum of what those platforms report.

muashair.com ↗
3daysaccepted spec to verified engine
158tests passing51 against real PostgreSQL 17
32database tables11 migrations
7platforms modelledads and stores
01

Why

Every ad platform claims credit for the same sale. Add up what they report and a store appears to have sold more than it did.

So the owner sees spend rising and sales rising, while the cash does not. The real question is where to spend, not whether to spend more.

One rule came out of that: never show a zero when a source failed. A missing number that looks like a real zero is how bad decisions get made.

02

How

One hero metric

Marketing efficiency: total ad spend against real store revenue, not revenue as each platform reports it.

Attribution with the gaps shown

Sales are matched by click id, then by UTM. Whatever cannot be matched stays visible as its own row and is never quietly redistributed.

Coverage recorded at sync time

Which data arrived and which did not is written as it happens, because it cannot be reconstructed later.

Arabic first, secure by default

Right to left from the start, with row-level security on every read and every write.

03

What

  • A metric registry with one validated query path
  • Append-only raw data with revisions, so history is never overwritten
  • Goals and forecasts that refuse to evaluate incomplete date ranges
  • An AI analyst that turns a plain question into a structured query
  • Attribution and profit views, with campaign and decision modes
  • An encrypted credential service for connected accounts
Next.js · PostgreSQL 17 · Drizzle · row-level security · BullMQ · Anthropic Claude
04

Impact

By numbers

3days from accepted spec (19 August 2026) to an engine verified against a real database
25,074lines of code
186interface strings in Arabic and English
22screens
Not measured yet

The engine currently runs on seeded demo data. Live ad-platform connectors come next, so there are no customer results to report yet. The waitlist is open at muashair.com.

What it proves

A data product with real correctness guarantees does not need a quarter of scoping before the first honest number appears.

Next case

Shakila

How you work with AI

Your turn

What are you still doing by hand?

Seven short sections. We reply within one working day.