"I think" becomes "the data shows" or "our hypothesis is."
The manifesto
Commerce is a system.
Most companies treat it like a series of tasks. Launch the listing. Run the ads. Chase the stockout. Repeat until margin disappears.
We don't do tasks. We build the operating system underneath them. Every decision on evidence. Every workflow designed to be repeated. Every automation built to make a human operator faster, sharper, and harder to compete with.
We are not an agency. We are the engineering department your commerce business never had.
East Star Commerce — the system behind the growthOrigin logic
Agencies sell hours. Hours don't compound. Systems do.
East Star began with a pattern its founders kept seeing inside 7- and 8-figure Amazon businesses: the growth ceiling was never talent or capital. It was the operating model. Brands were buying disconnected services — an ads person, a listings person, a supply-chain spreadsheet — while competitors built connected systems where inventory, advertising, and pricing informed each other automatically.
So East Star was built the way a software company is built: a core intelligence layer, documented workflows, automation pipelines, and human strategists who operate the system rather than improvise around it. The name is deliberate — the east star is the navigator's fixed reference. Markets move; the reference point doesn't.
01ObservationServices don't scale outcomes.
02ThesisCommerce is an engineering problem.
03SystemOne intelligence layer, many workflows.
04Compounding12-month curves, not launch spikes.
Direction
1 mission. 1 vision. 1 purpose.
Mission
Engineer the systems that make ecommerce businesses measurably more profitable, more predictable, and less dependent on manual effort.
Vision
A world where every serious commerce brand runs on an intelligence layer — inventory, advertising, pricing, and cash decided together.
Purpose
Operators built modern commerce. They deserve better tools than spreadsheets and guesswork.
Measurable — every engagement carries defined metrics
Predictable — forecast accuracy is a deliverable
Automated — coverage is reported like a KPI
Core values — decision rules, not wall posters
Each value carries a proof, so it can be hired for, promoted on, and designed with.
Evidence over opinion
In practice
Enforced by
No decision ships without data or a stated hypothesis.
Systems over heroics
In practice
A brilliant one-off fix is a failure of process.
Enforced by
Post-mortems ask "what system prevents this," never "who saved it."
Automation with judgment
In practice
Machines execute; humans decide.
Enforced by
Every automation has a named human owner and an override path.
Clarity is a feature
In practice
If the client can't explain it, we haven't finished it.
Enforced by
Deliverables are read aloud to a non-specialist before shipping.
Compounding beats bursts
In practice
We optimize the 12-month curve, not the launch spike.
Enforced by
Reporting defaults to trailing-90-day views before weekly views.
Every decision, measurable.Every process, repeatable.Every result, reported — including the misses.
In contracts — metrics before price · forecast vs actual in every report · the constraint in every case study
The operators
Operators, not account managers.
Bios below state what each person owns and the number they answer for. Photography is documentary — people at work, not people at lens.
Amina Raza
Head of forecastingOwns the demand model — every reorder gate crosses her desk.
96.1% days-of-cover accuracy — trailing 90dDecision right: every reorder approvalDaniyal Khan
Head of pricingOwns unit economics — COGS current, elasticity observed.
14 fee types audited weeklyDecision right: guardrail changesSara Iqbal
PPC leadOwns the rule engine and everything it isn't allowed to do.
2.4K bid decisions/day · 12 rules liveDecision right: the review queueOmar Siddiqui
Listings leadOwns the test queue. Ends listing debates with test IDs.
214 tests run in 2025 · 57% win rateDecision right: ship or killHina Baig
Head of engineeringOwns the pipelines and the override paths on all of them.
14.2K records/day ingestedDecision right: what gets automatedBilal Ahmed
Client systems directorOwns the scorecards — and the obligation to publish the misses.
Forecast vs actual — published monthlyDecision right: what gets reportedGovernance
We audit our own brand. The score is public.
Every quarter, 10 live artifacts — pages, proposals, charts, social tiles — are sampled against the brand system's own checklists. The score is published, with the violations. A brand built on reported misses reports its own.
Q2 2026 brand auditSELF-ASSESSED · A.03 CHECKLIST
0/10010 artifacts sampled
12 checks each
12 checks each
Violations found — 02
- A proposal chart shipped with a forecast line drawn solid. Corrected — the template now renders forecasts dashed by default.
- A social tile used 2 copper elements in 1 frame. One of them was wrong (3.03). Retired.
Next audit — Oct 2026Method: A.03 launch checklist, sampled retroactively
Q2 2026 brand audit score: 92 out of 100, from 10 artifacts and 12 checks each. Two violations found: a solid forecast line, corrected; and a tile with two copper elements, retired.
Work with the engineering department you never had.
The audit is where every engagement starts — metrics stated before price.