Wired into Meta and Google, not beside them
Kepra talks to the ad platforms through typed MCP tools – pause a campaign, set a budget, adjust a bid cap. The fix executes against the platform's own API in about a second.
● Kepra for e-commerce
Built for shops that live on ads.
Shopify, Meta and Google in one profit view – margin per ad after COGS, repeat-purchase truth, and fixes that execute inside the ad platforms the moment you approve them.
Sample data · no signup · OAuth setup in minutes
Meta says ROAS 4.2. Your bank account disagrees.
Ad platforms report revenue against their own attribution and grade their own homework. Kepra matches orders back to the ad clicks behind them and judges on profit – with an honest unattributed line for what cannot be matched.
COGS never makes it into the dashboard.
A ROAS of 2.5 on a 40% margin product is exactly break-even – it contributes nothing. We pull product costs from Shopify, so every campaign is judged on what is left after the goods are paid.
First orders look great. One-time customers aren't.
A campaign can fill the order feed and still bleed money if most buyers never return. Cohort-based verdicts catch the difference between growth and churn-and-burn – usually weeks before the trend does.
Finding the leak is half the job
The fix ships from the same screen.
Every other tool ends with a chart and leaves you the homework: open Ads Manager, find the campaign, make the change, hope you typed it right. Kepra is wired into Meta and Google – so when you approve a fix, it executes against the platform’s own API in about a second, and the loop is closed before the tab would have finished loading.
The twenty minutes you skip
No switching to Ads Manager, no hunting for the campaign in the wrong account, no typing a budget with one zero too many. The change you approved is exactly the change that ships.
You stay the decision-maker
Nothing reaches your ad account without your approval of that specific change. That is a commitment in our terms, not a toggle – there is no autopilot mode to accidentally leave on.
Logged, and revertible in one click
Every executed action stores who approved it, the evidence it cited, the platform's response, and the state it replaced. Undo is a click, not an archaeology exercise.
◆ The loop, on Meta Ads
Meta Ads → Kepra
K
Reading
spend + matched orders
Deciding
profit after COGS and LTV
Your call
nothing moves until you say so
Pushing
executed over MCP, logged
Verdict: Prospecting – Interests spends €9,220/mo. Meta reports ROAS 2.8; matched against real orders it returned €6,450, and 71% of those buyers never came back. After COGS it is €5,200/mo underwater.
Pause campaignawaiting your approval
Live from the demo – a running-gear shop on sample data
Demo data⚠ LeakMeta · Prospecting – Interests · ad 2901-B
You're spending €9,220/mo on this campaign. Meta reports a ROAS of 2.8. Matched against orders that actually originated here it returned €6,450 – and 71% of those buyers never came back. After COGS the campaign is €5,200/mo underwater.
You approve – Kepra pushes it to Meta
▲ Scale MeterDemo data
Can I floor it?
Yes
You can likely increase spend 18% without marginal ROAS dropping below break-even.
current spendceiling
Expected extra profit: +€12,100/mo
Marginal ROAS 2.90 vs. break-even 1.61.
What the first month looks like
- Day 1
Connect Shopify, Meta and Google over OAuth – minutes, no code, no pixels to paste. Orders and spend start flowing the same day.
- Week 1
Contribution margin lands on every order, and the first verdicts appear: which campaigns keep money, which only move it. First-order economics are solid from day one.
- Week 4+
Cohorts mature. Now the verdicts see repeat behaviour – the campaign that looked like a winner on first orders gets re-judged on what its customers actually did next.
Honest note: cohort verdicts need history. The first weeks judge on first-order economics, and get sharper as repeat data arrives.
Fair questions.
▸How is this different from a tracking pixel or a CAPI tool?
Tracking tools improve the data going INTO the ad platforms so their algorithms optimise better – and you still judge on ROAS. Kepra is the judge: it pulls spend and orders out, subtracts COGS, refunds and fees, and tells you which campaigns keep money. The two are complementary, not substitutes.
▸Does Kepra change my campaigns automatically?
No, and it never will without asking you first. The AI proposes a specific change with its evidence; nothing executes until you approve exactly that change. Every action is logged and revertible.
▸What does setup involve?
OAuth sign-in to Shopify, Meta and Google – minutes, no code. Product costs come from Shopify where you maintain them; a blended margin covers the gap until then.
▸What if my orders can't all be attributed?
They can't – nobody's can. Customers clear cookies, switch devices and touch several channels. What cannot be matched is reported as unattributed rather than guessed, because an honest gap beats a confident fiction.
See it on your kind of shop first.
The demo is open – no signup. When the beta opens, early shops lock their price for life.
Also selling subscriptions? We put Stripe in the same profit view →