The scoreboard this product is built to move - true ROAS.
Platform-reported ROAS looks generous. Bank statements do not. The scaling engine measures the honest one - real revenue net of real cost - and compounds spend toward the campaigns actually earning it.
Real revenue, real cost, netted.
Platform-reported ROAS includes attribution that other platforms also claim. Add them all up and you get a return that is impossible on paper. The scaling engine ignores that game.
What counts as revenue is what actually landed in your store's order data. What counts as cost is what actually left your ad accounts. The engine reconciles them per product, per campaign, per creative - and only reports what nets out.
- Store-order revenue as the source of truth, not platform attribution
- Ad spend pulled directly from ad account billing, not estimates
- Refunds and returns subtracted from revenue on the same window
- Attribution windows are honest and disclosed - no post-hoc changes
Winners scale. Everything else gets tested or retired.
A product with strong true ROAS gets more budget. A product with a strong creative but poor conversion gets a landing-page signal, not more spend. A product with weak signal gets a fresh generation pass before being paused.
Scaling is compounding, not chasing. The engine biases toward proof - what has already earned spend, based on real revenue, gets more of it, at a controlled rate.
- Scale steps are capped - no overnight 10x that trips ad account risk
- Scaling factors include margin, stock depth, and refund rate - not just gross revenue
- Underperforming products get retagged, not just cut - so a re-test is possible
- Cross-product signal shared - a hook that wins on one SKU gets tested on siblings
- 1Established ROAS floor cleared
- 2Sustained over 3 days
- 3Winner promoted, budget +25%
- 4Held for 7 days, budget +25% again
- 5Compounding - cross-product test kicks in
Meaningful scaling takes weeks, not hours - and varies by product and market.
A campaign needs real spend and real signal before scaling decisions can compound. Products with tight margin, thin creative source material, or narrow addressable audiences will scale more slowly. Products that have already found product-market fit tend to scale faster.
First 7 days - signal gathering
week 1Baseline CTR, add-to-cart rate, and net revenue per creative are established. Scaling holds while data collects.
Week 2-3 - compounding starts
week 2-3Winning creatives and audiences accumulate budget. Fatigued variants get replaced. True ROAS starts to trend upward.
Week 4+ - scaling engine active
week 4+Cross-product signal, lookalike expansion, and multi-platform balance kick in. Compounding continues at a controlled rate.
Some things we will not do, on purpose.
Report platform-inflated ROAS as reality
honestyEvery ROAS figure the scaling engine acts on is store-verified. Platform reporting is shown for context - not used for decisions.
Scale past what your margin can carry
disciplineProducts with thin margin get lower spend ceilings by default. High-margin winners scale further, controllably.
Break platform account health
safetyScaling steps stay within platform-safe bounds - no overnight 10x that trips risk models and pauses your ad accounts.
Chase a metric that stopped mattering
restraintIf real revenue stops moving, scaling stops - even if platform reporting still looks good.
About the number that decides everything else.
How does Mablens actually know what my revenue is?
+
What attribution window do you use?
+
Can I see the difference between platform-reported and true ROAS?
+
What happens if the numbers look bad?
+
Move the number that actually pays the bills.
True ROAS. Netted. Honest. That is the scoreboard this whole factory is built to move.