Scale spend without crashing ROAS
Most brands that "can't scale past $X/day" don't have a budget problem, they have a pacing and creative problem. Jump spend too fast and you reset the platform's learning phase, starve the algorithm of stable signal, and watch CPA climb right when you're trying to prove the channel.
AI can turn your last two weeks of stats into a scale/hold/cut call per campaign; the learning-reset rules and your ROAS must come from the platform and your own dashboard, never a guess.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are my scaling analyst. Here are the last 14 days per campaign, pasted from Ads Manager: [campaign | daily budget | spend | revenue | ROAS | CPA | frequency]. My break-even ROAS is [break-even].
1. For EACH campaign decide SCALE, HOLD, or CUT: scale only if it clears break-even with room and frequency is healthy; cut if ROAS is below break-even or frequency is climbing with falling CTR.
2. For scale campaigns, propose the next budget step at no more than +20%, and confirm the current learning-phase-reset guidance by opening the platform's help page. Do NOT invent the reset rule.
Output ONE markdown table: Campaign | ROAS vs break-even | Frequency | Decision | Next budget.
Use ONLY the numbers I pasted; the reset guidance from the platform page. Do not guess or fill in a missing campaign; leave blanks and ask me. Cite the platform page for the reset rule. If you can't browse, leave that rule blank and name the page.
Or do it in 5 steps
- Raise budget ~20% at a time. Step up in ~20% moves, at most once every 2-3 days, and only after the current level holds. A bigger jump, or changing bidding/targeting/creative at once, restarts the learning phase (confirm the reset rule on the platform's help page). At this pace, $100/day climbs to ~$300/day over a few weeks with no reset.
- Scale horizontally before vertically. Before pushing one campaign's budget higher, add new audiences, lookalikes, placements and fresh creative angles. Widening the base lets the account absorb spend without frequency and fatigue spiking; only then push budget up.
- Watch marginal ROAS, not blended. Each budget increase reaches a slightly colder audience, so the next dollar earns less than the last. Track contribution profit at every step; stop raising when marginal ROAS approaches your break-even line.
- Feed the creative engine. Creative volume is the real ceiling. Ship fresh variants continuously, graduate winners into the scaling budget within ~48 hours, cut losers within a few days.
- Run a weekly scale/hold/cut ritual. Once a week, sort campaigns: scale the ones clearing target with headroom, hold the borderline, cut the ones below break-even. Also check that backend (Shopify/CAPI) numbers track platform-reported ones within ~10-15%.
Done looks like: a paced budget ladder, audiences widened before budget pushed, and a standing weekly scale/hold/cut review.
Worked example: a 4-week $100 → $300/day path with gates (illustrative)
| Week | Daily budget | Gate before increasing |
|---|
| 1 | $100 | baseline holds above break-even, frequency < ~2/week |
| 2 | $150 (+50 over 2 steps) | marginal ROAS still above break-even |
| 3 | $220 | new audiences/creatives added; CTR steady |
| 4 | $300 | contribution profit still positive at higher spend |
If any gate fails, hold at the current step until it clears; don't average through a CPA spike hoping it self-corrects. Numbers are illustrative, run your own 14 days through the prompt above.
Review weekly; patience beats a big jump.