Creative Strategy · Testing & Scaling
Restspan — Creative Testing & Scaling Plan
Continuing the Restspan creative brief: turning its three hooks into a structured test — budget, duration, and success thresholds set before launch — then a scale, iterate, or kill call for each, and a specific plan for scaling the winner.
What this is: the Restspan brief produced three hooks off one angle, built specifically so a test would isolate the hook as the only variable. This exercise runs that test: define the setup and the decision threshold first, read the results honestly even when the loudest-looking hook isn't the best one, and lay out exactly how the winner gets scaled — because "turn up the budget" is not, on its own, a scaling plan.
Before the test
Setup
Everything except the hook is held identical across all three ad sets, so the comparison is fair — same audience, same budget, same placements, same body and CTA.
What's isolated
Only the first 3 seconds (the hook) differs between ad sets — body, CTA, audience, and placements are identical across all three
Budget & duration
$50/day per ad set, 5 days minimum ($250/hook, $750 total) — no read before day 5, regardless of how a hook looks on day 1 or 2
Primary decision metric
Hold rate (thru-play ÷ 3-sec views) and CPA — not hook rate alone, since hook rate only measures whether people stop, not whether they stay or buy
Kill threshold, set before launch
Any hook with a Hold rate under 10% after the full $250 minimum spend is killed regardless of its Hook rate or CTR — a hook that can't hold attention past the open won't be rescued by more spend.
After the test window
Results
All figures below are illustrative example results for this exercise, not a real ad account.
| Hook | Spend | Impressions | Hook rate | Hold rate | Link CTR | LP views | ATC | Purchases | CPA | ROAS |
|---|
| 1 — "I used to think…" | $250 | 24,000 | 22% | 21% | 1.9% | 410 | 47 | 8 | $31.25 | 2.85x |
| 2 — "POV: it's 1am…" | $250 | 21,500 | 31% | 8% | 2.4% | 480 | 34 | 4 | $62.50 | 1.42x |
| 3 — "Nobody tells you…" | $250 | 22,600 | 26% | 14% | 2.1% | 445 | 41 | 6 | $41.67 | 2.14x |
The hook with the loudest open isn't the winner. Hook 2 has the best Hook rate (31%) and the best CTR (2.4%) of the three — the most scroll-stopping opening — but the worst Hold rate (8%, below the pre-set kill threshold), the fewest add-to-carts, and by far the worst CPA and ROAS. The same pattern shows up in
the FAST diagnostic exercise: the metric that's easiest to see first isn't the one that predicts the business result.
Decision
Scale, iterate, or kill
Hook 1 — "I used to think…"Scale
Hold 21% (best) · CPA $31.25 (best) · ROAS 2.85x (best)
Wins on every metric that actually matters once attention is spent — hold, CPA, and ROAS — despite the lowest Hook rate of the three. It's also the hook closest to the brief's stated voice (quiet, confessional), which is likely exactly why people who stop for it stay for it.
Hook 3 — "Nobody tells you…"Iterate
Hold 14% · CPA $41.67 · ROAS 2.14x
Profitable and roughly balanced across the funnel, but at about half of Hook 1's efficiency. Worth a second-round variant that keeps this hook's "nobody tells you" reveal but borrows Hook 1's slower, more confessional pacing — not worth discarding the angle outright.
Hook 2 — "POV: it's 1am…"Kill
Hold 8% (below the 10% kill threshold) · CPA $62.50 (worst) · ROAS 1.42x (worst)
Falls below the kill threshold set before the test started, so the call was made in advance, not in the moment. The punchy, meme-style open doesn't match the brief's "quiet confession, not influencer-polished" voice — and everything downstream of the hook shows it.
Scaling plan
How Hook 1 gets scaled, specifically
"Increase the budget" is not a scaling plan on its own — each step below exists to avoid losing the stable result the test just found.
1
Let it finish exiting the learning phase at the current $50/day for at least 3 more days before touching the budget — don't scale on day-5 data alone.
2
Raise budget in 20–30% increments every 48–72 hours, not in one large jump — a bigger move risks resetting the learning phase and losing the CPA just achieved.
3
Duplicate the winning ad set at similar budget rather than only inflating the original — this checks whether the result holds through a fresh learning phase, and protects the account if the original's performance drifts once spend rises.
4
Open a second test round once Hook 1 is stable at higher spend: two new hooks written in the same confessional voice, tested against Hook 1 as the new control — testing continues instead of stopping the moment one hook wins.
5
Kill Hook 2's spend immediately rather than letting it keep running "just in case," and carry Hook 3's pacing idea into the next test round rather than the current one.
Takeaway
What this teaches
A test only means something if the variable is isolated and the decision threshold is set before the results come in — otherwise it's easy to talk yourself into keeping whichever hook felt most exciting to write. And scaling is a separate discipline from testing: the winner still needs a careful, incremental rollout, or the same budget increase that should reward a good hook can just as easily break it.