Welcome back to Beyond ROAS, CE Digital's weekly newsletter on what's actually working in paid media right now.
This week we're getting into the test budget math most brands have never run, why AI will scale your weaknesses if you let it, a look at the persona tool we've been building inside AdSignl, how we think a media buyer should actually be evaluated, and the three-pillar audit to run before you make any hiring decisions.
Let's get into it.
01 — The test budget math most brands haven't run
Most brands have no idea how much of their ad budget is actually going toward finding new winners.
We built a model to answer that question for every account we manage. The number that comes back is almost always uncomfortable.
The account this data is pulled from is currently allocating 5.7% of total spend to testing. At that rate, the math works out to roughly 62 tests per month, 12 expected hits, and 3.5 creatives graduating to scale. Per month.
3.5 new scale ads. In an account churning through dozens.
That's not a creative quality problem. That's a volume problem created by underfunding the one part of the system that feeds everything else.
Here's what the model shows when you move the lever:
At 10% — 124 tests per month, 7 creatives graduating to scale At 15% — 186 tests per month, 10.6 graduates At 20% — 248 tests per month, 14 graduates
The output nearly quadruples. The total budget doesn't change. You're just redistributing where it goes.
Most brands treat test budget as a cost. The accounts that actually scale treat it as the rate-limiting input, because it is. Everything downstream depends on what you're feeding into the top of the pipeline.
The model tells you exactly what your current allocation is producing and what a different number would buy you. Most teams just haven't run the math.
02 — AI scales your weaknesses if you let it
AI is not Artificial Intelligence. It's Augmented Intelligence.
"Artificial Intelligence" implies the intelligence is manufactured from nothing. That if you give it to your team, the output improves automatically. It doesn't work that way. The output will only ever be as good as the logic you put in.
Which means if your inputs are weak, AI scales your weaknesses.
We audit every client account across five input dimensions before we touch a model:
Data quality. Are you working from clean attribution and a single source of truth? Or mixing click windows, view windows, and Meta's reported numbers and then wondering why the recommendations feel off?
Strategy clarity. Do you have defined KPIs with decision rules attached? Or goals that exist as statements in a deck nobody revisits?
Persona depth. Demographics are not a persona. Age and gender tell you who might see the ad. Psychographics, objections, and language triggers tell you what to say to them.
Brief structure. Is your brief format linked to performance data? Or a creative wishlist that changes based on who's in the room that week?
Feedback loops. Does what performs well automatically inform the next brief? Or does every brief start from scratch?
In almost every account we audit, the weakest area is the same one: feedback loops. The context layer isn't getting richer over time. It resets. And that's the one that compounds everything else.
The variable isn't the model. It's what you're feeding it.
03 — What if you could ask your customer "would you click this?"
What if you could pitch a hook to your target customer before you ever spent a dollar?
That's what we've been building at CE Digital with AdSignl. We build each buyer persona as an interactive AI model trained on everything we know about that customer. Not just demographics. Psychographics, buying objections, creative preferences, the language that turns them off, what social proof actually moves them.
You pitch it an idea and it responds as that person. Which means you can stress-test a hook, a concept, or an angle against your actual customer before anything goes into paid media.
Real example. One of the personas we built for a baby brand is called "The Mindful Mom." Her filter for everything: "Can I do this while my kids are screaming in the background?" If it doesn't make the morning routine faster or put the baby down easier, it's not for her.
When we pitch that persona a hook, it tells us if it lands.
Hook: "Zip. Done. Back to sleep, for both of you." Response: That's my life. I've snapped onesie buttons at 3am half-asleep. A zipper that actually works is not a luxury. I'd stop scrolling. Score: 91.
Hook: "Wrap your little one in the softest bamboo, they deserve the best." Response: "They deserve the best" is guilt-bait and I see through it immediately. Just tell me it washes well and doesn't shrink. Score: 17.
Same product. Same account. Two completely different outcomes based on language alone, and we knew before spending anything.
It's not replacing the strategist. It's giving your team a gut check against the actual customer instead of the loudest voice in the room. And it gets sharper over time as we keep feeding it top-performing hooks, winning concepts, performance data.
04 — How we actually evaluate a media buyer
Most media buyers are being graded on the wrong thing.
ROAS up? Good month. ROAS down? Bad month.
That's not performance management. That's holding someone accountable for market conditions they didn't create.
Here's what we actually measure:
Cut timing. Most tools are recommending cuts at 5 to 7 days in. The bigger issue we keep finding in audits isn't slow cuts, it's false kills. Ads getting pulled before they've had a real chance to spend, because they didn't show strong ROAS immediately. Cutting too early costs you just as much as cutting too late.
Graduation rate. This is the one most teams completely miss. Of all the creatives that hit your targets in testing, how many actually made it into the scale campaign? We cross-reference HIT ads against creative names in the scale campaign and the gap is almost always bigger than expected. Winners sitting in testing, never moved, never amplified. That's money being left on the table in the most literal way possible.
Test velocity. New creatives launched per week. Most agencies ignore this entirely. We target 50. The accounts that hit it consistently outperform the ones that don't, regardless of budget level.
Hit rate trend. Not the raw number, the direction it's moving. Rising means the input layer is improving. Flat or declining means something upstream is broken.
These are operational metrics. They measure decision quality, not market outcomes. That distinction matters because outcomes are partly luck. Decision quality is entirely within your control.
05 — The three-pillar audit: run this before you hire
Before you hire another media buyer, run this audit.
Creative. Are you launching enough volume to generate real signal? Can you turn a winning insight into a new creative within 48 hours? If not, the pillar is weak, regardless of how good individual ads feel.
Data. Do you have one source of truth, or are you averaging across conflicting attribution windows? Are your cut and scale thresholds written down and followed, or do they live in someone's head?
Capital. Can you move budget to a winner within hours, or does it require a multi-step approval chain? What percentage of spend is in testing versus scaling? Under 15% and you're underfueling the discovery engine.
Strong creative, weak capital allocation means you're finding winners and can't amplify them. Lots of capital, weak creative means you're spending fast to confirm what doesn't work.
Most brands in that second pattern think they have a media buying problem. They have a pipeline problem. Not enough creative volume.
In almost every account we take over, one pillar is significantly weaker than the other two. The brand is blaming performance. The problem is infrastructure.
Identify the weakness. Fix your scaling problems.