Ninety days into a new marketing spend, somebody in the business asks the question. Maybe it’s your ops manager. Maybe it’s you, at a red light, thinking about the invoice that just cleared. Is this working?
Here’s what usually happens next. Nobody has a clean answer, because nobody wrote down, back on day one, what “working” was supposed to look like at this point. So the answer comes from mood instead of measurement. A good week makes you feel like it’s paying off. A slow week makes you feel like you got taken. Neither feeling is actually evidence.
That’s the gap this piece is for. Not a pitch for patience, not a checklist for catching bad vendors. A fair test you can run on any marketing investment, decided before the money moves, so the verdict at day 90 comes from a scorecard instead of a hunch. It applies to a new ad platform, a new agency, a new in-house hire running your marketing. It applies to us.
Decide the Test Before You Start
Most marketing verdicts get rendered the wrong way twice. Once too early, once too late.
The early version is the panic verdict. Three weeks in, cost per lead looks rough, and the whole thing gets written off as a waste before the campaign has even finished its first real learning cycle. You cut the thing that might have worked because the first data point looked bad, and first data points almost always look a little rough.
The late version is worse, because it’s quieter. It’s the sunk-cost patience that lets a channel run for nine months on the promise that it’s “still building” with nothing you can point to and measure. Nobody wants to be the one who pulled the plug too soon, so nobody pulls it at all, and the spend keeps going because stopping feels like admitting the first eight months were wasted.
Both failure modes have the same root cause: nobody agreed in advance what proof looks like, and at which date. Without that agreement, every check-in becomes a debate about whether now is “too early to tell” or “long enough to know,” and that debate never resolves, because there’s no pre-committed standard to resolve it against.
The fix is boring and it works. Before you spend the first dollar, write down what should be demonstrably true at day 30, day 60, and day 90. Not a wish list, a specific set of signals you’ll actually be able to check. Then judge the investment against that list, not against how the week is going. A verdict against a standard you set when you were thinking clearly beats a verdict against this morning’s mood every time.
The Scorecard: 30, 60, 90
This is the shape we use internally, and it’s built for paid acquisition specifically, since that’s the channel that’s supposed to move fastest. Treat the dates as checkpoints for a type of proof, not a countdown to a guaranteed number. Markets, budgets, and starting points vary too much for a single number to mean the same thing for every business.
Day 30: Infrastructure Proof
At 30 days, you’re not looking for volume. You’re looking for whether the plumbing is real.
Is tracking actually live and verified, meaning you can trace a specific lead back to a specific ad, not just trust that it’s happening somewhere in the background? Are the ads actually serving, with impression and click data flowing where you can see it yourself, not just in a screenshot someone sent you? Is the early cost-per-click number anywhere close to what was planned for, and if it isn’t, is there a plain explanation you can repeat back in your own words?
A pass at day 30 looks like: you can log in and see the numbers yourself, tracking matches a call you know happened, and the person running it can explain the early cost reality without reaching for a vague industry line. A fail looks like: nobody can show you raw numbers, tracking is “still being set up,” or the explanation for slow costs is a shrug.
Leads showing up this early is a bonus, not the test. Day 30 is asking whether the foundation is honest, not whether the machine is fast yet.
Day 60: Direction Proof
By day 60, the question changes from “is the plumbing real” to “which way is this trending.”
Lead flow should be established by now, meaning it’s happening on a regular basis, not arriving in one lucky burst. Cost per lead should have a direction, ideally down or flat as targeting sharpens, and you should be able to see which channel or campaign each lead actually came from. Waste should be visibly getting cut, meaning the obviously bad-performing pieces from week one are gone or changed, not still running on autopilot because nobody looked.
The trend line matters a lot more than the level at day 60. A cost per lead that’s still higher than you’d like but has been dropping for four straight weeks is a healthier sign than a lower number that’s been flat since day one, because flat this early usually means nobody’s actually working the account.
Day 90: Economics Proof
Day 90 is where you get to talk about money honestly, because you finally have enough volume to say something real.
By now you should have enough data to talk about cost per qualified lead, not just cost per lead, and ideally cost per booked job. Source attribution should hold up: you can point to which channel produced which job. And here’s the sentence that has to be on the table by day 90, stated plainly: marketing’s job is producing the calls, the close rate on those calls is a sales-side number, and if the calls are qualified but the close rate is weak, that’s a different problem than a marketing problem, and it deserves a different fix.
A day-90 keep, fix, or kill decision should write itself once you have those numbers in front of you. Keep it if cost per booked job pencils out against what the job is worth to you. Fix the targeting or the reporting if the picture is still murky through no fault of the channel itself. Kill it if 90 days of real spend produced neither volume nor a believable trend, and nobody can explain why.
What Doesn’t Count
A fair test also has to be specific about what it ignores, or every noisy week becomes a false alarm.
Single-day or single-week cost swings don’t count. Ad platforms are volatile day to day; a bad Tuesday means nothing on its own. One bad batch of leads doesn’t count either, whether that’s a run of tire-kickers or a string of out-of-area calls; every channel has an off week sometimes.
Vanity counts don’t count. Impressions as a primary metric is the tell to watch for specifically: if a report leads with how many people saw an ad instead of how many real calls it produced, that’s a report built to look busy, not a report built to tell you the truth. The same goes for follower counts and profile visits when they’re presented as evidence a channel is working.
And the traceability rule matters more than any single number: if a result can’t be traced back to a real call you can point to, it isn’t evidence yet, it’s a claim. We’ve written about this exact gap before: reports that can’t survive being checked against the phone log aren’t reporting, they’re theater.
When Not to Wait for Day 90
The 30/60/90 scorecard is the standard case. Some situations don’t earn the right to wait that long.
Pull the plug early if tracking still isn’t working by day 30 despite you asking about it more than once. Pull it early if nobody will show you raw numbers, only summarized slides. Pull it early if spend is climbing and the explanation for why keeps changing or never fully makes sense, twice in a row.
But there’s a false-alarm list too, and it matters just as much, because this is where fairness gets tested. A slow first two weeks is normal for a freshly launched campaign; platforms need a short settling-in period before performance steadies, and that’s real. The difference between a legitimate learning period and a stall tactic is simple: a legitimate one comes with actual dates and actual data attached to it. “The campaign needs more time” is a brush-off. “Here’s what we’re seeing week over week, and here’s when we expect it to stabilize” is a plan you can hold someone to.
The Channels This Test Doesn’t Fit
This test is built for paid acquisition, and it’s worth saying plainly where it doesn’t apply, because pretending it applies everywhere is how good work gets killed.
SEO and content don’t run on this clock. That work compounds over quarters, not weeks, the same honest timeline we’ve laid out here. Judging a content program at day 90 with the same standard you’d use on a paid campaign is one of the more common ways a business kills a channel that was about to start paying off, right before it would have shown up.
But that honesty cuts both directions. A vendor who claims everything they touch is “SEO, so it just takes time” is using the same slow-channel language to hide work that should be showing signals much sooner. If a paid channel is being defended with a content-timeline excuse, that’s a mismatch worth naming out loud, not a reason to keep waiting.
Fair on Day One, Sharp on Day Ninety
A fair test does two things at once. It clears a channel or a vendor doing honest work from a verdict rendered too fast on too little. And it catches the one that’s been hiding behind vague language for three months with nothing to show for it. Both outcomes come from the same scorecard, decided in advance, applied the same way regardless of who’s on the other end of it.
That’s the whole point of writing it down before the money moves. You’re not hoping for a good outcome. You’re setting a standard good work can clear and bad work can’t hide from.
Hold Us to the Same Test
Run this on your next marketing investment, and run it on the one you already have. If you want to see what the 30/60/90 scorecard looks like applied to a real plan for your business, that’s what kodiakdigital.co/plans/ is built to show you: what should be true at each checkpoint, in writing, before you spend a dollar. Hold us to the same standard this piece just laid out. That’s the deal.