How long should you wait before judging a new campaign?

“It has been two weeks. Is it working?” It might be the most common question in paid media, and it usually gets asked at exactly the wrong moment. Two weeks feels like a fair trial. Sometimes it is. Often it’s a verdict on a campaign that hasn’t collected enough data to say anything yet, and switching it off then throws away the money it took to get that far.

The honest answer isn’t a number of days. It’s a number of conversions, plus the time your customers take to buy.

Why this matters

Why every answer you find is different

Search the question and the answers range from a week to a year. LeadEnforce suggests 7 to 10 days for a conversion campaign on Meta (LeadEnforce). Brew Digital recommends at least a month, and up to three, before making significant changes (Brew Digital). Impremis puts most campaigns at 45 to 90 days, with milestones every 30 (Impremis). Olivetree says most campaigns take three to six months (Olivetree), Zimmer Communications says to allow at least three months before making significant changes to a digital campaign (Zimmer Communications), and Insil talks about six to nine months, with 12 before you reassess, though it notes paid search can show some results almost immediately (Insil).

They aren’t really disagreeing. They’re answering different questions. Some are talking about brand building, which Zimmer itself says can take 12 months or more to work in a measurable way (Zimmer Communications). Some fold SEO, which Olivetree puts at four to 12 months, into the same answer as paid ads (Olivetree). And most blur two separate questions: when will this campaign pay for itself, and how soon can you tell whether it will?

For paid ads, the second question has a much sharper answer. Olivetree puts it well: paid ads bring traffic within hours, but need several weeks of data and optimisation before they become cost-efficient (Olivetree). The traffic is instant. The evidence isn’t.

Count conversions, not days

Rook Media says it most bluntly: a campaign with three conversions has told you almost nothing however long it ran, while one with forty has told you a great deal in a week (Rook Media). We agree, and the platforms say much the same.

On Meta, an ad set usually leaves the learning phase after about 50 results in the week after its last significant edit (Meta). LeadEnforce suggests 50 to 100 conversions per ad set before judging a conversion campaign (LeadEnforce). We covered why most accounts never get there in our piece on the learning phase.

Google is more specific than most people realise. Its help centre says Smart Bidding typically takes one to two conversion cycles to settle on a new goal, and that performance should then be judged over two conversion cycles or two weeks, whichever is longer, allowing for conversion delay (Google). It also suggests looking at a month of data for a clearer picture (Google). And the “Learning” label isn’t the verdict: it shows for up to seven days, while the bidding keeps learning in the background long after it disappears (Google).

Your budget sets the speed

So the real deadline is set by your budget and your cost per result, not the calendar.

Take a lead gen campaign at $60 a lead. On $3,000 a month it buys about 50 leads, roughly 12 a week, so it takes about a month to collect enough to read. On $15,000 a month, the same campaign gets about 58 leads a week and has that much data in under a week. Same campaign, same question. One answer arrives five times faster than the other.

The buying cycle still sets a floor. A big budget gets you the data sooner. It doesn’t make your customers decide any faster.

Why the last few days always look worst

One more trap, and it catches experienced marketers too. Google counts a conversion against the date of the ad, not the date of the sale, so the most recent days are always missing sales that are still on their way (Google). That conversion delay makes your cost per acquisition look higher and your ROAS look lower than they really are (Google). Some conversions take days, or even weeks, to come in (Google).

If your customers usually take ten days to buy, a two-week review is judging a fortnight in which only the first few days are anywhere near complete. LeadEnforce gives the example of a software free trial that may need ten days before results show, because people browse, compare and come back later (LeadEnforce).

What to look at, and when

Here’s the rhythm we use. The calendar points are rough. The real trigger is whichever comes later: enough conversions to read, or enough time for your customers to buy.

Week one: is it working, not is it worth it. Check the machinery. Is tracking firing? Is the budget spending? Are the ads approved, the search terms sensible, the landing page loading properly on a phone? Brew Digital frames its daily checks the same way, around delivery and budget rather than results (Brew Digital). Fix anything broken. Leave everything else alone.

Month one: the early signals. By now, most campaigns on a reasonable budget have enough data to show a direction. Is the cost per result moving towards target? Are people clicking, and is the page converting them? For lead gen, are the leads any good? Olivetree calls these leading indicators, the signals that show up before the revenue does (Olivetree). This is where small adjustments start to make sense (Brew Digital).

Month three: the business verdict. Judge the campaign on what the business made, not on what the platform reports. Brew Digital also puts the better-informed decisions at the two to three month mark (Brew Digital). If you sell to other businesses, even month three can be early. Olivetree, citing Gartner, puts a typical B2B purchase at six to ten decision-makers over four to six months (Olivetree). In that case, month three tells you about the pipeline, not yet about the revenue.

We saw how much the measure matters in a US lead gen pilot on Meta. The client wanted to know whether Meta could deliver leads at a defensible cost before committing more budget. The existing setup was already producing leads. The sales team just couldn’t close them (Dadek Digital). Over a 56-day period, high-quality leads went from 72 to 139 and the cost per lead fell 28%, from $41.97 to $30.18, as spend grew from $3,026 to $4,195 (Dadek Digital). The pilot was judged on high-quality leads, and on whether the close rate held as spend grew. It did (Dadek Digital).

What people overestimate

That two weeks is a test

Two weeks is a fine window for a campaign with plenty of volume and a short buying cycle. For most campaigns on a $5,000 to $30,000 budget, it’s a sample, not a test.

The damage usually comes from what happens next. Rook Media describes the pattern it finds in the change history of every struggling account it takes over: judge during the learning period, change everything, then judge the result of that (Rook Media). Each round of changes restarts the clock. On Meta, significant edits, including frequent budget changes, can send an ad set back into learning (Meta). On Google, budget and target changes don’t trigger a “Learning” status, but performance still takes a conversion cycle or two to settle on the new setting (Google).

That patience is always the safe choice

The opposite mistake is just as expensive. “Give it three months” can be good advice for brand campaigns, and Zimmer is right that many campaigns start to gain traction just as clients lose patience (Zimmer Communications). But a campaign spending well past break-even with no sales doesn’t need three months to give you an answer.

Rook Media’s rule makes a good backstop: spend two to three times your break-even cost per acquisition with no conversions, and that is an answer too (Rook Media). LeadEnforce makes the same point: there’s no value in keeping an ad running just because it might improve later (LeadEnforce).

Patience without a kill number isn’t a strategy. It’s hope with a budget.

That the platform’s number is the verdict

Every platform reports the conversions it can see and gives itself the credit for them. That’s useful for steering a campaign and a poor way to decide whether it was worth it. Judge a new campaign on the whole business: total revenue against total ad spend over the same weeks, new customers, and for lead gen, what the sales team actually closed.

Don’t judge it on the cheap metrics either. Google says clicks, cost per click and click-through rate can move around while Smart Bidding works towards your main goal (Google), and tells advertisers not to evaluate on trailing metrics like cost per click and impressions (Google).

If the platform looks weak but total revenue rose, look harder before you switch anything off. If the platform looks brilliant and the bank account doesn’t, that’s the bigger warning.

Before you launch the next one

Write three things down before the campaign spends its first dollar. The result that means it worked. The result that means stop. And the point at which you’ll decide: a conversion count (around 50, the number Meta uses for learning, is a sensible floor) or two full buying cycles, whichever comes later. Put that date in the calendar and leave the campaign alone until it arrives, unless something is broken.

That turns “is it working?” from a feeling into a check. And the next time someone asks at the two-week mark, you’ll already know whether it’s too early to answer.

Dadek Digital agrees the target, the stop number and the review date before a campaign launches, and judges it on the business’s own numbers rather than the platform’s. If you’re two weeks into a campaign and not sure whether to hold it or switch it off, a paid strategy session is the place to start.

Next
Next

How many ads should you test? The honest answer for a $5k to $30k budget