There's a tension anyone managing online advertising lives with, from a small business owner reviewing their own campaigns to an agency manager handling six-figure accounts: the campaign has gone a few days without hitting the expected numbers, and a decision has to be made about whether it's time to pause it or whether to hold on a bit longer. Pausing too soon throws away the accumulated learning and the budget invested in getting there. Holding on too long burns money on something that, with enough data, was already clearly not going to work.
There's no infallible mathematical formula, but there are fairly reliable signals that help distinguish a normal rough start from a real problem, and that's exactly what this article aims to give: concrete criteria, not a vague intuition of "give it a bit more time".
Why almost every campaign performs poorly at first
Auto-bidding ad platforms (Google Ads, Meta Ads) go through a learning phase when launching a new campaign or making a significant change to an existing one (a big budget change, new targeting, a new optimisation goal). During that phase, which usually lasts between five and seven days or until reaching roughly 50 conversions (whichever comes first), the algorithm is still feeling out who to show the ad to, and performance tends to be more expensive and inconsistent than it will be once stabilised. Judging a campaign within this window is the most frequent and most costly miscalculation businesses managing their own ads make.
Signs that you should hold on
The first sign it's worth waiting is that the campaign is inside or just past the learning phase: if it's been running less than a week or has accumulated fewer than 50 conversions, any conclusion is premature. The second is that, even if cost per result is high, the trend within that first week is improving day by day, not worsening: that indicates the algorithm is finding its footing, it just needs more time. The third sign, more qualitative, is that the negative comes alongside positive signals in other metrics (good CTR, good engagement rate) suggesting interest does exist, and that the problem may lie further down the funnel (landing page, checkout process) rather than in the campaign itself.
Signs that you should pause
The first clear sign to pause is a cost per result that, once past the learning phase (a week or 50 conversions), remains well above what the business can afford to pay for that conversion, with a stable or worsening trend, not improving. The second is an abnormally low CTR from day one that never picks up: that indicates the creative or message isn't connecting with the audience at all, a problem that time alone rarely fixes, and that requires an active change, not patience. The third is spend that has already exceeded, without results, the figure you set as a reasonable test limit before starting (something worth setting before launching any campaign, precisely so you don't make this decision in the heat of the moment and out of anxiety).
The reasonable test limit: set it before launching, not during
The best defence against indecision (both pausing too soon and holding on too long) is deciding, before launching the campaign, how much budget you're willing to "risk" as a learning cost, with a clear head, without the pressure of the moment. A reasonable reference: the equivalent of 10 to 15 conversions at the cost per result you consider acceptable. If you reach that limit without clear signs of improvement, pause without guilt; if you haven't reached it, resist the temptation to touch the campaign too early.
Pausing doesn't always mean deleting
Pausing a campaign and deleting it are different decisions, and it's worth not confusing them. A paused campaign keeps its learning history for a while (several weeks, depending on the platform), so if the problem turns out to be temporary (a slow time of year, a one-off website issue) it can be reactivated later without losing all the previous work. Deleting a campaign and creating a new one from scratch, on the other hand, restarts learning completely.
What to do instead of pausing entirely: the middle-ground adjustment
Before pausing completely, sometimes an intermediate step is worth it: cutting the budget in half instead of eliminating it, to keep gathering data with less financial risk while preparing a creative or offer change. This works especially well when the warning sign is ambiguous (neither clearly positive nor clearly negative) and when the business can afford that reduced spend margin while deciding the next step.
A case where holding on one more week changed the outcome entirely
A baby products store launched a new Meta Ads campaign for a product recently added to its catalogue. Over the first five days, cost per sale hovered around 45 euros, well above the 20-euro ceiling the business had set as a reasonable limit. The team's initial reaction was to pause the campaign immediately, but since the pre-set test spending limit (the equivalent of 15 conversions at the target cost, around 300 euros) hadn't yet been reached, they decided to hold on two more days before making the final call.
In those two extra days, coinciding with the end of the algorithm's learning phase, cost per sale dropped sharply to 16 euros, below the target, and stayed stable in that range for the following weeks. Had the team paused the campaign on day five, as their initial gut reaction suggested, they would have written off a product that ended up being one of the most profitable of the quarter, simply for not having waited for the algorithm to finish finding its audience.
How to tell a normal rough start from a real problem without relying on gut feeling alone
Beyond the signals already mentioned, a practical trick is watching the daily trend within the learning phase itself, not just the cumulative average. If day five's cost per result is better than day three's, and day three's was better than day one's, that downward trend is a more reliable sign the algorithm is finding its footing than the week's overall average, which can still look bad even though the direction has already turned.
The emotional cost of indecision, and how to neutralise it
Beyond the technical criteria, there's a human component to this decision rarely mentioned: reviewing a campaign that's going badly generates anxiety, and that anxiety pushes toward faster-than-reasonable decisions, usually toward pausing too soon. Setting the test limit before launching, as explained, is precisely how you neutralise that emotional bias: the decision is already made in advance, with a clear head, and reviewing the campaign becomes checking whether an already-agreed number has been reached, not deciding under pressure every time you open the results panel.
An additional trick that helps maintain that discipline: share the test limit with someone else on the team before launching the campaign, so the decision to pause or hold doesn't depend solely on one person's mood on a given day.
How to document the decision so you don't repeat the same debate every time
Every time a campaign goes through this dilemma, it's worth briefly noting what was decided and why, even if the decision was simply "wait two more days". Over time, that record becomes your own history of good and bad calls, far more valuable than any generic rule from an article, because it reflects the specific behaviour of your own account, your sector and your product, which rarely matches the general average one hundred percent.
Comparison: pausing entirely versus gradually reducing budget
When a campaign shows mixed signals (neither clearly positive nor clearly negative), pausing it entirely isn't the only middle-ground option available. Cutting the budget in half, instead of pausing completely, has an underrated advantage: it keeps the algorithm's learning phase alive at lower financial risk, instead of resetting it entirely when the decision is made to reactivate later. The trade-off is that a reduced budget also slows how fast data accumulates, so the final decision (keep, scale, or truly pause) takes somewhat longer to make with confidence. For businesses with tight cash margins, where every euro in a questionable campaign carries a lot of weight, a full pause is usually the more prudent option despite this downside; for businesses with more financial cushion, a gradual reduction usually better protects the learning investment already made.
Frequently asked questions
What's the minimum time to give a new campaign before evaluating it?
At least seven full days (to cover a complete weekly cycle) or until accumulating around 50 conversions, whichever comes later. Evaluating before that almost always leads to decisions based on statistical noise, not a real trend.
Is the criterion for pausing a brand awareness campaign different from a direct sales one?
Yes. In awareness campaigns, the relevant metrics (reach, frequency, brand recall) take longer to show tangible results, so the reasonable patience margin is usually longer, three to four weeks instead of one or two.
What if I have to pause due to lack of budget, not because the campaign is underperforming?
That's a different scenario and it's worth documenting it as such (not confusing it with a campaign failure), so that when you resume it once budget is available you can still benefit from the previous learning, provided the history is preserved within the platform's retention window.
Does pausing and reactivating a campaign several times hurt its performance?
Yes, each reactivation after a prolonged pause usually resets part of the learning phase, so it's worth avoiding frequent pauses and reactivations if possible, and preferring a downward budget adjustment over a full pause when the goal is only to temporarily reduce risk.
Should I pause the whole campaign or just the ad or ad set that isn't working?
Whenever possible, it's better to pause only the specific element (an ad, a creative) that isn't performing, keeping the rest of the structure that is working active, rather than pausing the entire campaign and losing the accumulated learning in the parts that are doing fine.
What role does seasonality play in this decision?
An important one: a campaign that seems not to be working might simply coincide with a normal low-demand period in your sector (August for many B2B businesses, for example). Before pausing due to weak performance, it's worth checking whether the same pattern repeated the previous year around similar dates.