When to switch a campaign off instead of tuning it
Tell the difference between a campaign that needs more work and one that should be stopped, and set the kill criteria before you have money at stake.
The campaign that gets fixed every week
It has been running for two months. Every week brings a change. New headline. Tighter radius. Bid strategy swapped. Audience narrowed and then widened again. Cost per lead moves a little in both directions and nothing settles. Everybody involved is working hard and the account is going nowhere.
This is the most common failure in paid media and it is not a skill problem. It is the absence of a stopping rule. Without one agreed in advance a campaign is never dead. It is only between adjustments and it will consume budget for as long as somebody has an idea left.
What continuing costs you beyond the spend
Tuning cannot fix things that sit outside the ad account. It will not repair an offer the market rejects. It cannot create an audience that does not exist on the platform. It will not produce a profit on a product whose margin is smaller than the cost of a click. Begin by establishing that the problem is in fact inside the account because every week spent tuning assumes it is.
The direct cost of continuing is the daily spend. The larger cost is opportunity. The same money applied to a different channel or a better offer might have told you something. There is also a data cost. A long-running campaign with weak conversion history teaches the platform's bidding to find the wrong people and that learning does not vanish when you finally change the creative.
Which campaigns deserve patience and which do not
Leave it alone if it launched recently. A campaign still gathering conversions is unstable by design and judging it early then changing it resets the process. Leave seasonal campaigns running through their season because mid-season performance says little about the full window. Never switch off branded search. It defends demand that already exists and it is usually the cheapest traffic you own.
Switch it off when the constraint is outside your control and will stay there. If the enquiries are good and nobody is calling them back then the campaign is not the problem and spending while that is true is pure waste. Switch it off when the product is being discontinued and when the geography has no delivery capacity and when you cannot name a single untried structural change. A business that keeps a campaign alive because stopping would mean admitting the quarter's budget achieved nothing is paying for comfort rather than customers.
How to make the decision mechanically
Write the kill criteria before launch. A cost per qualified outcome you will not exceed. A minimum number of conversions you need before judging anything. A review date. All three written where other people can see them.
Then check volume before you check performance. With very few conversions a week the swings you are reacting to are noise and no change you make is measurable. If the budget cannot produce enough events to judge then the honest conclusion is that this campaign cannot be optimised at this spend and you either raise the spend or stop.
Separate structural changes from cosmetic ones. Offer, audience premise, platform and landing page are structural. Headlines, bids and schedules are cosmetic. Allow yourself a fixed number of structural attempts and keep a one-line record of each with its result. When that list is exhausted the campaign is finished and the record makes saying so straightforward rather than personal.
The number that decides and the number that stalls you
The deciding number is cost per qualified outcome against the ceiling you set in advance. Qualified means the sales team agrees it was worth their time. The ceiling comes from your margin and not from a benchmark somebody quoted.
The number that stalls you is week-on-week percentage change. On small volumes it is almost always impressive in one direction or the other and it is nearly always noise. Four leads against seven leads reads as a large improvement and means nothing. Click-through rate is the other trap. It can be improved almost indefinitely by better copy while cost per customer stays exactly where it was.
The common mistake
The expensive error is tuning cosmetics when the problem is structural. Bids get adjusted and headlines rewritten for months while the real issue is that the landing page never mentions the price or that the audience was chosen by assumption and never tested against a different premise. Cosmetic changes are easy to make and easy to report which is exactly why they fill the weeks.
The sunk cost version is worse. Once a meaningful budget has gone in, stopping starts to feel like writing off the investment, so the campaign is defended rather than assessed. The money already spent is gone either way. The only live question is whether the next rupee has a better home.
Switching off without losing what you learned
Pause and archive rather than delete. Keep the audiences and the conversion history and the creative library because a later campaign can use all three. Write down what you tried and what it cost so the next person who has the same idea inherits the answer instead of repeating the experiment.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.