Campaign Kill-Point Calculator: When to Cut an Offer

How much should you spend before killing a campaign?

The floor for judging a cold-traffic campaign is 2 to 3 times the offer's payout, spent with zero conversions and no supporting signal further up the funnel. If a $40 payout offer has burned $100 and the landing page is converting at your historical average, that is a small-sample fluctuation, not a verdict. If the same $100 spent shows landing-page conversion at half of normal, kill it now. You already have your answer.

Buyers who run this daily, the kind who post testing budgets on STM Forum rather than guess out loud, set the ceiling before the campaign launches, not while watching the dashboard refresh mid-afternoon. That habit alone prevents the most common leak in affiliate accounts: extending a losing campaign because the last $20 spent produced a nibble of hope. Payout size sets the unit. A $15 payout offer earns a $30-45 kill window before you even open funnel data. A $150 payout nutra offer on native buys $300-450 of runway, because the sample needed to see one conversion scales with payout.

Calculate this the night before launch, not during. Write the number down somewhere outside the ad platform. Walk away from the dashboard until you hit it.

What kill thresholds do professional buyers pre-commit to?

Professional buyers commit to threshold tiers by traffic type before the first dollar spends, because cold, warm and retargeting traffic need different sample sizes before the numbers mean anything. A retargeting pool of people who already visited the landing page converts at multiples of cold traffic, so a single flat rule punishes it unfairly. The tiers below reflect discussion common in practitioner forums and roughly track what Meta's own documentation implies about minimum sample size: its learning-phase guidance calls for around 50 optimization events in a rolling 7-day window before an ad set's delivery stabilizes, which is itself a floor on how much data you need before trusting anything the dashboard shows you.

Treat these as starting ranges, not law. A buyer with three years of data on one vertical can tighten them safely. A buyer testing a brand-new niche should widen them, because the true conversion rate is genuinely unknown, not just unlucky today.

When is zero conversions actually still normal?

Zero conversions stays statistically normal for longer than most affiliates assume, and the window depends on the true conversion rate you have not observed yet. Statisticians use a shortcut called the rule of three: with zero events across n trials, the upper bound of the 95% confidence interval for the true rate is roughly 3 divided by n. Translated to media buying: if a landing page needs around 100 visitors to reliably convert once, seeing zero sales in 60 visitors sits comfortably inside normal variance. Seeing zero in 400 visitors does not.

This is why spend-based and click-based kill rules disagree at low volume and agree at high volume. A $40 payout at $1.20 per click needs roughly 33 clicks to reach a 1x-payout spend threshold. If landing pages in that category typically convert around 2%, expected conversions at 33 clicks sits well under one, so zero conversions there is close to meaningless. The same zero conversions at 300 clicks, with roughly 6 expected, is a real signal worth acting on.

Nobody runs this arithmetic live in their head, and nobody needs to. A calculator that takes payout, spend and conversion count and checks it against a category's typical conversion rate does that work automatically, and returns a plain verdict on whether zero is still normal or already a decision.

How do you separate a bad offer from bad creatives?

Separate the offer from the creative using two numbers instead of total ROAS: click-through rate against your own account average, and landing-page conversion rate against the category norm for that type of offer. Strong CTR paired with landing-page conversion under half the category norm points at the offer or its page, not the ad. Weak CTR paired with landing-page conversion that matches or beats the norm points at the creative.

Most affiliates skip this split and test blind. They run one creative, see a bad ROAS, and conclude the offer is dead, when the ad itself was the weak link the entire time. Pulling click and landing-page numbers apart is the only way to know which lever actually failed, and it costs nothing beyond reading numbers already sitting in the dashboard.

Here is the harder part. Once that split is done, stop iterating creative on a single offer past two or three variants. The sample size needed to detect a genuine difference between two creatives at typical affiliate spend is larger than almost anyone's testing budget allows — the same rule-of-three math above cuts both ways, and a 2-point CTR gap on 40 clicks per variant is noise dressed up as insight. Dollar for dollar, a second offer in the same vertical is usually cheaper information than a fourth creative variant on the same one. Most buyers resist that trade, because a new creative feels like progress and a new offer feels like starting over, but the underlying arithmetic is identical to the arithmetic that sets kill thresholds in the first place.

What should you salvage from a killed campaign?

A killed campaign still owes you three things: audience data, landing-page learnings, and a documented reason for the kill. Save the pixel or engagement audience where the platform allows it. Meta lets you save a custom audience from ad engagement even after a campaign pauses, and that audience can seed a lookalike for the next test in the same vertical. Landing-page click-to-lead or click-to-sale data tells you whether traffic quality was the real problem across two different offers, which is a durable signal about your traffic source, not just about the one campaign that died.

Write down the reason it died in one line: creative fatigue, landing-page conversion below norm, payout cut mid-run, network changed the offer's terms. Six months from now, when a similar offer from the same vertical shows up, that line is the only thing standing between you and repeating the exact same test.

Do not salvage the ad account's spend history as a badge of effort. A long list of killed tests is not a strategy. It is a record of guesses, unless each one comes with a specific, written reason attached.

How do you pick the next offer with better odds?

Pick the next offer by checking two things: a payout-to-cost ratio that clears your kill-threshold math comfortably, and evidence the offer still has room before saturation. ClickBank's marketplace gravity score and Digistore24's bestseller ranking are the two most-cited public proxies for how much competing traffic an offer already carries. Neither is a conversion-rate guarantee, but both indicate whether you would be the fifth buyer on an offer or the five-hundredth.

Newer offers with rising gravity and a payout clearing 3x your typical cost-per-click-to-conversion give the best odds, because you are testing against less accumulated market fatigue. An offer with a high, flat gravity score sitting unchanged for eight months has likely been tested by every buyer who could find it; whatever creative angles worked have already been copied into the ground by competitors with bigger budgets than yours.

Spy tools like AdSpy, priced by seat on a published pricing page and updated on a scheduled crawl, are useful for surfacing volume and rough creative angles on a given offer. The number of live ads visible in a tool is not the same as evidence of live profitability, so cross-reference that view with the network's own trending or gravity data before committing budget you just freed up from the offer you killed.

Move the exact dollar figure you set as this campaign's kill threshold onto the next offer's test plan, unchanged. If $300 of runway was needed to judge fairly last time, the next offer gets the same $300 before you touch the multiplier. Consistency in the threshold, not the offer, is what turns a string of kills into a working process instead of a string of guesses.

Frequently asked questions

What is a good kill threshold for a cold-traffic ad campaign?

Kill a cold-traffic campaign once spend reaches 2 to 3 times the offer's payout with zero conversions and no supporting signal in click-through rate or landing-page conversion. Retargeting and warm audiences can run to 4-5x payout before the same verdict applies, since that traffic is smaller and more qualified per click.

Is zero conversions after $100 in spend always a bad sign?

Not on its own. Whether zero conversions means anything depends on how many clicks that spend produced relative to your category's typical conversion rate. At low click counts, zero conversions falls inside normal statistical variance, and only becomes a real signal once expected conversions at that volume clearly exceed one.

How do you know if a losing campaign is a bad offer or bad creative?

Compare click-through rate against your account average and landing-page conversion rate against the offer's category norm. Strong clicks paired with weak landing-page conversion points to the offer or its page. Weak clicks paired with normal landing-page conversion points to the creative itself, not the underlying offer.

What should you do with data from a campaign you just killed?

Save any audience or pixel data the platform allows, record the landing-page conversion rate for future comparison, and write one line stating exactly why the campaign died. That written reason is what stops you from re-testing the same failed combination six months later under a different creative.

Sources

Named rather than linked — verify before relying on any figure below.

• Meta Business Help Center — Ads Manager learning phase documentation
• STM Forum media-buying community testing threads
• ClickBank Marketplace gravity score documentation
• AdSpy published pricing page