Checklist for Evaluating Which Campaigns Are Worth Keeping

campaign evaluation ad spend performance review

Heads up — this post may include links to things I use or like, and I might earn a little something if you shop through them. Doesn’t cost you anything extra, and I only mention stuff I’d actually recommend.

Most campaign post-mortems happen too early. A cost-per-conversion spike on day three triggers a pause. A slow first week feels like failure. But the frameworks used in professional digital advertising audits suggest a different starting point: you need 50–100 conversions per month before the data is statistically reliable enough to make a keep-or-cut call. That threshold alone changes which campaigns deserve patience and which ones genuinely need to go.

How much data is enough before you decide

The single most common mistake in campaign evaluation is judging performance before the sample size supports a conclusion. Multiple audit frameworks converge on the same range: 50–100 conversions per month is the minimum for statistical significance. Below that, what looks like a trend is usually noise.

The same logic applies to bidding strategy transitions. Automated bidding — Target CPA or Target ROAS — needs at least 30–50 conversions within a 30-day window before it can optimize reliably. Switching earlier means the algorithm is guessing, not learning. One source I reviewed recommends starting with Manual CPC or Maximize Clicks to build conversion history, then moving to Smart Bidding after that threshold is met.

This has a practical consequence for evaluation: if a campaign hasn’t accumulated enough conversions, the question isn’t “is this working?” — it’s “have we given it enough runway?” The answer, in most cases, is no.

What your tracking setup is actually saying

Before you evaluate whether a campaign is worth keeping, you need to know whether the data you’re looking at is accurate. Misconfigured tracking is the invisible budget leak that gets blamed on creative, audience, or offer.

The digital advertising checklist from Beacon-Ads emphasizes installing Meta Pixel and Conversions API in parallel with identical event_id values for deduplication. Without that, the platform can double-count conversions, inflating what looks like performance. On the Google side, enabling Enhanced Conversions delivers an average 8% ROAS improvement on Search campaigns — but only if the setup is correct.

Landing page speed is another data-quality factor that masquerades as a campaign problem. Core Web Vitals benchmarks — Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200ms, Cumulative Layout Shift under 0.1 — directly affect Quality Score and cost per click. A campaign with strong creative and weak landing page speed will look like it’s failing when the real issue is technical.

The V9 Digital marketing audit framework recommends auditing conversion path friction specifically: look at bounce rates, step abandonment, and device-level differences. If mobile users drop off at twice the rate of desktop, the campaign might be fine — the landing page experience isn’t.

The first two weeks tell a different story

The post-launch period has its own evaluation logic. Multiple sources agree that the first 14 days are a monitoring window, not a performance window. During this phase, the priority is signal quality, not results.

The BigFinSEO paid advertising checklist recommends reviewing the Search Terms report and adding negative keywords daily during this period. It also flags that Quality Scores below 5 in the first week signal a keyword–ad group mismatch or a landing page problem — not a campaign that needs to be killed.

Impression share tells you whether the campaign is hitting budget or rank limitations. CTR by device reveals creative or landing page experience gaps. One framework explicitly warns against making bid changes during the first 7–14 days of Smart Bidding unless something is clearly broken — because the algorithm is still in its learning phase and needs stable input to calibrate.

The Bullseye Strategy PPC audit checklist notes that AI-driven campaigns like Performance Max have a 2–4 week learning phase where performance may fluctuate, and major changes during this window reset the learning process entirely. The same source recommends uploading at least 5 text variations, high-quality images, and at least one video for PMax asset groups — otherwise Google auto-generates video that usually underperforms.

The decision: cut, hold, or scale

Once you have reliable data — enough conversions, clean tracking, and a full learning period — the evaluation framework becomes clearer. The research points to three actions, each with specific triggers.

Cut. The BigFinSEO checklist recommends cutting ads with no conversions after reaching statistical significance, typically 100+ clicks. The Janzen Marketing year-end review framework suggests looking at ROI and cost-per-click across channels to identify the lowest performers. But cutting should be based on conversion data, not impression volume — a campaign with high impressions and low conversions is wasting budget, while a campaign with low impressions and a strong conversion rate may just need more reach.

Hold. Campaigns that meet conversion thresholds but haven’t stabilized yet need patience. The Bullseye Strategy audit notes that AI optimization requires 2–4 weeks of learning, and performance may fluctuate during that period. Holding also applies to campaigns where the tracking setup was fixed mid-flight — the data before the fix isn’t comparable to data after.

Scale. When a campaign proves its cost per acquisition at a small budget, the recommended approach is to increase spend in 20% increments while watching CPA closely. Shift budget toward campaigns with the lowest cost per conversion, not the highest impressions. The HostingCT small business checklist also recommends retargeting non-converters as a scaling strategy — sometimes the campaign that looks like it’s failing is actually building an audience that converts on the second touch.

What happens before the ad ever runs

The most honest finding across all the research I reviewed is this: most campaign problems are setup problems. A 30-minute pre-launch checklist can prevent multi-day troubleshooting of conversion data mismatches. The infrastructure behind the ads — CRM integration, landing page readiness, audience segmentation, budget logic — determines whether the campaign has a fair chance to perform.

The V9 Digital audit framework recommends gathering 12 months of performance data before making strategic decisions, involving a small group (one marketing decision owner, one budget owner, optionally a sales or media manager), and focusing on patterns rather than isolated metrics. That last point is worth sitting with: a single bad week isn’t a pattern. A consistent direction over months is.

The landing page best practices guide on this site covers the message-match principle that every ad audit checklist emphasizes: the ad headline and the page headline need to say the same thing. When they don’t, the campaign looks broken even when the targeting and creative are fine.

For anyone running their own campaigns — especially solo operators and small teams — the question isn’t just “which campaigns are worth keeping.” It’s also “which campaigns were set up in a way that let them actually work?” The answer to the second question often answers the first.