The Paid Ads Diagnostic Playbook: What to Check Before You Blame the Algorithm

Performance drops and the first instinct is to blame the platform — "the algorithm changed," "iOS broke attribution," "the auction got more expensive." Sometimes that's true. Usually it's one of five specific, checkable things. Here's the order to check them in, and which checks you can hand to a dashboard versus which ones need a human looking at the actual ads.

1. Confirm it's actually a performance problem, not a measurement problem

Before touching the account, check whether the drop is real or just invisible. Pixel firing correctly? Conversion API still connected after that site migration last month? UTM parameters intact on the latest landing page? A surprising share of "ad performance crashes" are tracking outages wearing a performance costume. This check is mechanical — set up an automated tag-health monitor that pings you the moment a pixel stops firing, and you'll rule this out in minutes instead of a week of panicked bidding changes.

2. Separate creative fatigue from audience saturation

If the same ads have been running for six-plus weeks to the same audience, frequency is your first read. Climbing frequency with falling click-through is creative fatigue — the fix is new creative, not a new strategy. Falling reach and rising cost-per-result at a fixed budget is closer to audience saturation — you've shown the ad to everyone worth showing it to, and the platform is reaching deeper into a colder pool to spend your budget. These look similar on a single chart but need opposite fixes, so pull frequency and reach trends, not just the headline conversion number. A model can flag "frequency above 4 and CTR down 20%" automatically; deciding whether the answer is a new hook or a new audience is the part that still needs a person who's actually looked at the creative.

3. Check the landing page before you touch the ad account

Ad metrics and site metrics get diagnosed separately way too often. If click-through rate is steady but conversion rate has fallen, the problem probably isn't upstream in targeting at all — it's the page. A pricing change, a slower load time after a CMS update, a broken form on mobile: any of these will quietly tank an account that the ad platform itself is reporting as healthy. Five minutes on the actual landing page, on the actual device your audience uses, catches more "ad problems" than another week of bid adjustments.

4. Look at the calendar before you look at the budget

Auction costs move with demand, and demand moves with the calendar — pay periods, competitor promotions, seasonal categories, even local events. Before assuming your targeting broke, overlay this month's cost-per-result against the same week last year and against your category's typical seasonality. If costs are up across the board for everyone bidding in your space, the fix isn't a new audience — it's a budget or bid-strategy conversation, and possibly patience. Automated alerts can flag "CPM up 30% week over week"; only someone who knows the business knows whether that's worth fighting through or waiting out.

5. Audit attribution last, not first

Attribution windows and cross-device tracking are real and have gotten messier, but they're also the easiest excuse to reach for because they're the hardest to fully verify. Save this check for last. If you've ruled out tracking, fatigue, saturation, the landing page, and seasonality, and the numbers still don't add up, that's when it's worth reconciling platform-reported conversions against your CRM or order data directly. Most of the time you won't need to — you'll have found the actual cause two or three checks earlier.

Build the checklist once, run it automatically

None of this requires guesswork each time performance dips — it requires a checklist run in order, every time, before anyone touches a bid. The monitoring (pixel health, frequency, CPM trends, page speed) is exactly the kind of repetitive, always-on watching that automation handles better than a person checking a dashboard once a week. The judgement — is this fatigue or saturation, is this seasonality worth waiting out, is this landing page change worth reverting — still needs someone who understands the account and the business behind it. Automate the watching. Keep the deciding.