What do I do when a performance campaign doesn’t work: 6-step checklist

Once every two weeks something breaks in one of my campaigns. CPL jumps 3 times, conversions drop, budget is burning. In 9 years I have developed 6-step diagnostic algorithm, which boils down “doesn’t work” to a specific reason in 90 minutes. Not “let’s update the creatives and see,” but an analytical process - which in 8 out of 10 cases gives an accurate answer.

It annoys me when my colleagues write in chats “the campaign has fallen, what should I do?” and then “perhaps the creative has burned out, try updating it.” This is not a diagnosis - this is fortune telling. Real diagnostics: you go through a fixed list of sources of the problem and in 90 minutes you localize exactly where it’s broken. I'm sharing my list.

“The campaign is not going well” is not just one problem. This is a symptom that has 6 typical causes. Find the one that happened - the fix takes 2 hours.

When to use this algorithm

Symptoms for which the checklist works:

• CPL increased by 30%+ in 5–7 days with the same budget
• Conversion to target action dropped by 20%+
• The auction is heating up (CPC is growing), but traffic is not growing
• Quality indicators (bounces, time on site) have deteriorated for no apparent reason
• “Everything was fine, then it broke” (and there is no obvious trigger)

If the campaign “never started” is a different scenario, a full audit is needed. This algorithm is for the case when there was a moment when it worked, and there is a moment when it broke.

Step 01. Check tracking to see if the data is broken

The main rule of diagnosis is to first make sure that you are looking at the truth. In 30% of cases “the campaign fell” it turns out that “the reporting was broken.”

What we check in 10 minutes:

• Pixel/VK tag / Ya.Metrica / GA4 - does it fire standard events (PageView, Purchase, Lead)?
• Are the UTM tags in active campaigns the same as they were, or did someone change them?
• Webhook between the advertising platform and CRM - are new leads coming in?
• Didn’t a new funnel appear in CRM to which some of the leads were redirected?

One of the latest failures: on the E-com project, I noticed “-30% of sales for the week.” 20 minutes of checking - it turned out that the developers updated the checkout and Pixel.Purchase was shooting on the intermediate page, and not the final one. Conversion in reality did not fall - it fell measurement. If I started “updating creatives”, I would lose a day and 100K rubles of budget on a false hypothesis.

Step 02. Compare cohorts - is it a failure for everyone or a specific segment?

After checking the data - not “the campaign fell”, but where exactly did it fall. Divide metrics by:

• Channels (VK / Yandex Direct / TG / organic)
• Geo (if several regions)
• Devices (mobile vs desktop)
• Audience types (look-alike, retargeting, cold)
• Segments by gender/age

In 60% of cases the failure is localized in one segment, the rest are stable. This changes the entire diagnostic strategy.

Specific example: in EZ KATKA in 2024, CPL increased by 40% across the entire network. After the division by geo, it turned out that only Moscow Center and St. Petersburg Center sank. Regions are stable. Further, the investigation went in the direction of “what has changed in big cities” (answer: competitors raised bids in the VK auction after the opening of the new season; more about the Russian performance market in 2026 - separate analysis).

Step 03. Check bids and auction to see if the market has heated up

If the symptom is “CPC has increased,” the first suspicion is competition in the auction. Especially for VK Ads and Yandex Direct, where bids are auctioned dynamically.

What we are watching:

• CPC chart in your account for the last 30 days - a sharp jump or a smooth increase?
• Bid forecast (if the platform provides it) - has the recommended bid changed significantly?
• Seasonality - maybe now is the “hot season” in your niche?
• Market news - a competitor has launched a large-scale campaign that is stealing the auction?

If the market is really overheated, the answer options are:

1. Raise the bid and accept the new CPL (if unit-economics still allows)
2. Shift to a less competitive audience segment
3. Switch to another channel where the auction is less overheated
4. Wait (if it's a seasonal peak) and scale later

Step 04. Running creatives - burned out or not

Only at the 4th step do we reach the point of “creativity burnt out.” This is because burnout is a common cause, but not the main one. And diagnosing burnout should be based on numbers, and not “let’s update, just in case.”

What we check:

• CTR for the last 14 days vs previous 14 – has it dropped by more than 20%?
• Frequency (impressions per user) – is 4–5 higher for a cold campaign?
• Is the oldest active creative more than 3 weeks old in active rotation?
• Reach - audience saturation (if you have reached 80%+ ICP - is it time to expand or change)?

If 3 out of 4 answers are “yes,” then the creatives have really burned out. Solution: 4–6 new creatives per week, put them in an A/B test with the current ones, after 7 days - leave the top 2 and disable the old ones.

Step 05. Check the landing page and funnel to see if the “after click” is broken

Often, a failed “campaign” actually fell after the advertising platform - in the funnel. If the CTR has not changed, but the sales conversion rate has dropped, the problem is not in advertising, but in landing/CRM.

Funnel checklist:

• Landing page loading speed (PageSpeed Insights / Lighthouse) - has it dropped?
• Are the forms broken? Are applications being sent?
• Are 404 / 500 errors appearing on key pages?
• The response time of sales managers has not increased (rest, vacations)?
• Has the price on the website changed (it happens: marketing leads to a “−20% offer”, but on the website there is already a standard price)?

One of the most expensive failures in my practice was the client changing the price on the landing page without approval from marketing. CPL remained the same, but sales conversion fell by 50%. We spent a month and a half investigating before we found the source.

Step 06. External factors - what has changed outside the campaign

If all 5 steps above showed “everything is fine”, but the symptoms remain, it’s time to look around.

What could be:

• The platform has changed its algorithms (VK / Yandex Direct / TG release updates regularly)
• A competitor launched an aggressive promotion, attracting the audience
• The political/economic situation has changed (macro factor)
• Seasonality (which you did not take into account)
• Something has changed inside the company - a new offer, a new pricing policy, a change in positioning

This step is the most difficult, because external factors are often diagnosed retrospectively (“a month later they realized that it was due to a change in the VK algorithm in March”). But you still need to formulate a hypothesis to decide “what to do next.”

What to do after diagnosis

When I found the reason - actions by type:

What I never do

A list of “crutches” that I came across in other people’s campaigns. They don't help:

• “Restart the campaign” (turn it off and on) - this deprives the algorithm of the collected training sample. It often gets worse.
• “Sharply increase the budget by 2×” - without diagnostics, this will simply speed up the burning of money.
• “Update all creatives at once” - you lose your starting point, it’s not clear what worked.
• “Disable targeting by interests, leave only look-alike” - for no reason, simply because “look-alike is fashionable.”

What to pick up on Monday

If your campaign doesn’t work tomorrow, open this checklist and follow the 6 steps in this order:

1. Check tracking to see if the data is broken (10 minutes)
2. Divide the metrics into cohorts - where exactly is the failure (15 minutes)
3. Look at the bids and auction to see if the market has heated up (15 minutes)
4. I ran the creatives through - did they burn out or not (15 minutes)
5. Check the funnel to see if it breaks after clicking (15 minutes)
6. Look for external factors - what has changed outside the campaign (20 minutes)

Cumulatively - about 90 minutes. In 8 cases out of 10 - the exact answer. In 2 cases, deepening in a specific direction will be required.

If you want to sort out your current case with a failed campaign, write @dipustovalov. Often such an analysis is done in 30 minutes on a call, then a fix plan.

Related materials: ROAS/ROMI calculator, CPA calculator, a complete guide to performance marketing 2026.