Common mistakes in performance 2026
Top 10 mistakes: wide audience, lack of UTM, no attribution, no LAL, no retargeting and others.
Over four years, I conducted about 60 performance audits of other people's offices. 80% errors are repeated from project to project - regardless of niche, budget and qualifications previous contractor. This chapter is the top 10 of these mistakes in descending order. soreness. If you recognize three or more in yourself, it’s time to change your approach (or contractor).
mistake 1: one campaign for “everyone”
One campaign “brand + category + competitors + retargeting” in one bottle. Budget turns to where CPC is cheaper - usually this is a brand (their employees Google, and Direct “converts” them). Real new clients do not receive impressions.
Treatment: at least 4 different campaigns - Brand, Category, Competitors, Retarget. The budget for each is separate. Creatives and texts are different.
mistake 2: wide audience without segmentation
Target “18–65 years old, Russia, interests - everything.” The algorithm finds the cheapest users who are easier to “buy” at the cost of a click, but they are irrelevant to the business. CR drops three to four times compared to a correctly segmented target.
Treatment: divide the audience into at least 3-5 segments with different creatives and budgets. See Chapter 8 about look-alike and segments.
error 3: there is no UTM markup or it’s just for show
utm_source=yandex / utm_medium=cpc / utm_campaign=test1 / utm_content=test2. Via For six months no one can remember what was in “test1” and “test2”. Analytics impossible, optimization is impossible.
Treatment: clear markup standard, convention document, Google UTM generator Sheets for each channel. Basic scheme: utm_source=yandex / utm_medium=cpc / utm_campaign=brand-search-msk / utm_content=ad1-banner1.
error 4: end-to-end analytics is not configured
Leads are counted in Metrica, sales are counted in CRM, the report is collected once a month in Excel manually. By the time the problem is visible, the budget has already burned out.
Treatment: dashboard in Looker Studio / DataLens / Metabase, updated automatically. Metrica+CRM combination via UTM. See chapter 11.
Mistake 5: Ignoring look-alikes
“We only use Search Direct, YAN makes no sense, we don’t use look-alike.” Result: the hot demand limit has been exhausted in a month, further growth due to rate increases. CPA grows linearly, budget scales for cold traffic impossible.
Treatment: Every month 20-30% of the budget for LAL and cold segments. This is an investment in scaling, not "wasting".
error 6: no retargeting or it’s for “everyone”
Either retargeting is completely turned off (“search only, retargeting does not work”), or included in the “all visitors 365 days” segment with one creative. Abandoned cart receives the same ad as a person who accidentally came from a search.
Treatment: at least 5 retargeting segments with different creatives and frequency caps. See chapter 7.
Mistake 7: Same creative for 3 months in a row
“This creative worked in March, let’s leave it in April, May, June.” By June, CTR drops by 50%, CPC grows by 60%, conversions fall, but “the creativity is the same - that means it’s a matter of something else." This is creative burnout.
Treatment: rotation of creatives once every 2-3 weeks in the cold, once a week on retarget. See chapter 9.
Mistake 8: Launching an Auto Strategy Without Conversions
We launched the “Maximum conversions with pay-per-conversions” campaign, CPA goal 1,500 ₽. Conversions per week - 3. The algorithm cannot learn, the campaign is “sleeping”, the budget not consumed. Two weeks later, the marketer complains “autostrategies are not working.”
Treatment: we turn on auto strategies if there are at least 30 conversions per week on campaign. Before this - “Maximum clicks” with a daily limit, or manual bids.
Mistake 9: Home Page Landing
The advertisement leads to the main page of the site, where there are ten product categories and five services. The user comes to “buy a dance mat”, sees “we make stands for exhibitions” and leaves. CR on the main one is usually 3–5 times lower than on the special one landing page.
Treatment: separate landing pages for each campaign. See chapter 10.
error 10: lack of conversions from CRM to direct
Direct sees “lead left phone number” as a conversion. CRM sees “lead bought” - this a completely different event. Without transferring the fact of payment back to the Direct algorithm optimizes for “cheap leads” who then don’t buy. CPL is beautiful, CAC - a disaster.
Treatment: Conversion API from CRM to Metrica and Direct. The algorithm begins to optimize for those who pay, the CAC levels out.
bonus: second order errors
These are not the main ones, but I see them regularly:
- Budget “1000 ₽ per day.” Algorithms do not allow less than 3-5k ₽ per day have time to study. A week at 30k is better than a month at 5k.
- Launch without negative keywords. Minimum 200 negative keywords on day one.
- One office for 5 projects. The history of one project influences quality score of others. Each project is its own office.
- Ignoring mobile. 65–75% of traffic is mobile, and The landing page was tested only on the desktop.
- They don't count LTV. All CPA decisions are made without understanding real value of the client.
- There is no backup channel. If Direct breaks down for a week - business will decline. At least two channels in operation.
- Creativity is generated “by inspiration.” Without hypotheses and systematic A/B tests are an accident, not work.
how to quickly conduct a self-audit
Sit in front of the monitor with this list and put a “+” or “−” next to each errors. If you get 5+ minuses, it’s time to clean performance. If 3+ - definitely worth it do an audit. Less than 3 - you are doing better than 80% of the market.
For deep auditing you can use calculators on the site - they will quickly show “whether the unit economics are healthy” and “whether the target CPA is correct.” Or you can make an appointment with me for analysis via contact page — 30 minutes, we analyze the funnel, I come up with hypotheses.
in the next chapter
Chapter 16 - AI tools in performance 2026. ChatGPT for copywriting, Midjourney and Nano Banana for creatives, n8n for automation, Claude Code for landing pages. A specific stack that I use myself.