Introduction: what is performance and who needs it

Performance marketing is marketing in which every ruble of advertising can be tied to result. Not “reach”, not “recognition”, not “involvement”, but a live transaction: lead, application, payment, registration. If an advertising channel does not provide measurable action, for performance it doesn't exist. Dot. Everything else is brand-marketing, and its task completely different (to warm up future demand, and not close today’s).

Eight years in performance have taught me one thing: half of a customer's problems start not from a “bad Direct”, but from an attempt to solve a problem with performance tools that not about performance at all. Therefore, the first chapter is not about setting up campaigns, but about who needs performance, to whom it is strictly contraindicated, and what five metrics distinguish “run ads” from “manage performance marketing.”

performance vs brand: where is the line?

Brand marketing works with memory. I saw an advertisement for Alfa-Bank in the metro, after four I wanted a mortgage for a month - I went to Alpha first. There is a delay between touch and action days, weeks and months. It is impossible to measure ROI using direct attribution; it is measured through brand-lift research, funnel capacity, share of mentions.

Performance works with demand that is already there. The client googles “buy dance rug”, sees my ad, goes to the landing page, leaves a request. From the first touch to measurable action - minutes or hours. Each penny is attached to a specific action through UTM, pixel and end-to-end analytics. This is the watershed: brand builds demand, performance collects it.

Practical criterion: if you can state the expected cost per lead (CPL) or customer cost (CAC) before launching the campaign and check the plan after two weeks - you in performance. If the answer sounds like “well, let’s wait a couple of blocks and see how it goes” - you are in brand. It's not bad. It's a different game with different rules.

who needs performance right now

The performance approach gives the maximum where there are three conditions at the same time. First: formed demand in a niche (people are already looking for “buy X”, “order Y”, “service Z”). Without demand performance simply measures how slowly you are trying to run to empty store. Second: measurable action on the site or in the bot (application, order, registration, subscription). Third: sufficient margin for CAC to fit into LTV with a margin - usually You need an LTV/CAC ratio of 3:1, otherwise you pay more for a client than he brings.

Typical niches where performance works almost automatically: e-commerce with a receipt from 2,000 ₽, education and information products, self-service SaaS, local services businesses (dentistry, car services, repairs), MedTech and FinTech products with fast onboarding. Wherever less than a month passes between “saw an advertisement” and “paid”.

for whom performance is strictly contraindicated

The list is shorter and more important. If you hit even one point, your advertising money will be lost faster than you can blame the “adjuster”.

• Pet project with organic content of 50% and above. If SEO and word of mouth give half the traffic, any the paid channel will be “dirty”: it is impossible to separate “he would have bought it anyway” from “he bought precisely because of advertising.” First, complete the attribution (see chapter 11), then turn on traffic.
• B2B with a long cycle of more than 6 months. If the average trade closes in 240 days, performance metrics will lose touch with reality before you do get the first data. Doesn’t mean “not to use at all” - it means “not as main channel." Content marketing, ABM, long-term nurture helps.
• Product without PMF (product-market fit). If people don't buy what you sell, advertising will not fix the product. It will simply show that they are not buying in large numbers. Read more about this scenario in chapter 17.
• A business where the deal is closed by a live manager, but there are few managers and they slow down. Leads will go bad in CRM, sales conversion will collapse, performance metrics will show “channel doesn't work." What actually works is the front office that doesn't work.
• The audience is too thin. If your target is “clinic directors in cities 500,000+”, even the ideal target setting will not work: the algorithms need volume data. Up to 50–100 conversions per month per channel, autobid strategies will not start working.

the main metrics without which you can’t start

Five metrics are the minimum vocabulary of a performance marketer. We analyze deeply each in Chapter 2, here - at the level of “what to count in the first week.”

• CPL (Cost Per Lead) - how much one lead costs. Formula: expenses ÷ leads. We count it right away, because this is the figure that the CEO will see for the first time. Monday. CPL calculator on the website.
• CPA (Cost Per Action) - the cost of the target action. It could be application, registration, adding to cart, subscription. CPA is broader than CPL and replaces it in products where “lead” is too rich a concept. CPA calculator.
• CAC (Customer Acquisition Cost) - the cost of the buyer. Not Lida, not registration, but the paying person. Formula: marketing expenses ÷ customers. CAC is the main number that financiers and investors look at.
• ROAS (Return on Ad Spend) - return on advertising. Formula: revenue ÷ advertising expenses. ROAS 400% means “for every ruble of advertising, four were returned ruble of revenue." Not to be confused with profit. ROAS Calculator.
• LTV (Lifetime Value) - the lifetime value of the client. How much money the client will bring during his entire “life” in the company. Without LTV, it is impossible to calculate the “ceiling” CAC." LTV/CAC calculator.

For detailed definition and context, see glossaries, in the same place - related terms (attribution, conversion rate, frequency cap and thirty others).

where to start in the first two weeks

Before you launch your first ruble, check four things. If at least one answer no - performance is still early, you need to complete the base first.

1. Do you know your target CPL/CPA? Not “I want it cheaper”, but a specific figure with justification through LTV and margin. See chapter 12.
2. Are pixels installed (Ya.Metrika, VK pixel) and goals configured? Without this you won't you will see the data on day one.
3. Is there a landing page with at least a 2-3% conversion rate? Not a business card site, but a landing page for a specific offer. Chapter 10.
4. Is the sales team ready to accept leads? If the CRM is empty, and the manager is alone and on vacation - first close the hole, then turn on the traffic.

in the next chapter

Chapter 2 - an in-depth analysis of five metrics: formulas, typical values for niches in the Russian Federation 2026, how they are related to each other, which ones are most often confused with each other. Without this dictionary it will be noisy further.