New VK Ads account in 2026: a step-by-step guide for a marketer

The new VK Ads account (the one that used to be “Advertising myTarget”) in 2026 is not a “revived old account.” This is a different tool with a different goal-setting logic, different work with look-alike, different auto-strategies. Half of the clients I come to still work in it as in myTarget in 2019 - therefore they accumulate a CTR of 0.3% and a CPL of 800 ₽ where it should be 200 ₽.

This guide is not a retelling of the official VK documentation. In this office I launched campaigns in gaming, e-com, MedTech and B2B SaaS, at different times I simultaneously managed from 5 to 12 advertising accounts. Here is the sequence of steps that I myself use when logging into a new account, and a list of “seven mistakes” that I see in 80% of client accounts.

VK Ads is not “targeted advertising on VK”. This is a performance office with auto strategies, a retargeting stack and look-alike, which in 2026 are an order of magnitude better than in 2021. If you are still spinning money for show, you have missed the last five years.

1. Account structure: what is important to set up BEFORE the first campaign

This is the step that almost everyone skips. Before launching your first campaign in VK Ads, you need to configure three things in your account - otherwise you will lose ~30% of attribution accuracy and immediately fall into the “I don’t understand what works” mode.

1. VK Pixel + offline conversions. Pixel is placed in one line in the head of the landing page, but that’s half the battle. The other half is sending events through the API, especially on mobile where Pixel is often blocked. Coverage after this setup is 92–95% of events instead of 70%.
2. Goals and events. VK Ads has standard events (view, cart, purchase) and custom ones - the latter are written using the JS method. Without custom events, it is impossible to optimize for “deep” actions like “finished reading the FAQ” or “spent more than 2 minutes.”
3. Audiences - in advance. On the first day I create 5-7 audiences: site visitors 30/90/180 days, cart without purchase, customers last 60 days, look-alike from customers. They are needed for retargeting and for look-alike chains, and without them the first week of campaigns will pass without the main channels.

Time for this step: 2-3 hours on the first day. This will pay for itself 5–7 times in the first two weeks.

2. Campaign goals: what to optimize for in 2026

The main change here compared to myTarget is that VK Ads has switched to neural network auto-strategies. They work, but require the right choice of target.

Main rule: auto-strategies study 7–14 days. During this period, do not touch the settings. You see that the CPL is high for the first 5 days - this is normal. If you increase your bet after 2 days, the algorithm restarts training and you lose a week.

3. Look-alike: how to make it work

Look-alike in VK Ads in 2026 is not “similar to your subscribers.” This is a multidimensional search based on signals, which only works with a correctly selected seed audience.

The rules that I wrote down for myself after three unsuccessful look-alike campaigns:

• Minimum 1,000 people in seed. Less - the algorithm does not find enough signals. At 500 I caught a CPL 1.5–2x higher than planned.
• Seed only from quality events. Not “all site visitors,” but “purchased in the last 60 days” or “finished reading the article.” Seed quality = look-alike quality.
• The degree of similarity is 1–3%, not 5–10%. Wider - you lose accuracy. On gaming projects for NEMIFIST the best results are 1%, on e-com projects - 3%.
• Seed update every 2–4 weeks. Buyer behavior is seasonal. Look-alike for a two-month old seed is already distorted.
• Look-alike + retargeting together. Two separate campaigns in one group means doubling the reach on one audience and overspending the budget. Better - one campaign structure with two groups.

4. Creatives: what will be included in VK Ads in 2026

The main shift of the last two years is that VK Ads began to work better with short videos (5-15 seconds) and worse with static banners. This is due to the reorientation of the VK Clips feed and news feed to the video format.

What I see in the data for 30+ campaigns 2025–2026:

For the last year I have been generating most of my video creatives in Veo 3.1, banners and posters through Nano Banana Pro. A full analysis of the cost is in the article One marketer = team of 5 roles. It used to take a week to produce 50 creatives; now is day.

5. Retargeting chains: what works in 2026

Retargeting in VK Ads is the main “cheap channel”. On my projects, the CPL of retargeting audiences is usually 2.5–4x lower than that of cold traffic. But retargeting is easy to burn out if you don’t build the right chain.

A 5-step workflow that I reuse across projects:

1. Day 0–1: Visited site without action. Creativity is a reminder of the value of the product. Budget - 30% of the entire retargeting budget.
2. Days 2–7: Visited a product page or category. Creative - specific product + social proof. Budget - 25%.
3. Days 3–14: Added to cart without purchase. Creative - discount-trigger or urgency. Budget - 25%.
4. Day 14-30: Bought before but didn't return. Creative - new catalog or upsell. Budget - 15%.
5. Day 30+: haven't been for a long time. Creativity - what's new, brand mission. Budget - 5%.

The chain is built through VK Ads Custom Audiences with time-windows. Each segment is a separate ad group; users automatically flow from one to another over time. The most common mistake I see is that people make one group “visitors for 30 days” and run one ad on it. This is a loss of 60–70% of retargeting potential.

6. Seven mistakes that I see in 80% of offices

1. No Pixel or only partially installed. Without full Pixel+ offline conversions you lose 25%+ accuracy.
2. Auto-strategies are pulled every 2 days. The algorithm does not have time to learn. Do not touch for the first 14 days.
3. Look-alike made by 300 people. Too little. Below 1,000 it doesn’t work.
4. One creative for the entire campaign. Creative burnout begins within 5–7 days. At least 3 creatives in rotation.
5. Retargeting general “30 days”. Without behavioral segmentation, you lose 60% of efficiency.
6. There is no UTM markup or it is there, but it’s crooked. “utm_source=vk-ads&utm_campaign=mainCampaign” is not markup. We need a structure with medium, source, campaign, content and term.
7. Budget “for coverage” and not “for conversion”. Goal = coverage - for brand campaigns. For performance - only conversions or messages.

7. Niche strategies: where to start

Short recommendations for the most common niches in which I run VK Ads. These are starting configurations - always adjusted after 2 weeks of data.

• Gaming/eSports. The goal is messages to the community or subscribers. Look-alike 1% of active subscribers. Video 5–10 sec. Starting budget - from 1,500 ₽/day per group. More details in my article about EZ KATKA.
• E-com. The goal is conversions on the site. Catalog + dynamic retargeting. Look-alike 3% from buyers. Minimum 50 orders/week for optimization.
• MedTech. The goal is applications. Look-alike 1% + retargeting FAQ visitors. Video 30–60 sec, long text. High requirements for trust - creatives with real doctors.
• B2B SaaS. The goal is messages. Narrow audience based on interests + retargeting from pricing/cases pages. Creatives - client cases, video demos.
• EdTech. The goal is leads through the lead form. Look-alike from those who filled out the form. Launch during peak seasonality - 2-3 weeks before the start of recruitment.

If you currently keep VK Ads at CPL 1.5–3x higher than planned, most likely the problem is one of the seven errors above. Most are resolved in a week or two of reconfiguration. Would you like me to sort out your office for 30 minutes? discovery-colle — I’ll show you what to fix first, without selling the retainer.

Related materials: CTR benchmarks in VK Ads, CPL calculator, ROAS calculator.