How many goals do you need to set up in Ya.Metrika?

Minimum 5: main conversion (order/application), micro-conversion (feedback form), engagement (time on the site, scrolling), auxiliary (click on the phone, messenger), technical (refusals). For e-com, add e-commerce events via dataLayer.

How is a compound goal different from a regular goal?

The usual goal is one action (button press). Composite - a sequence of 2-5 steps (opened the form → filled out → sent). Compounds provide a funnel with the conversion of each step and help find bottlenecks in UX.

Do I need to use Ya.Metrika if I already have GA4?

In the Russian Federation - yes, definitely. Ya.Metrica communicates with Yandex Direct to transfer conversions and auto strategies, GA4 does not do this. Plus Ya.Metrica has a Webvisor (session recordings), a scroll map and heat maps, which is not available in GA4 without third-party services.

What is Webvisor and is it worth enabling?

Webvisor records user sessions emulating mouse movements and clicks. Useful for UX audits and finding problems in the conversion funnel. Cons: heavy code, can slow down loading by 100-200ms. Enable only when UX analysis is active, then disable.

How to set up conversion tracking from a Telegram bot?

Through offline conversions in Ya.Metrika API. When the bot detects a target action (payment, application), it sends an event with user_id or ClientID saved via javascript to the Metrics endpoint. Yandex Direct sees this conversion and optimizes the campaign.

What's the difference between ClientID and UserID?

ClientID is a unique browser identifier that Metrica automatically assigns. UserID is the user identifier in your system (login, email, phone), which you transfer to Metrica. UserID works cross-device (if one user logs in from two devices).

Is it possible to transfer data from CRM to Metrica?

Yes, via offline conversions API. When a deal in CRM is closed, send an event to Metrica with ClientID and the fact of purchase. This provides the most accurate optimization in Yandex Direct: the algorithm sees not “left an application”, but “left an application that became a sale.”

How long does it take to set up Ya.Metrica on a new project?

The minimum setting with one goal is 1 hour. Full setup with e-commerce, compound goals, GTM, audiences and reports - 8-16 hours depending on the complexity of the site. From scratch without experience - budget 2-3 working days plus testing.

All articles

PerformanceMay 11, 202613 min

Yandex Metrica 2026: setting goals, e-commerce and linking with Direct

Ya.Metrika is the central data hub for performance in the Russian Federation. Without correctly configured goals, Ya.Direct overpays for CPL by 30-50%. Full guide: what 8-12 goals are needed, e-commerce through dataLayer, connection with YA Audiences and CRM, automatic alerts in Telegram. With an analysis of 5 typical mistakes.

Ya.Metrika is not just a “statistics counter”. In 2026, this is a tool that feeds data to Kolibri in Yandex Direct and determines how much you will pay per click. Poorly set goals = 30-50% overpayment for CPL. Well configured = auto strategies work, the neural network quickly finds your audience. This article is about a setup that will withstand the performance load, and not a “tick” counter for the sake of formal presence.

I set up Metrica on about 40 projects: e-com, MedTech, EdTech, B2B software, e-sports. In my work projects, Metrica is the central hub: data from Yandex Direct, VK Ads, TG Ads, CRM and custom Telegram bots flows there. All figures in the article are from practice.

The main mistake of 80% of sites in the Russian Federation is that the site contains the Metrica code, but either there are no configured goals, or they are formal (one-page-visit, time on site 10 seconds). This is the equivalent of putting a camera in a store and never looking at it.

1. Why set up goals in Ya.Metrika at all?

Metric goals do three useful things. First, they allow Yandex Direct to optimize for conversions, not clicks. This is the difference between “the neural network brought traffic” and “the neural network brought customers.” Secondly, they collect audiences for retargeting and look-alike. Third, they allow you to see the conversion funnel and find bottlenecks.

Without goals, Metrica is a beautiful dashboard without business sense. You can look at traffic and draw conclusions about sources, but you cannot optimize advertising for results. And most importantly, you cannot compare channels based on the real cost of the result.

2. What goals are really needed (and how many)

I usually set 8-12 goals for an average project. Less - do not cover the funnel. There is more confusion and noise in the reports.

Primary conversion (1 goal) - the main target action. For e-com - placing an order, for leadgen - submitting a form, for SaaS - registering for a trial. This is the goal against which all auto strategies will be optimized.

Micro conversions (2-3 goals) - actions that lead to the main one, but are not it: clicking “buy” (without payment), adding to cart, opening a form. Helps us understand where people fall off.

Engagement goals (2-3 goals) — scrolling 75% of the page, time on the site more than 2 minutes, viewing more than 3 pages. Used for look-alike and retargeting to a “warmed-up” audience.

Auxiliary (2-3 goals) - click on the phone, click on the messenger (WhatsApp/Telegram), go to the YouTube channel. Often they close some conversions that are not visible in the main goal.

Technical (1-2 goals) — click on 404 page, form submission error, cookie refusal. Helps find technical problems.

Recommended structure of goals by niche - what I set at the start of the project:

Niche	Main	Micro	Engagement	Aux.
E-com	purchase	add_to_cart, begin_checkout	view 3+ items, time 2min+	click to chat
B2B software	form_submit	open_demo_video, scroll 75%	view 3+ pages	click "contact"
MedTech clinic	booking_submit	click_phone, scroll 75%	view services 2+	click whatsapp
EdTech	signup_trial	watch_intro_video, scroll 80%	view 2+ courses	click to TG channel
SaaS	trial_start	signup_step_2, view_pricing	session 3min+	dox open
Local services	form_submit	click_phone, scroll 75%	view contacts	click Ya.Maps

3. E-commerce setup via dataLayer

For e-com projects without e-commerce settings, Metrica is useless. Standard JavaScript code only counts page transitions, but what is needed is what was put in the cart, for what amount, what product.

The standard is Google Enhanced Ecommerce dataLayer (Ya.Metrika understands it). Minimum set of events:

view_item — viewing the product card. Passes product_id, price, category.
add_to_cart — adding to cart.
begin_checkout — transition to registration.
purchase - successful payment. Passes transaction_id, revenue, list of products.

Implemented through a developer on the backend (if SSR) or through GTM on the frontend. On my e-com projects, development takes 2-3 days, then it works for a year without intervention.

4. Connection with Yandex Direct, Ya.Audiences and CRM

The goals themselves are half the work. The other half is to connect them with advertising and CRM sources.

Yandex Direct: in the settings of each campaign, specify the counter ID and list of goals. After this, Kolibri begins to optimize for conversions, not clicks. Auto strategies (“pay for conversions”, “target share of advertising costs”) work only with connected goals. More details in guide to Yandex Direct 2026.

I.Audience: segments are created based on goals - “bought”, “put in cart but did not buy”, “spent more than 2 minutes”, “haven’t been on the site for 30 days”. These segments are used for retargeting and look-alike.

CRM: via the offline conversions API, the events “lead became a client” and “client made a repeat purchase” are sent to Metrica. This gives the most accurate optimization: Yandex Direct sees not “left a request”, but “left a request that brought 50,000 ₽”.

5. Automatic alerts and reports

Ya.Metrica has email alerts out of the box: a drop in conversions by N%, an increase in 404 errors, abnormal traffic. This is the basic minimum - set up in 10 minutes.

Next, through the Metrica API + n8n or your own script, you can collect a daily report and send it to Telegram. On my projects, such a report arrives at 9 am and contains: traffic for yesterday, conversions by main goal, CPL by source, top 3 pages. Time for analysis - 2 minutes a day instead of 20 minutes in the office.

I posted the n8n-flow template for a report in Telegram in article about 10 n8n recipes.

6. Top 5 mistakes in setup

Over 40+ projects, I counted approximately the same failures:

1. Target only the “thank you for ordering” page. If the user does not have this page (modal window or AJAX form), the conversion is not recorded. Solution: events via dataLayer or JS call yaCounterXXX.reachGoal('purchase').

2. Duplicate goals. Same conversion, same goal. If there are 3 “submit a request” buttons on the site, that’s one goal, not three. Otherwise, Yandex Direct counts 3 conversions instead of 1.

3. Lack of CRM conversion tracking. Application ≠ sale. If you do not transmit the fact of sale to Metrica, the neural network is optimized for the number of leads, and not for quality. After 2-3 months - there are leads, but no sales.

4. Ignoring micro-conversions. Without them, Yandex Direct takes a long time to learn. If the volume of main conversions is small (less than 30 per month), auto strategies do not work—micro-conversions are needed for training.

5. Enabled Webvisor on a combat project. Webvisor slows down loading by 100-200ms and collects a huge amount of data. Turn it on only during UX analysis, then turn it off.

7. Conclusion

Ya.Metrika in 2026 is a central data hub for performance projects in the Russian Federation. Without correctly configured goals, Yandex Direct overpays for CPL by 30-50%, look-alike does not work, reports are useless. Setup takes 1-3 business days and pays for itself in the first month through a reduction in CPL.

If you need an audit of your current Metrics for the quality of settings - Telegram or form. Basic audit - 0 ₽.