New format
Web analytics tools: tags, logs, and dashboards
Web analytics is collecting and reading data on visitors and how the site behaves — so you can decide where budget goes, what to fix on pages, and which channels pay off.
Below: what systems report, how tags differ from log analyzers, and how to start with Yandex Metrica and Google Analytics. The analyst role and dashboards live in related pieces; here the focus is tools and data.
What web analytics is
It is a loop: data → meaning → action. You watch audience, sources, paths, conversions, and tech signals (speed, errors) to change ads, content, and UX.
Without regular review, analytics becomes “we installed a tag and forgot.” Agree business goals before you wire the reports.
What it is for:
- channel and CPA evaluation
- finding weak landings
- checking hypotheses and releases
- reporting for the team and client
What systems report
Base layer: visits and users, sources and queries (where available), devices, geo, pageviews, time, events and goals. Then funnels, ecommerce, cohorts.
Tech slice: load speed as users feel it, failures, sometimes overlap with server data. Marketing slice: which ads and landings bring leads.
Minimum weekly review:
- traffic by channel
- conversions and CR
- top landings and bounces
- anomalies after releases or promos
Tags and log analyzers
A tag is a script in pages: handy reports, segments, goals, session replay. Downside: it depends on JS, blockers, and cookie consent.
A log analyzer reads server journals: full request flow, bots, response statuses, load. Downside: weaker on human in-browser behavior. For SEO tech, logs are often essential.
When to use which:
- tag — marketing and CRO
- logs — bots, crawling, 5xx
- both — disputed traffic spikes
- do not confuse bots with “organic growth”
Yandex Metrica and Google Analytics
In RU, teams often start with Metrica: goals, segments, session replay, easy link to Yandex Direct. Google Analytics (GA4) is the Google ecosystem standard and useful for cross-project comparisons.
Install the code correctly, set key events or goals, mark your own visits. Then UTMs and end-to-end logic with CRM if you count sales.
Day-one start:
- tag on site templates
- 2–5 goals (lead, call, purchase)
- real-time check
- team access with the right roles
How not to drown in reports
Pick one main goal for the period and 5–7 metrics. Everything else — on demand. Put KPIs on a simple dashboard so debates are about numbers, not screenshots.
Tools speed decisions but do not replace site and ad work. Growing organic visibility for a query core is a separate horizon — typically months after SEO work starts, following prep.
Rhythm:
- daily — anomalies and ads
- weekly — channels and goals
- monthly — trends and hypotheses
- after release — traffic and error smoke checks
FAQ
Tag or log analyzer?
A JS tag sees in-browser behavior and suits marketing. Server logs are stronger on tech and bots. In practice you often need both.
Is Metrica alone enough?
For many RU projects — yes as a base. GA4 helps in parallel if you watch Google traffic and that report ecosystem.
What should you set up first?
Code on every URL that matters, goals or events, filter your own IPs, UTMs on ads. Without goals, pretty charts do not decide anything.
Does analytics replace an SEO audit?
No. It shows symptoms — bounces, sources, load speed as users feel it. Tech and indexing still need a crawler and Webmaster / Search Console.
Does everyone need session replay?
On problem URLs, yes — selectively. Watching every session — no. Numbers first, then a sample of recordings.
Why do Metrica and GA numbers diverge?
Different session models, bots, cookie consent, and definitions. Compare trends inside one system.
When do you need BI / a dashboard?
When sources multiply and reports are assembled by hand every week. Start with 5–7 KPIs, then complicate.
Will analytics put you on page one?
Not directly. It speeds decisions. Rankings come from work on the site; growing a query core is typically planned over months after SEO starts.
Counters are in — and weekly reports are still screenshots without goals?
We’ll wire goals, UTMs, and a short KPI set — decisions from numbers, not from pretty graphs alone.
Discuss the task
