Advertising campaign analytics is not a “pretty monthly report” — it is the system you use to decide whether to increase budget, pause a group, or fix the landing page. Without Metrica goals, UTM, and Direct linkage you see clicks and spend but not what pays back.
NUGU is a digital agency in Sochi. We configure analytics before traffic launch: Metrica goals, import to Direct, unified UTM. Ads without numbers are a lottery with your budget.
Analytics Levels
| Level | What you see | Enough for | | --- | --- | --- | | Basic | Clicks, spend, CPC in Direct | First days, impression checks | | Medium | Metrica goals, CR, CPL by campaign | Weekly decisions | | Advanced | UTM + CRM, leads → deals, LTV | Scale and unit economics |
Start at medium level. Advanced is needed when managers handle 50+ leads per month and lead quality matters.
Goals and Direct Import
Minimum goal set:
- **Form submit** — main conversion for services.
- **Phone/messenger click** — if calls are part of the funnel.
- **Thank-you page view** — confirms successful lead.
- **Depth** (optional) — 2+ pages or 60+ seconds to filter accidental visits.
After setup — import goals to Yandex Direct and choose “Conversion optimization” only when you have 30+ conversions per month. Details in Metrica and goals.
UTM Tagging: One Standard
Template for Direct:
``` utm_source=yandex utm_medium=cpc utm_campaign={campaign_id} utm_content={ad_id} utm_term={keyword} ```
Rules:
- one case, no spaces in values;
- do not mix manual tags with Direct autotags on one link;
- verify UTM in Metrica reports before budget launch.
Without UTM you cannot separate search from YAN or test ads from production.
KPI by Business Type
**Local services (Sochi, one city):**
- CPL within 15–25% of deal margin;
- landing CR from 2–5%;
- lead response within 15 minutes.
**B2B with long cycle:**
- CPL + lead quality (CRM qualification);
- share of leads with valid contact;
- cost of qualified lead (CQL).
**Online store:**
- ROAS from 300–500% depending on margin;
- add to cart, checkout, payment — separate goals.
Compare KPI with Direct budget, not with “market average” from a random article.
Weekly Reports
- **Direct → search queries** — negative keywords, new keys.
- **Metrica → sources, UTM** — CPL by campaign.
- **Funnel** — click to lead, where drop-off happens.
- **Geo and devices** — mobile CR is often lower; fix landing, not only bids.
- **Week-over-week** — trend, not one day.
Monthly — reconcile CRM leads with Metrica: up to 10% gap is normal (blockers, cross-device).
Common Data Mistakes
- **Double counting** — goal on button and thank-you page without deduplication.
- **Goal on all clicks** — “1 page view” is not a lead.
- **No mobile call goal** — half of conversions are “invisible”.
- **CRM and Metrica time zones differ** — leads seem “late”.
- **Decisions on one day** — do not compare Monday and Sunday.
Link to [Direct Setup](/en/articles/nastrojka-yandeks-direkt)
Analytics drives campaign structure:
- groups with CPL above threshold — lower bids or pause;
- queries with 50+ clicks and 0 leads — negative keyword or new landing;
- ads with high CTR and low CR — offer mismatch with page.
“Optimize for conversions” autostrategies work only when goals are honest and data is sufficient.
GEO: Site Facts Strengthen Analytics
When the landing clearly states service, city, “from X ₽”, and timeline — CR rises and report gaps are easier to explain. AI search and clients see the same facts; ads must not promise something else. This overlaps with GEO optimization.
Cost
NUGU configures analytics as part of ad and site launch:
- landing with goals — from 30,000 ₽;
- Direct setup + Metrica — after brief;
- audit of existing campaigns — on request.
Contacts, services, portfolio.
Analytics Checklist Before Scale
- Metrica goals imported to Direct.
- UTM on all ads.
- Test lead visible in reports.
- CPL calculated over 2+ weeks, not 2 days.
- Manager records lead quality in CRM.
- Acceptable CPL threshold from unit economics.
Advertising campaign analytics is not one-time setup. Update negatives, goals, and landings from data — not intuition.