Funnel

Funnel (/funnel) shows where traffic drops along the path from source to conversion. Unlike Reports (cut tables), Funnel answers where the funnel breaks and shows transition rates between steps.

Funnel
Funnel builder: mode, steps, visualization, and result.

Purpose

When campaign CR drops, Reports can show bad by country or sub_id but not whether the break is at landing, offer, or conversion. Funnel builds a step chain and counts how many reached each step. Modes: conversion statuses, path entities, or custom events.

Prerequisites

  • Clicks exist for the period; later steps need landing/offer views and/or conversions.
  • Permission for the funnels page.
  • For events mode — familiarity with funnel registry tables/fields.

How to open

  1. Sidebar → Funnel.
  2. Path: /funnel.
  3. Optional: shared link with ?token=... loads a shared preset config.

Funnel modes

status — steps by conversion funnel statuses (e.g. click → lead → sale → registration → deposit → rebill → rejected).

path — tracking entity stages in path order:

  1. traffic_source
  2. campaign
  3. flow
  4. landing_page
  5. offer
  6. conversion

You can select a subset of steps; order reflects the traffic path.

events — custom step conditions (table / field / operator / value) when status/path are not enough.

Choosing a mode

If the question is how many leads reached sale/deposit — use status. If it is which tracking entity drops (landing vs offer) — path. If you need a condition outside standard steps — events. Do not mix expectations: status does not draw a landing_page stage; path does not replace CPA status analysis.

Step-by-step workflow

  1. Set the period. Start with Last 7 days: too narrow is noisy, too wide slows Run on large accounts.
  2. Pick mode: status, path, or events.
  3. In status/path select steps. In path you may use a subset (e.g. campaign → landing_page → offer → conversion) while keeping path order. In events add step conditions one by one: table / field / operator / value.
  4. Add filters: campaign, offer, traffic source, flow, landing, country, device, sub_id, UTM, etc. Without a campaign filter, path on the whole account mixes unrelated bundles.
  5. Optionally set a breakdown (e.g. country or device) to compare segments on one funnel.
  6. Choose visualization: bars/funnel for drop overview; table for exact counts/%; trend for day dynamics; sankey/segmented for flow comparison. Weekly review usually needs funnel + table.
  7. Click Run. Read counts and transition %. The largest % drop between neighbors is the first hypothesis.
  8. Save a preset named like Path / campaign X / 7d.

Path fields in context

traffic_source — entry from the source. campaign — click already in the campaign. flow — in-campaign routing. landing_page — landing pass; sharp drops often mean prelander/A-B or a dead URL (check Monitoring). offer — offer transition; drops point to flow redirects/filters or offer downtime. conversion — recorded conversion; with healthy offers check postbacks, goals, and traffic quality (Logs, Fraud).

In status, steps like click → lead → sale → registration → deposit → rebill → rejected follow the catalog. A zero lead→sale with live leads in Logs usually means postback status/goal mapping, not a Funnel bug.

How to interpret

Find the largest relative drop between neighbors, then confirm absolute numbers in Reports for the same period and campaign. If early-step counts are already near zero, the problem is before the funnel (no clicks, filter, timezone). If early steps are live and conversion is empty, follow Logs → Clicks → Postbacks → Conversions. With breakdown, compare segments only at comparable volume.

Presets and sharing

Save working configs with a clear name. A shared ?token=... loads another preset for view/run. If the link fails, rebuild and save your own. A preset does not freeze numbers — each Run uses current data for the selected period.

Typical scenarios

  • Single campaign, path. Filter campaign = X, path steps, viz = funnel. On a landing_page drop open Monitoring and a test click in Logs.
  • Geo compare. Same path + country filter or country breakdown; then deepen bad geos in Reports.
  • CPA status funnel. Mode status: click → lead → sale (deposit/rebill if needed); filter by affiliate network.
  • After a flow/landing change. Period around the change + the same path preset; compare landing→offer and offer→conversion %.
  • Events for a non-standard step. One condition per step, Run, cross-check counts with Logs.

Troubleshooting

  • All steps zero: period, account timezone, campaign filter; clicks in Logs for the same dates.
  • Early steps live, conversion = 0: Incoming postback, Conversion Catalog / Postback Mapping, test postback.
  • Drop at offer while Monitoring is green: check flow filters/branches, not only HTTP 200.
  • Events empty/odd: simplify the condition; compare with path/status on the same slice.

Common mistakes

  • Running path with no campaign filter on a large account.
  • Confusing status vs path: status does not show landing_page as an entity stage.
  • Treating a zero step as a Funnel bug before checking Logs for the same period.
  • In events, setting a condition without understanding the field.
  • Optimizing CR from Funnel while Monitoring shows the offer down.