How do I compare attribution models for a conversion?

Short answer

Open All-platform analytics › Behavior › Behavior analysis and choose the attribution view. Set the goal (one event or page), choose what counts as a touch (channel by default) and the lookback window (30 days by default), then tick the models to compare: first touch, last touch, linear, position and time decay. Each ticked model appears as a column of credited conversions for every touch value.

Set up the comparison

1. Go to All-platform analytics › Behavior › Behavior analysis and switch to attribution.
2. Goal (required). Pick one event or page, written the same way as a funnel step, for example the purchase event or the /thank-you page. If you have not defined it yet, see set up goals.
3. Touch. Channel is the default; when a session has no channel, its referrer channel is used. You can switch to referrer channel, UTM source, UTM campaign, referrer domain or landing page.
4. Lookback window. 1 to 90 days, default 30. It includes the session in which the conversion happened.
5. Models. Tick at least one:
   • First touch: all credit to the first session in the window.
   • Last touch: all credit to the last session.
   • Linear: equal shares across sessions.
   • Position: one touch gets everything; two split it in half; with more, first and last get 40% each and the middle shares 20%.
   • Time decay: closer sessions weigh more; half-life 7 days by default, 1 to 90.
6. Optional earlier event. Keep only touches that happen after this event and before the conversion, for example only touches after signup when the goal is purchase.
7. Run the analysis. Identity defaults to person: the conversion and its sessions are joined on the same person.

Read the result

At the top are four numbers:

Below, each row is one touch value (for example "Organic search" or "spring_sale"):

• One column per model with the conversions credited to that value.
• Paths: conversions that included this touch at all.
• Touch sessions: sessions in the range with this value.
• Touch rate: paths ÷ touch sessions.

Compare the columns rather than reading one: a value strong in first touch but weak in last touch opens journeys that other channels close. A worked comparison is in multi-touch attribution.

Which model to look at

Limits

• Up to 50,000 conversions are scored per query. The headline conversion count is the full total, so above 50,000 the unattributed number looks inflated.
• Rows are ordered by last-touch credit and capped at 200. An empty touch value shows as none.
• There is no breakdown and no compare period in this view.
• Time decay uses the seconds from each session's start to the conversion.
• The analysis does not rewrite the source fields already stored on events.
• It uses on-site sessions. App install attribution from an outside provider is a separate report under App analytics.

Why the numbers differ from ad platforms

Ad platforms credit conversions to their own ads, often including views and longer windows, and each platform counts independently, so their totals usually add up to more than your real conversions. TapCub credits only sessions it observed on your site within the window you chose. Untagged links fall into direct or referral, so consistent UTM tags make the report more useful; see sources and UTM.

Related

• Multi-touch attribution: models compared on one journey
• Attribution
• Set up goals
• Sources and UTM