Scale or Kill? Running Last-Click Against Linear on the Same Channel Table
Objective: Given a real 6-channel conversion-credit table computed two ways (last-click and linear), apply the lesson's 25%-delta rule to decide which channels are discovery engines versus closers.
You're the paid media analyst at Warby Parker deciding Q3 budget across six channels using last quarter's multi-model attribution export.
Compare last-click and linear credit per channel, flag any channel with more than a 25-point delta between the two, and recommend scale, hold, or kill for each.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, handles a 6-row comparison with no setup
The process
2 steps
Step 01 of 02
The lesson's playbook says: run two models in parallel and look at the delta. If linear and last-click disagree by more than 25 percent on a channel, that channel is either a discovery engine or a closer, treat it differently.
Six channels are listed below with their last-click and linear credit share. Which channels cross the 25-point delta threshold, and in which direction?
Procedure
- Import channel-attribution-export.csv (6 rows: channel, last-click %, linear %)
- Add a delta column = last-click % minus linear %
- Sort by absolute delta, descending
- Flag any row where |delta| > 25 points
Channel Last-click Linear Delta Retargeting 34% 9% +25 pts (flag) Branded Search 22% 11% +11 pts Display (prospecting) 4% 18% -14 pts Organic Social 8% 16% -8 pts Email 19% 14% +5 pts Affiliate 13% 32% -19 pts
Healthy
Retargeting is correctly flagged: it closes deals at 34% last-click but only creates 9% of linear-weighted demand, that is a closer, not a discovery channel.
Unhealthy
Reading Retargeting's 34% last-click share as proof it should get more prospecting budget, when the linear number shows it is intercepting demand created elsewhere.
What this means
A channel that scores high on last-click and low on linear is a closer; scale it in the final-touch position, not as a top-of-funnel driver.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| A channel's last-click share is used alone to justify a prospecting budget increase | Check its linear share first; a wide positive delta means it is closing, not creating, demand | 5 min |
Step 02 of 02
The lesson says linear gives equal credit to every touch, and lies about importance, a throwaway display impression gets the same weight as a 20-minute webinar, honest only when touches really are interchangeable.
Affiliate scores 32% under linear but only 13% under last-click. Before recommending Affiliate get more budget, what does the lesson say to check about whether its touches are actually interchangeable with the others?
Procedure
- Isolate the Affiliate row and its raw touch count across conversion paths
- Check whether Affiliate touches cluster near the start of paths (discovery) or are spread evenly
- Note that linear alone cannot distinguish 'many small assist touches' from 'one genuinely important touch counted many times'
Affiliate: 32% linear credit, appears in 61% of converting paths, average position: touch 1 of 4 (early) Interpretation: high touch frequency early in the path, likely real discovery value, not an artifact
Healthy
Affiliate's high linear score is corroborated by it appearing early and frequently across paths, a genuine discovery signal worth the budget recommendation.
Unhealthy
Accepting Affiliate's 32% linear number at face value without checking path position, when linear treats every touch as equally important by construction.
What this means
Linear's number is only trustworthy once you've confirmed the touches it's crediting are actually similar in importance to each other.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| A channel scores unexpectedly high under linear attribution | Check its typical path position before trusting the number for a budget call | 5 min |
Final deliverable
A one-page budget recommendation memo: scale, hold, or kill per channel, with each call justified by its delta and path position.
See a reference example
Lenskart, Q3 channel call (excerpt) Retargeting: HOLD as closer, do not fund as a prospecting channel (delta +25pts) Affiliate: SCALE, confirmed early-path discovery role (32% linear, touch 1 of 4 avg) Display: SCALE cautiously, low last-click but real linear assist role (delta -14pts)
Success criteria
You're done when you can:
- Correctly flags every channel with a delta greater than 25 points
- Distinguishes a closer (high last-click, low linear) from a discovery channel (low last-click, high linear)
- Checks Affiliate's path position before trusting its linear score