W-Shaped Attribution
W-shaped attribution is a multi-touch model that gives major credit to three milestones: the first touch, the lead-creation touch, and the opportunity-creation touch, with the rest spread across other interactions.
Key takeaways
- Credit anchors to three dated milestones: first touch, lead creation and opportunity creation.
- Each milestone takes roughly thirty percent, leaving ten for all remaining interactions.
- The model depends on lifecycle stamps being set when the event actually happened.
- Manually created or reopened opportunities break the milestone logic entirely.
- Self-serve businesses without an opportunity stage cannot apply the three-peak shape.
In depth
W-shaped attribution anchors credit to three dated events on the record rather than to positions in a click path: the first touch, the touch present when the contact became a known lead, and the touch present when the record became an opportunity. Each of those three takes roughly thirty percent and the remaining ten is shared among every other interaction. The model therefore needs two things positional models do not, namely reliable lifecycle date stamps and a way to find which touch sits at each stamp.
Everything depends on how those milestones are defined and enforced. If lead status is set by a bulk import or by an operator tidying records, the middle peak lands on nothing meaningful. If opportunity creation is logged days after the conversation that produced it, the third peak credits a follow-up email instead of the demo. Tightening definitions improves the model but demands sales discipline, while widening the small residual pool gives nurture touches visible credit at the cost of blurring the milestone story.
The shape is built for organisations where marketing and sales share one pipeline number. Marketing can defend the first two peaks, sales owns the third, and the argument shifts from who deserves credit to whether the milestones are stamped correctly. Where a scorecard quiz creates and scores the lead, it usually takes the middle peak while the ad that drove the visit takes the first, so a single quiz-driven journey shows demand creation and qualification side by side in the same report.
W-shaped attribution measures process stages rather than persuasion, so it rewards whichever touch happened to be present when a status field changed. Deals that skip a stage, get reopened, or are created manually by a representative break the logic entirely, and self-serve businesses with no opportunity concept cannot apply it at all. Concentrating ninety percent of credit into three moments also makes long nurture sequences and any marketing that happens after opportunity creation effectively invisible.
Example in practice
How to measure it
Report the three peak totals separately and read them as a funnel rather than as a ranking: which channels sit at first touch, which at lead creation, which at opportunity creation. A channel appearing under two peaks is doing two jobs. Divide spend by the credit at each peak to get both a cost per originated lead and a cost per influenced opportunity from the same model.
Then audit the joins. Track the share of opportunities with no identifiable touch at their creation timestamp, and the share of leads whose creation touch is the same event as the first touch. High numbers in either case mean the peaks are collapsing into one another, and the model is describing a two-peak journey while still claiming three.
Common mistakes
The common failure is trusting lifecycle timestamps that nobody audits. A weekly batch job flipping records to lead status every Monday concentrates the middle peak onto Monday's traffic sources, and the report quietly becomes a chart of when the automation runs. Check the distribution of milestone timestamps across days and hours; unnatural clustering means the stamp reflects an internal process rather than anything a buyer did.
The second is extending the shape without saying so. Teams add a fourth peak at the closed-won moment, keep calling the result W-shaped, and produce numbers nobody else can reproduce. If you add a milestone you are running a full-path variant, so name it that way and restate the weights. Attribution disputes are almost always disputes about undocumented configuration rather than about the underlying data.
Frequently asked questions
What are the three touchpoints in W-shaped attribution?
The first touch that introduced the person, the touch present when they became a known lead by identifying themselves, and the touch present when the record was converted into a sales opportunity. Everything else on the path shares a small residual. The three are milestones in your lifecycle data, not positions counted from either end of the path.
What weights does W-shaped attribution use?
The conventional split is thirty percent to each of the three milestones and ten percent shared across all remaining touches. The exact figures are a convention rather than a standard, so some teams use a slightly larger residual to keep nurture visible. Whatever you choose, apply it consistently and state it beside every report.
What do I need in my CRM before using W-shaped attribution?
Two reliable date fields, one for when a contact became a lead and one for when an opportunity was created, plus a marketing touch history that can be joined to the same record. Without accurate timestamps the middle and third peaks land arbitrarily. Fix the stamping process before configuring any weights.
When should I use W-shaped instead of U-shaped attribution?
When there is a real handover between marketing and sales and both need credit for the stage they influenced. The extra peak exists to represent that handover. If your funnel has no opportunity concept, or the stage is stamped inconsistently, the U-shaped model with two peaks will describe reality more honestly.
Does W-shaped attribution work for self-serve or ecommerce businesses?
Rarely, because there is no opportunity-creation moment to anchor the third peak. Some teams substitute a proxy milestone such as trial start or first meaningful product action, which can work if the event is logged reliably and genuinely marks a change in intent. Otherwise a two-peak or time-weighted model is a better fit.
What is the difference between W-shaped and full-path attribution?
Full-path adds a fourth milestone at the closed-won moment, so credit is spread across four peaks instead of three, usually around twenty-two percent each with a small residual. It suits teams whose marketing continues to influence late-stage deals. The trade-off is another lifecycle stamp that has to be accurate for the model to mean anything.