How do you attribute a closed deal?
First-touch, last-touch and multi-touch each answer a different question about the same deal. Multi-touch is the most honest but requires every channel to actually be logged.
The short answer
- First-touch attribution credits whatever channel first brought the lead into the system; last-touch credits whatever channel or action happened right before the deal closed. Neither is wrong, they measure different things.
- Multi-touch attribution credits some or all of the touches across the deal's lifecycle, giving a more complete picture of the journey, but it only works if every touch, email, call, LinkedIn message, form fill, was actually logged somewhere trackable.
- The model to use depends on the decision it needs to inform: understanding which channels generate initial interest calls for first-touch data, while understanding which activities actually close deals calls for last-touch or multi-touch data.
- The most common failure in any attribution model is a silent tracking gap, a real touch that happened, often a phone call or an in-person conversation, that never got logged, which understates that channel's real contribution without anyone noticing.
Why 'what closed the deal' is the wrong single question
Asking which channel or action gets credit for a closed deal assumes there is one right answer, but a deal that closes after a cold email, three follow-up calls, and a LinkedIn message did not close because of any single one of those touches in isolation, it closed because of the sequence. First-touch attribution answers "what got this person's attention in the first place," which is useful for evaluating top-of-funnel channels. Last-touch attribution answers "what happened right before they said yes," which is useful for evaluating closing activities. Neither answers the full question of what the whole journey actually looked like.
Multi-touch attribution tries to answer that fuller question by distributing credit across some or all of the touches in a deal's history, but it depends entirely on those touches actually existing as data. A phone call that happened but was never logged, a LinkedIn message sent outside any tracked system, an in-person conversation at an event, none of these show up in a multi-touch model if they were never recorded, which silently understates whatever channel carried that untracked activity.
Attribution models compared
| Model | Answers | Weakness |
|---|---|---|
| First-touch | What first brought this lead in | Ignores everything that happened after |
| Last-touch | What happened right before it closed | Ignores everything that built up to that point |
| Multi-touch | The full journey across all touches | Only as complete as your tracking |
What to check before trusting attribution data
Audit whether every channel your team actually uses, email, calls, LinkedIn, in-person, forms, is logged against the deal record consistently, not only the channels that happen to be easy to track automatically. A model built on incomplete data will confidently point to the wrong conclusion, since it has no way to represent activity it never saw.
Disclosure: SalesCrew is our product. Deals, contacts, timeline, and channel activity already live in one database today, which is what makes multi-touch attribution possible without combining exports from separate tools. The completeness point above still applies: any gap in what gets logged is a gap in the resulting attribution, regardless of which system tracks it.
A tracking gap is invisible until you go looking for it
Questions
- Which attribution model is 'correct'?
- None of them is objectively correct; they answer different questions. First-touch answers what got attention, last-touch answers what closed the deal, multi-touch answers what the whole journey looked like. Pick based on what decision the attribution is meant to inform.
- Can you run more than one model at once?
- Yes, and many teams do, looking at first-touch for channel discovery decisions and last-touch or multi-touch for closing-channel decisions. The models are not mutually exclusive; they just answer different questions from the same data.
- What breaks multi-touch attribution most often?
- Gaps in tracking, a touch that happened but was never logged, most commonly a phone call, a LinkedIn message, or an in-person conversation that never made it into any system, which silently understates that channel's real contribution.