Measurement
Marketing attribution reality check
Marketing attribution reality check is licensed CC BY 4.0. Attribution: Katafacts (katafacts.com).
Customise with AISkill file ↓1 · What it is
What it is
A marketing attribution reality check computes how much credit for a deal shifts across three deliberately simple, honest models — first-attributed, last-attributed, and an even split — applied to real touchpoint journeys instead of a shop-floor outcome. It doesn't replace your CRM's attribution model with a better one; it pressure-tests the one you already have against two equally simple alternatives, so a channel that looks authoritative under today's default doesn't quietly pass as validated when it's never actually been checked.
2 · When to use it
When to use it — and when not to
Use it when
- Marketing and Sales disagree about which channel actually deserves credit for a deal, and the CRM's default model has never been checked against an alternative.
- A budget conversation is about to lean on a channel-credit number, and you want to know honestly how much that number would change under a different, equally reasonable model.
- You want to know which channels are robustly credited regardless of model choice, and which ones only look strong under the specific model you happen to be using today.
Not when
- You already ran a real experiment (a geo holdout, incrementality test) that isolated a channel's true effect — trust that over any of this tool's three models.
- The deal genuinely had one touchpoint — there's no real attribution question to reality-check.
3 · How to fill it in
How to fill it in
- Scope area
- What deals, and what period, does this reality check cover?
- Current model
- How is channel credit actually assigned today? Name the real convention in use — including a CRM default nobody deliberately chose — not the model you wish you had.
- Cases
- For each deal: its real pipeline value, and every touchpoint with the real order it occurred in — your own record, never guessed. Credit by model and the swing between them are computed for you.
- Narrative
- Drafted from the computed swing, naming which channels are robust and which are contested.
4 · What good looks like
What good looks like
The example below continues the Beacon Analytics storyline and is directly grounded in the gap the marketing metric definition sheet already named without resolving it — the CRM's first-touch default for channel attribution — pressure-testing that exact default against two other honest models rather than inventing a new scenario.
Same example, as a downloadable xlsx workbook.
Download .xlsxAttribution reality check
Beacon Analytics — attribution reality check
SQL-generating deals — Q2 channel credit
Priya Anand, RevOps · 2026-05-15
Scope area
SQL-generating deals — Q2 channel credit
Current model
CRM first-touch default
Whichever channel's UTM tag is present on a lead's very first form fill gets full credit. Nobody at Beacon Analytics deliberately chose this — it's just what the CRM defaults to, and it's never been checked against an alternative model.
Cases
Acme Corp — SQL — outcome value 42000
Organic search (blog post) → Webinar (paid social) → Branded search / direct
Fenwick & Holt — SQL — outcome value 28000
Paid search → Case study download (organic search) → Branded search / direct
Credit by model
Narrative
Under Beacon's actual current model — first-touch — Webinar (paid social) and the case study download get zero credit in every single deal, regardless of how much they actually influenced the buyer, because neither one was literally the first touch. That's not a one-off miss; it's a systematic blind spot built into the model itself. Branded search / direct is the sharpest finding: first-touch already (correctly) assigns it zero credit, since it never opens a journey — but had Beacon used a last-touch model instead, the exact same touchpoint would swing to claiming 100% of both deals' credit, despite showing up only after organic search and paid search had already done the real work of introducing each buyer. That 0-to-100 swing, on a channel that does nothing but confirm a decision the buyer already made, is exactly the fragility this reality check exists to expose — not to hand RevOps a replacement model, but to make clear that the one in use today has never been checked against even the simplest alternative.
5 · Common mistakes
Common mistakes
Treating a CRM's default attribution field as though it were a deliberately chosen, validated model.
A default is not a decision — this tool exists specifically to check whether that default actually holds up against an equally simple alternative.
Using this tool to pick a new 'correct' attribution model and rolling it out as the new default.
Three simple models agreeing or disagreeing isn't the same as a controlled experiment — treat a large swing as a reason to investigate further, not as license to declare a new winner.
Ignoring a channel that scores zero under every model without asking why.
A channel that's genuinely never present isn't the same as a channel that's present but always in a position (like the very middle of a journey) that every simple model happens to undercount.
6 · What it connects to
What it connects to
upstream
Marketing metric definition sheet
The metric definition sheet is where the CRM's first-touch default first got named as an unowned, unvalidated model; this reality check is where that default actually gets pressure-tested.
Marketing KPI tree
A driver metric left without a real number because of the attribution gap is exactly what this reality check gives an honest, computed sensitivity check against.
downstream
7 · Where AI helps
Where AI helps
Judgement — stays yours
- Deciding whether a channel's swing is genuinely worth flagging to Sales leadership
- Deciding what a real incrementality test to actually settle a contested channel would need to look like
Analysis — AI helps
- Tightening the current-model description from raw notes into specific, checkable language
- Drafting the narrative naming which channels are robust and which are contested, from the computed swing
Drudgery — automated
- Computing credit under all three models and the swing between them
- Exporting to xlsx in the house format
9 · Rate this kata