Most attribution guides hand you a model and tell you how to divide the credit. This one does the opposite. “Wittgenstein’s attribution model” is the idea that before you trust any model, you should ask whether the instrument is reliable in the first place, because a bad one quietly tells you more about yourself than about your marketing.
The phrase comes from Alex Birkett of Omniscient Digital, who borrowed a concept known as Wittgenstein’s ruler and pointed it at marketing measurement. It is less a model to implement than a lens to keep you honest.
What is Wittgenstein’s attribution model?
Wittgenstein’s attribution model is the principle that an unreliable attribution model measures your own assumptions as much as it measures reality. If you don’t trust the instrument, you can no longer tell whether it is measuring your marketing or your marketing is exposing the flaws in the instrument.
The idea originates not with Ludwig Wittgenstein directly but with Nassim Taleb, who coined “Wittgenstein’s ruler” in Fooled by Randomness:
- “Unless you have confidence in the ruler’s reliability, if you use a ruler to measure a table you may also be using the table to measure the ruler.” - Nassim Nicholas Taleb, Fooled by Randomness
Apply that to attribution and the implication is uncomfortable. When a last-click report tells you paid search drives all your revenue, that finding might reflect how buyers behave. Or it might just reflect the fact that last-click can only ever see the final touch. Without confidence in the ruler, you cannot tell which.
Why it matters for marketers
Attribution models are routinely treated as objective instruments, like a thermometer or a set of scales. They aren’t. Every model encodes an assumption about how buying works, and that assumption shapes the answer before any data arrives. Birkett’s argument is that in messy, real-world conditions, especially B2B with its long cycles and dark social, the model often reveals the analyst’s bias more than the buyer’s behaviour.
This creates three failure modes worth naming:
- Attribution as office politics. Once everyone knows the model decides who gets credit, the debate stops being about truth and starts being about budget. Whoever picks the model picks the winner.
- Over-engineering small problems. Teams with narrow markets and a handful of channels reach for elaborate multi-touch models, generating precise-looking numbers built on far too little data.
- Mistaking the map for the territory. Hours vanish into debating model shapes while the metrics that actually signal business health, win rates and pipeline growth, go unwatched.
What to do instead
The point isn’t to abandon attribution. It’s to stop treating any single model as the verdict. Birkett’s prescription is closer to good scientific hygiene than to a new measurement technique:
- Lead with business KPIs. Start from revenue, pipeline, and win rate. Use attribution to interrogate those numbers, not to replace them.
- Point attribution at specific questions. “Why did win rates drop last quarter?” is a question a model can help answer. “What is our true attribution?” is not.
- Match the model to the context. Tailor the approach to your actual business rather than importing a standardised model because a vendor recommended it.
- Protect long-term brand investment. Channels that are easy to measure crowd out channels that are hard to measure but valuable. Don’t let the ruler decide your strategy.
Questions to ask yourself
As a modern growth or agile marketing professional, hold your attribution up to Wittgenstein’s ruler:
- Do I actually trust this model, or am I trusting it because it is the default in my analytics tool?
- If this report flatters a particular channel, is that because the channel works, or because the model can only see that channel?
- Am I using attribution to make a decision, or to win an internal argument?
- Have I checked the model’s story against an incrementality test or a simple “How did you hear about us?” question?
- Is the complexity of my measurement justified by the volume and complexity of my marketing?
Related articles
- Marketing Attribution Models: The Definitive Guide
- Uncovering the attribution mirage
- What is last-click attribution?
- What is data-driven attribution?
- What is self-reported attribution?
- What is multi-touch attribution?
About Growth Method
Growth Method is the growth platform designed for experiment-led and data-driven marketers.
Learn more on our homepage, connect with me on LinkedIn or Twitter, or book a call here.