How Modern Buying Broke Attribution Models

And What You Should Do About It in 2026

Before we get into it, a quick note…. I’m back.

I haven’t written here since mid to late September. After two and a half years of writing consistently, I took a much needed break. Not from the work, but from publishing. Stepping back gave me some distance and honestly, more conviction about what actually matters.

This year, I want this newsletter to be more in-depth. Fewer takes. More systems thinking. More things you can actually use to make better decisions.

Which brings us to attribution.

Let’s kick off 2026 with an argument we have apparently decided to have every single year.Can we finally admit attribution is mostly broken by modern buying behavior and that companies waste an incredible amount of time pretending it is more precise than it is.

2 stats I saw recently that prompted this article:

• 91 percent of marketers say attribution is critical to success.

• Only 31 percent say they are confident in their models.

That gap tells you everything.

At the same time, CMOs use attribution outputs to directly drive budget decisions. That is the risky part. When uncertain inputs drive real money, you do not just get noise. You get misallocation at scale.

Teams are not irrational. They chase attribution because leadership wants certainty. Boards want clean answers. Finance wants line items that roll up neatly to revenue.

But buying does not behave neatly.

How Buying Actually Happens Now

Most attribution models still assume three things.

1. Buying is linear.

2. Influence is observable.

3. The most recent measurable action deserves the most credit.

Here is what actually happens instead.

• An exec sees a founder post on LinkedIn three months ago.

• A teammate shares a blog post in Slack.

• Someone asks ChatGPT or another AI tool, “What are the best tools for X?” and your company shows up in the answer.

• They click nothing. They just remember the name.

• Later, a peer mentions you at a dinner.

• Weeks after that, they search your brand directly and click a branded paid search ad.

• Attribution assigns most of the credit to paid search because it is the last measurable touch.

• Everything else gets treated as background noise, even though that is where conviction was built.

This is where attribution models are falling even further behind reality. This is where attribution models are falling even further behind reality.

And let me guess what every other marketer will say next.

“Just ask them on the form!”

As if that is some kind of magic bullet. Self-reported attribution feels comforting because it looks direct. In reality, it is one of the most biased data sources we rely on.

Multiple studies have shown that buyer recall is unreliable, especially in complex decisions. People over-index on the most recent interaction, the most socially acceptable answer, or the option that feels easiest to explain in a sentence.

In B2B buying environments with long sales cycles, buyers are often exposed to dozens of touches before converting. Yet when asked “how did you hear about us,” most responses collapse all of that into a single answer.

Research has consistently shown that self-reported source data skews heavily toward last-touch channels like search, referrals, or direct, even when earlier influences played a meaningful role. In some analyses, over half of respondents misattribute their first or most influential touch when compared to behavioral data.

This gets even worse when AI enters the picture.

If someone asks ChatGPT or another LLM, “What are the best tools for X,” sees your brand listed, remembers the name, and converts weeks later, what do you think they put on the form.

“Google.”

“Referral.”

“Not sure.”

They are not lying. They are answering to the best of their memory. The system just cannot capture what actually happened. So now we have attribution models that cannot see AI-assisted discovery and form data that confidently reinforces the wrong conclusion.

Which means teams feel even more justified over-investing in the channels that show up cleanly and under-investing in the ones that actually shape demand.

That is not a data problem. That is a systems misunderstanding.

AI-assisted discovery is now part of the buying journey, but it produces almost no clean, attributable signals. Marketers know it matters. They are just not sure how to manage or measure it yet. So the system quietly ignores it.

The Hidden Cost of Attribution Obsession

The biggest cost of attribution is not imperfect reporting. It is the behavior it incentivizes.

Three failure modes show up over and over again.

• First, teams over-optimize for what is measurable. Demand capture gets funded. Demand creation gets questioned. Over time, the funnel looks efficient right up until it dries up.

• Second, strategy gets replaced by debate. Teams spend hours arguing whether a channel influenced a deal by 12 percent or 18 percent instead of asking whether the message actually resonates with buyers.

• Third, creativity gets penalized. New ideas look bad in attribution models early, so teams default to incremental tweaks on what already shows up cleanly in dashboards.

Execution slows. Learning slows. The system becomes fragile.

What More Seasoned Teams Do Instead

The best teams I work with do not abandon attribution. They downgrade its authority. They use attribution directionally, not as a source of truth.

That means a few things.

They pick a simple model and document its assumptions. Everyone knows what it captures and what it does not. The goal is not accuracy. The goal is shared understanding.

They align on how buyers actually buy. Where they spend time. What reduces perceived risk. What builds trust over weeks and months.

If you sell to marketers, you do not need an attribution model to tell you that consistent thought leadership on LinkedIn influences outcomes. Your buyers are there every day. The mistake is waiting for a dashboard to give permission to invest.

Most importantly, they build signal systems, not credit systems.

Instead of asking which channel gets credit, they ask better questions.

• Are we showing up consistently in the places that matter?

• Are sales hearing our name unprompted more often?

• Is deal velocity improving for accounts exposed to our content?

• Are branded searches trending up after major initiatives?

No single metric proves success. Patterns over time do.

That is how more experienced marketing teams make decisions when the system cannot be fully observed.

The Practical Reframe for 2026

Stop chasing precision. It does not exist. Align on a reasonable, imperfect model. Agree on which channels matter and why. Accept that AI-driven discovery is now part of the system, even if it is not yet measurable.

Then execute relentlessly. Pick the channels that matter.mStop over-justifying obvious bets.

Spend less time defending the model and more time building conviction in the market.

The teams that win in 2026 will not be the ones with the cleanest dashboards. They will be the ones that understand the system best and move faster because of it.

Thanks for reading!

Adam