Attribution is the most consequential problem in marketing that nobody has solved. Which part of your strategy is actually producing results? Most teams answer that question with a model they know is wrong, because the alternative is answering it with nothing.

On a recent episode of the SaaS Stories podcast, I spoke with Jeff Greenfield, founder of Provalytics, who has built a career on this exact problem. We covered the mechanics of attribution, what privacy law did to it, and how marketers can operate in a world with far less data than they were promised.

Listen to the full episode of SaaS Stories with Jeff Greenfield

The conundrum

Marketers have wrestled with one question for a century: where did the sale come from? Jeff opened with the adage everyone in this industry knows — half the money spent on advertising is wasted, and nobody knows which half. Technology has not retired it. Omnichannel marketing has made it worse.

Last-click and first-click models are oversimplifications dressed as answers, and Jeff has the best description of the problem I've heard.

"Attribution models like these are like giving all the credit for buying a six-pack of beer to the checkout clerk who scanned it."

They ignore the touchpoints that did the persuading, which means budget follows the wrong signal. In B2B the effect is sharper still, because lead source is often a single CRM field, and that field quietly decides an argument between marketing and sales about who earned the number. E-commerce marketers have the mirror version, trying to attribute conversions across marketplaces, social platforms and their own site.

What privacy law changed

GDPR, Apple's iOS updates and the regulations that followed limited the tracking of individual behaviour. That's a good outcome for consumers and a genuine problem for anyone whose measurement depended on user-level data.

Jeff's team at Provalytics responded by blending the useful parts of media mix modelling with multi-touch attribution, producing a privacy-safe method of understanding campaign performance that doesn't need to follow individuals around the internet. The constraint forced a better approach, which is usually how this goes.

Think like a consumer brand

B2B marketers default to events, white papers and LinkedIn ads. They work. Jeff's argument is that stopping there costs you reach you could have cheaply.

"Even in B2B, you're marketing to people, not businesses. These decision-makers are also consumers, spending their downtime on platforms like Facebook, YouTube, and Reddit."

Connected TV advertising reaching decision-makers at home creates an impression that a sponsored post doesn't. Jeff points to LinkedIn's combination of professional targeting with CTV as a genuine shift for B2B marketers who've never had a way to buy that audience outside work hours.

Where AI fits

Provalytics uses a custom-trained model to run its four-step attribution process without human intervention at each stage, which produces faster and more consistent insight.

Jeff's view of where this goes is more interesting than the automation. He expects AI to move from analysing data to executing campaigns — dynamically allocating budget across platforms and optimising spend in real time. The distinction he draws matters: Google's and Meta's AI campaign tools optimise inside their own ecosystems, where they have an obvious interest in the answer. An independent attribution platform can give you a holistic and unbiased view.

"The ultimate goal is to create a single source of truth that aligns marketing, finance, and operations teams. When everyone is working from the same data set, it's easier to make confident, data-driven decisions."

The founder question

Jeff also talked about work and life, which isn't the usual detour on a marketing podcast. Early in his career he worked constantly and let his health and personal life absorb the cost. He now takes weekends and protects time with family.

"What good is the journey if you get there and you're alone and unhealthy?"

His point is that the recovery made him a better leader, not merely a happier one. Worth sitting with if you're building something.

Where attribution goes

As privacy regulation tightens further, measurement has to work without intrusive collection, and AI and advanced modelling will carry much of that transition.

Technology won't finish the job on its own, though. Marketers need to think differently — more creative, more collaborative, more willing to run an experiment that might fail. Combine the modelling with actual judgement and attribution stops being a reporting argument and starts being a growth advantage.

If you're still running last-click, replace it with anything that acknowledges more than one touchpoint. That single change will tell you more than another quarter of the same reports.