Urban Sprout’s 2026 AI Attribution Challenge

By 2026, Anya Sharma, Head of Growth at the sustainable home goods brand “Urban Sprout,” had a problem. Ad spend was up but returns were dropping, and none of her team’s old attribution models could explain it. The real issue was assigning credit for sales that started with AI-assembled consideration sets, those product lists from recommendation engines or personalized searches customers saw long before an ad. Last-click was useless here. Even their multi-touch models couldn’t map these invisible paths. This blind spot around AI attribution was burning through their marketing budget because they couldn’t see what was actually working, forcing Anya to figure out how to measure the real impact of these AI-driven touchpoints.

Where AI’s Influence Hides in Your Analytics

Urban Sprout’s customers, who were buying everything from bamboo kitchenware to solar-powered garden lights, were the type to do a ton of research first. They weren’t clicking banner ads. They were browsing “eco-friendly living” guides on AI content platforms, getting product recommendations from their smart home devices, or using AI chatbots to find products that matched their values. Anya laid it out for her team in a meeting: “We get a big organic traffic spike after some AI lifestyle blog features us, but our analytics can’t connect that specific interaction to a sale. It just gets dumped into ‘direct’ or ‘organic,’ which is useless for deciding where the next dollar goes.”

AI-assembled consideration sets work invisibly, influencing decisions days or weeks before a purchase, which is why old models break. Someone might see an Urban Sprout compost bin suggested by a smart assistant when they ask about “sustainable waste solutions,” then a week later search directly for “Urban Sprout compost bin” on Google. That first AI suggestion planted the seed but gets zero credit in any standard model. This is exactly why zero-click attribution is so important. It’s built to see the value in these moments *before* a click happens, which is where AI does a lot of its work.

An IAB report on attribution from 2025 backs this up, estimating that almost 40% of digital customer journeys had at least one AI-driven touchpoint that never got a click but still heavily influenced the final purchase. That’s a massive blind spot if you’re only counting clicks. For Anya, the job wasn’t just to spot these AI touchpoints, but to finally put a number on their value.

Building an AI Attribution System That Actually Works

Anya knew a patchwork solution wasn’t going to cut it, so Urban Sprout set out to build a single system to pull in data from all their AI-driven sources. They started by connecting data from their on-site recommendation engine and their customer service chatbot, but they also started monitoring their brand’s appearance on third-party AI content platforms. With privacy rules getting tighter, they focused on understanding the aggregate impact of AI exposure instead of trying to track individual users across the web. As Anya put it, “We had to get over ‘who clicked what’ and start asking ‘what built the consideration set’.” That meant they started tracking new things like time spent on AI product lists, how far people scrolled on recommendation carousels, and engagement with AI summaries.

The analytics team got to work, segmenting site visitors by how they interacted with AI features. They built cohorts for users who had asked the chatbot product questions versus those who hadn’t. They also tracked visitors landing on product pages from AI-powered search snippets or widgets on partner sites. Once anonymized and aggregated, this data started to reveal the shape of AI’s pre-click influence.

Assigning a dollar value was the next big challenge. Linear or time-decay models were completely out of their depth. So, the team looked into more advanced stats, landing on Shapley values. It’s a concept from game theory for splitting up a prize in a team game, which turned out to be perfect for assigning credit across marketing channels. Every AI touchpoint, a product showing up in an AI top-10 list, a smart assistant suggestion, was treated like a “player” on the team. The model then calculates its specific contribution to the final sale by looking at every possible sequence of events. For the first time, Anya could prove that while a paid ad got the last click, an early AI recommendation was often the real MVP that kicked off the entire journey.

The Data Plumbing: Turning AI Engagement into Usable Insights

Running Shapley values meant they needed a serious data setup capable of handling huge volumes of raw interaction data from all their AI tools. Urban Sprout invested in a platform to pull everything together, from their Customer Data Platform (CDP) logs on on-site AI use to API feeds from third-party aggregators showing how often their products popped up in AI lists. “The amount of data was intimidating,” Anya admitted, “but we were flying blind without it.”

Defining new KPIs for this AI engagement was a key piece of the puzzle. They stopped looking only at conversion rates and started tracking:

• AI-influenced consideration rate: What percentage of users saw an AI recommendation and then visited that product page within 72 hours?
• AI-assisted search conversion rate: The conversion rate for people whose first search queries were obviously influenced by a prior AI interaction, even if they didn’t click an ad.
• Average time to conversion for AI-exposed users: How much faster did users exposed to AI-curated lists buy something compared to those who weren’t?

Putting these new metrics together with the Shapley model uncovered something huge: while paid search ads were still good for quick conversions, customers who were first exposed to Urban Sprout products via AI recommendations had a 15% shorter journey to purchase and an 8% higher average order value. This was the proof. AI wasn’t just creating awareness. It was building lists of highly-qualified customers who were primed to buy before they ever clicked an ad.

Tuning the Model and Proving It Works

The first version of the model had its problems, mainly in separating real AI influence from other organic noise. The team kept tightening their data filters and even used machine learning to spot patterns that screamed “AI exposure.” For instance, they found that customers who used the AI chatbot’s “sustainable alternatives” feature were way more likely to buy a more expensive, ethically sourced product than people who just clicked through categories. That insight helped them give the chatbot more accurate credit in the Shapley model.

A perfect example of the new model’s power came up almost immediately. The team was about to cut spending on an AI product discovery platform because its click-through rates were terrible. But the new attribution model showed this platform was the very first touchpoint for a huge group of their best customers, planting the initial seed. These users would see a product there, then later do a direct brand search on Google or the Urban Sprout site, completely bypassing any trackable click from the platform. If they’d relied on the old click-based model, they would have killed a channel that was introducing them to their most valuable shoppers.

When Anya presented these findings to leadership, she could show a hard ROI for their AI discovery efforts. “We’re not guessing anymore,” she told them. “We can finally put a number on AI’s influence and invest in the platforms that are actually building our customer’s shopping list.” Based on that, leadership signed off on more budget for AI content partners and for improving their own recommendation engines, since they now had a reliable way to measure the return.

Switching to this AI-aware, zero-click model gave Urban Sprout more than just a better way to allocate budget. It gave them a real understanding of their customers in 2026. When AI is part of nearly every digital interaction, you can’t afford to ignore its effect on the consideration set, that’s like ignoring the top half of your funnel. This new system gave them the clarity to operate effectively in a marketing world that’s only getting more complicated.

It’s an approach that fits with where marketing is headed, as AI is completely changing what it means to be a marketer. By getting attribution right, Urban Sprout can now use AI to sharpen its customer focus and get better campaign results. Knowing these AI-driven paths is how you’re going to drive engagement and higher CTRs from now on.

What is AI attribution in the context of marketing?

It’s the process of assigning credit to AI-driven touchpoints that influence a customer’s decision, even when there’s no direct click. This means tracking how things like AI-generated recommendations, curated content, or chatbot chats help build a customer’s consideration set and eventually lead to a sale.

Why is traditional attribution insufficient for AI-assembled consideration sets?

They primarily rely on direct clicks to assign credit. AI often influences customers through subtle, “zero-click” interactions that happen long before a purchase, like seeing a product in an AI-generated list or getting a smart assistant suggestion. Conventional models completely miss these touchpoints, giving you a warped view of the customer journey.

What is zero-click attribution and how does it relate to AI?

It’s a framework that gives value to customer interactions that happen *before* a click but still help lead to a sale. It’s perfect for AI because it properly recognizes the impact of AI-generated content or recommendations that a user just views, helping marketers see how AI shapes a customer’s initial interest and builds their shopping list.

Which statistical models are effective for attributing AI influence?

Models like Shapley values, which come from game theory, work really well because they distribute credit fairly among all touchpoints by calculating each one’s specific contribution to the sale. Other methods, like Markov chains, can also be used to model the probable path a customer takes between different AI-influenced moments.

What data points are essential for building an AI attribution model?

You need to pull data from multiple sources: interaction logs from on-site recommendation engines, AI chatbots, and personalized search results are a start. You also have to monitor engagement on third-party AI content platforms, tracking metrics like view-throughs, scroll depth, time on AI-generated lists, and whether a user does a branded search after being exposed to AI content.