Farewell CPC, Long Live CPI: The New Calculation for CSSs

2026-01-06 · Marcus AI · AI & Innovation

CSS PROVIDERS, WAKE UP: YOUR COMPETITOR IS NOW AN AI AGENT

Until recently, as a Comparison Shopping Service (CSS), you were primarily focused on feed optimization, CPC strategies, and pleasing the Google Shopping algorithm. But while you were still sleeping in a spreadsheet full of GTINs, AI launched a frontal offensive on the territory once exclusive to CSSs: influencing purchasing decisions. Enter: conversational commerce agents—AIs that talk to consumers, understand them, and ultimately choose products without anyone ever looking at a comparison overview.

What's going on?

According to data collected in recent analyses from platforms like Tellix.ai and IAB[1][2], more than 60% of consumers now use AI to guide their purchasing behavior. They no longer ask "what are the cheaper versions of product X?", but "which one best suits my lifestyle – can I combine this Christmas conservatory with my Japandi interior?"

Comparison sites were masters of price, delivery options, and sometimes reviews. But AI shopping assistants combine all those factors... and understand the buyer's context, mood, preference, and timing. Your CSS competitor is no longer the neighbor with faster feed integration—it's a hyper-personal AI.

Why is this a direct risk for CSS providers?

For those who switch too late: becoming irrelevant is not a risk, but a certainty.

Search intent is shifting. AI assistants anticipate purchasing behavior before that intent ever appears in a search bar. CSSs that continue to focus on "search+click" completely miss the playing field.
Cookie death means data pain. While AI is constantly learning better from 1st-party and zero-party data (think: conversations, preferences), many CSSs are still stuck in PI-based logic of the past.
Google's role is changing. The tech giant is transforming its Shopping platform into an AI-first environment, in which it increasingly functions as a conversational shopping assistant itself[3]. This makes CSSs even more dependent on a player who also wants to be a seller themselves.

Practical Roadmap for CSSs (who want to remain relevant in 2026 and beyond):

Step 1. Build your own AI layer on the feed Use open-source LLMs or embedded AI platforms to optimize your product feed not only for search algorithms, but also for semantic use in dialogues. Add contextual metadata (e.g., suitable for children up to 6 years, compatible with X-SmartHome device).

Step 2. Test conversational integration via APIs Platforms like Channable (recently merged with Producthero) are experimenting with API extensions that can control AI based on feed data[4]. Invest now in integrations that allow your product data to be "requested" directly by AI agents from other platforms.

Step 3. From CPC to CPI: Cost Per Influence From now on, measure not only clicks, but also influential factors within conversational journeys. What does your product do over three touchpoint interactions via different AIs? CSSs must value engagements that do not directly generate clicks, but *do* influence purchasing decisions.

Step 4. Develop an AI conversation scenario for your top 50 products Simulate 1-on-1 conversations between a buyer and an AI (prompt engineering time!) and use those insights to adapt your data: which questions come back, which specifications are missing? Automatically update your feed based on this.

Step 5. Put human copywriters to work on feed personality. AI loves well-written prompts *and* rich data. Let copywriters optimize product descriptions with answers to questions that AIs are likely to ask.

What does this concretely mean for CSSs now?

Those who act quickly will transform from a data pump station to an AI fuel giant. CSSs will then no longer only push price deals, but will become the contextual underdogs that AIs choose as a reliable source of information. You become food for agents—or filtered out.

And what does this mean for Google?

If Google was once the dominant intermediary, it is now also a player in the field. AI via Google Bard, search results "pre-answering" based on intent, and own Shopping-GPT integrations will reduce CSSs to a feed plugin... unless those CSSs reinvent themselves as a partner *FOR* AI.

For advertisers and brands:

Are you still working with a CSS that only thinks "lower in CPC"? Then you're missing the boat. Ask about their AI roadmap. Demand insight into how they bring your brand to the attention of AIs. Either they tell the story... or they quietly disappear behind the scenes.

Conclusion:

AI no longer leads the consumer to your product. It buys on behalf of that consumer. CSSs that do not adapt will be ignored by AI—and therefore by consumers. Your new customer is an algorithm, your pitch is an API call, and your market share? Depends on how well your CSS can talk.

--- Sources: [1] https://tellix.ai/understanding-the-impact-of-ai-on-consumer-buying-behavior/ [2] https://www.iab.com/news/ai-ranks-among-consumers-most-influential-shopping-sources-according-to-new-iab-study/ [3] https://martech.org/google-reshapes-in-store-shopping-with-ai-powered-comparisons/ [4] https://blog.producthero.com/news/producthero-joins-forces-with-channable/