This article was sponsored by Fork. The opinions expressed in this article are those of the sponsor.
Does AI search really replace SEO, or should I budget for both?
How do I attribute conversions to ChatGPT and AI insights?
AI is progressing so fast that it’s hard to keep up with the changes, let alone know how to act.
This is why we surveyed 300 marketing managers from large companies to understand how they respond to AI search and where their organization stands.
The results highlight rapidly growing technology, a majority of leaders optimistic about change, and an infrastructure woefully ill-prepared to support the cacophony of technological change we are experiencing.
Finding 1: SEO is not dead and AI search is additive (not a replacement)
AI search is experiencing massive growth. From almost zero at the start of 2023, it now represents on average 35% of all website traffic.
In two years, AI research has managed to surpass decades of growth achieved by other channels. Naturally, the death of traditional SEO has become a popular prediction. If consumers could get contextually rich answers from a chatbot, why would they bother searching?
As with history, the results are more complex and subtle. Data shows that traditional SEO’s share of web traffic is also increasing. Respondents predict it will gain 8 points in traffic share, from 45% in 2025 to 53% in 2026.
What does this mean?
Think about your own interactions with a chatbot. You exchange ideas, are directed to recommended sites, and then often do your own follow-up research. Last night I asked ChatGPT to help me plan my trip to Iceland. After receiving a firm lecture about the inadequacy of my rain jacket, I headed to Google to find and purchase one. ChatGPT was responsible for two or three website visits, Google another two or three.
AI search adds a new mechanism to consumer discovery. Consumers can refine their ideas or recommendations in chatbots and move on to search with a more refined query. It is not surprising that after the emergence of chatbot, Google reports more complex, multimodal traditional research.
Accept that consumer behavior is (deliberately) hidden across channels
By the way, Google is at the heart of the difficulty in analyzing traditional SEO from AI search. It deliberately blurs the distinction between search, AI insights, and AI mode, and to protect its leading position in search, it has every reason to do so. Search for a coffee maker in AI mode and you will receive a sponsored post. Click on it and you will see a paid search campaign UTM tracking link. Advertisers start showing up in AI search results, without even realizing what’s happening.
ChatGPT (as of today) only offers a single UTM source reference with its traffic, letting marketers know that the traffic is coming from ChatGPT, but nothing more. Marketers see much higher intent traffic, but have no context for the referral. To get even a glimpse of the conversion funnel, marketers resort to reviewing search logs to understand ChatGPT bot behavior on their websites.
You can’t fight these trends. It’s best to build on your existing strategies while determining how to transition to new technologies. Google Gemini ads are simple; If you run search ads, Google has probably already chosen you to show them. Monitor your campaign results and don’t be surprised when some outliers change their behavior. Google will reuse your search ads to find what works in Gemini, you just need to provide the platform with the resources to iterate on the new medium.
ChatGPT is more difficult, but not impossible. Treat ChatGPT referral traffic as highly intentional users who are likely past the initial discovery phase and are well into the funnel. Don’t risk churn by forcing them through unnecessary funnels.
Beware of Conflicts Between SEO and AI Search
The technologies behind SEO and AI are very different. Search ranks content by relevance; AI aggregates multiple signals to distill a response. Often, the same fundamentals serve both technologies: machine readable textstandards-based schemes, clarity and social scores all the signal quality to the algorithms.
But sometimes they pull in opposite directions. In search, you can create two pages to target the exact opposite intent. One page presents an automobile as “luxurious”, while another presents the same car as “affordable”. The search will target each page with a distinct intent. An LLM will group together all the pages related to this product and will be confused by the mixed signals. Are you luxurious or affordable?
To prepare for AI search, be wary of situations where SEO strategies actually harm the new technology.
Finding 2: Marketers are betting huge sums on AI research, but are struggling to measure results
As the share of AI search grows, it’s no surprise that marketers are setting aside budget. What is surprising is just How much. Sixty-five percent of business leaders spend at least 25% of their total marketing budget to AIand 28% spend more than half. This is a significant commitment for a channel where advertising models are still being developed.
Marketers express confidence in measuring results from these budgets, but a closer look shows cracks. Two-thirds say they are very confident and 80% say AI attribution is clearer than traditional SEO.
But in a more detailed follow-up question, 66% also report difficulty with the basics of measurement. Less than one in five people say they have no measurement problems.
Mohammed Faizan of M&C Saatchi Performance suggests the reason is that current metrics simply aren’t up to snuff: “Teams have confidence in what they can see, and what they can see is a clean little edge of the funnel: clear referrals from AI platforms, last-click conversions. It’s not a metric. It’s about noticing the obvious. AI doesn’t show up in your attribution model; it hides in the growth of your branded search, your increase in direct traffic, your “unexplained” conversion peaks.»
This problem is about to get worse. Measuring ChatGPT referral traffic is one thing; paying for it is another. As AI search evolves into a paid channel, marketers will need attribution frameworks which does not yet exist.
If a consumer spends a week in conversations with a chatbot, doing research and comes across retargeting ads, how do you attribute that sale? The measurement gap that exists today will only widen as spending increases.
The good news is that you can take action now.
Embrace all channels; Measure everything you can
Advertising has become a black box. The algorithms run by major advertising platforms consume a huge amount of data to predict and serve the most relevant ads. As digital channels multiply, the number of potential touchpoints increases and metrics become fuzzier. Marketers will increasingly rely on algorithms to model and allocate spend across their channels.
To power these models, you need data. The more, the better. Measure organic traffic, paid search, LLM referrals and any other sources you can leverage. Modeled attribution of the future will need this basis.
Focus on end impact, not platform reporting
The more abstract your measurement model is from actual results, the more likely you are to misattribution. Advertising has progressed from CPM to CPC to CPA, with each change allowing marketers to find better performing media sources. But now, several channels are claiming the same action.
The best way to avoid duplicate attribution requests is not to model share based on what each platform brings in, but to model the actual sales result from the platform’s investment. OpenAI may not deserve 10% of your budget just because it wants 10% of your sales. A incrementality test could reveal that it actually generates 50% of sales. True performance reporting mitigates the impact of advertising on emerging technologies.
Findings 3-5 appear in the full report
Marketers are ready to act quickly with AI: the vast majority believe they will be running closed-loop transactions with chatbots by the end of this year.
And so far, despite the negative press, AI is a clear benefit for marketers: only 3% of respondents see negative marketing performance of AI. Yet when asked about future prospects, worry outweighs optimism.
Download the full report to see how your competitors are actually spending, measuring and planning on AI search this year.
Image credits
Featured Image: Image by branch used with permission.
In-Post Images: images by branch. Used with permission.
