The AI Search Attribution Trap (And How Forward-Thinking Marketers Can Actually Prove ROI)

---

---

Your buyers are asking AI before they ask you. According to Forrester research cited by Ahrefs, 94% of business buyers now use AI search when making purchase decisions, and they rate it above vendor websites, product experts and sales reps. Consumers are not far behind: NielsenIQ puts it at 42% of shoppers researching products with AI.

So budgets are moving toward AI search. And the moment money moves, leadership asks the obvious question: what are we getting back?

Interest in that question is climbing fast. Ahrefs’ Keywords Explorer data shows searches for AI attribution and ROI terms up 212% over the last 18 months.

Let’s be real: you will never prove AI search ROI to the penny. AI is one of the hardest channels to track that marketing has ever had.

• Someone can get a complete answer inside ChatGPT and never visit your website.
• When they do click through, the referrer is often stripped, so the visit shows up as direct traffic.
• And when someone acts on an AI recommendation, they rarely tell you that is what happened.

The good news is that leadership does not need a perfect number. They need a reasonable estimate they can trust and compare over time. Between what you can track and what your customers tell you, you have enough data to build one. Here is a practical method, adapted from Ahrefs’ approach, that you can repeat every quarter using data you mostly already have.

The AI search ROI formula

AI search ROI is the value you get from showing up in AI answers, minus what it cost you to get there, divided by that cost.

• Value includes sales, traffic, awareness among people who never click, conversions and better-informed buyers.
• Cost includes content, technical work, tools and team time.

Revenue is the number leadership cares about most, so that is what the steps below focus on:

AI search ROI = (AI revenue − AI cost) ÷ AI cost × 100

Anything above 0% means the work paid for itself. For example, $45,000 in AI revenue on $18,000 in AI cost is a 150% return.

Revenue is not the whole story, though. The awareness and shortlisting value of AI visibility matters too, and it should be reported alongside your ROI. More on that at the end.

Step 1: Add up what AI search costs you

There are two kinds of cost to account for:

• Pure AI costs only exist because of AI search, such as an AI visibility tracking tool or an agency retainer for AI work.
• Shared costs support both AI search and your broader SEO and content work, including your team, SEO tools, ChatGPT or Claude subscriptions, freelancers and agencies.

The easiest way to sort them is one question: would you still pay for this if you stopped AI search work tomorrow?

• If the answer is no, count all of it.
• If the answer is yes, count a share of it.

Do not burn a week trying to account for every cent. A solid ballpark is good enough, as long as you calculate it the same way every quarter.

Your team’s time

People are usually the biggest shared cost. You do not need timesheets. Just ask your team or vendors how they spent their time that quarter. Was about a third of it on AI content and optimization?

Say your three-person content team costs $54,000 a quarter. If they spent about a third of their time on AI search work, count $18,000.

Your tools

Most tools are shared too, and their usage is harder to pin down. The simplest approach is to use the same share as the team that uses them. If your content team spends a third of its time on AI search, count a third of its tool spend.

If a tool has a separately priced AI add-on, count the add-on and ignore the rest. And if your AI workload changes from quarter to quarter, adjust the share to match.

Put together, a quarter’s AI search costs might look like this:

Cost	Type	Quarterly cost	Share counted	AI cost
AI visibility tracking tool	Pure AI	$1,500	All	$1,500
AI-only agency retainer	Pure AI	$3,500	All	$3,500
Content team (3 people)	Shared	$54,000	A third	$18,000
Freelance writer	Shared	$6,000	Half	$3,000
SEO suite and AI subscriptions	Shared	$3,000	A third	$1,000
Total				$27,000

That $27,000 is your AI cost. Hold onto it for Step 3.

Step 2: Estimate the revenue AI search brings in

Much of AI’s influence never shows up in analytics, so there is no single clean revenue number. There are two good ways to estimate it. Use whichever fits your data, or use both.

Either way, be upfront in your report that both methods likely undercount the real value of AI.

Method 1: Tracked revenue

Tracked revenue comes from conversions you can tie directly to visits from AI tools. You can find those in GA4 or tools like Ahrefs Web Analytics.

1. Count the conversions that came from AI referral traffic.
2. Multiply them by your average customer value for that product or service, or use actual deal values if your CRM connects to your analytics.

If 40 AI-referred visitors converted last quarter and your average customer is worth $900, that is $36,000 in tracked revenue.

Tracked revenue only shows part of the picture. AI is usually one touchpoint in a longer buying journey. Plenty of people read an AI answer and come back later by searching your brand name or typing in your URL, and those visits never get credited to AI. The reverse happens too: your traditional content may have done the persuading, even if the final click came from ChatGPT.

Most analytics also cannot follow a visitor all the way to a closed deal. Declined cookie banners, cookieless tracking, device switching and long B2B sales cycles all break the chain. That is why average customer value is the practical choice for most teams. It turns the conversions you can see into a revenue estimate without needing every system to connect.

If your forms pass the traffic source into your CRM, you may be able to see the actual pipeline and deals tied to AI visitors who converted on their first visit. If those same 40 conversions carried $42,000 in deal value, use $42,000 instead of the $36,000 estimate. Just remember it only covers people who converted on that first visit.

Method 2: Self-reported revenue

This one is simple: ask new customers how they found you. A “How did you hear about us?” question at signup, checkout, onboarding or on your lead form does the job.

Because it does not depend on tracking, it catches many of the people your analytics misses, including buyers who saw you in an AI answer and came back later another way. Ahrefs does exactly this with its own free signups and pipes the answers into its analytics and Slack.

To turn survey answers into a revenue number:

1. Add AI tools (ChatGPT, Gemini, Perplexity, Copilot, Google AI Overviews) as answer options, and allow more than one answer.
2. Use actual deal values, since every answer is tied to a real customer.
3. Calculate what share of respondents’ revenue came from AI, then apply that share to all of your new-customer revenue for the quarter.

That last step matters because you will never get a 100% response rate. Start with the customers who did answer. Add up the revenue from those who picked an AI tool, and divide it by the total revenue from everyone who answered. Then apply that percentage to revenue from all new customers, including the ones who skipped the question.

For example, if respondents brought in $100,000 and $10,000 of it came from people who chose an AI tool, that is 10%. If all new customers brought in $400,000 that quarter, your estimated AI revenue is $40,000.

Now you have the value side of the equation.

Step 3: Calculate your ROI

Plug your numbers into the formula: subtract cost from revenue, divide by cost and multiply by 100.

AI search ROI = (AI revenue − AI cost) ÷ AI cost × 100

Using the $27,000 cost from Step 1:

• Tracked revenue of $36,000 gives you (36 − 27) ÷ 27 × 100 = 33% ROI.
• Self-reported revenue of $54,000 gives you (54 − 27) ÷ 27 × 100 = 100% ROI.

A 33% ROI is a good result. Every dollar you put in comes back with an extra 33 cents of profit.

If you used one method, that is your number. If you used both, lead with the higher figure and show the other next to it so leadership sees the full range. Then repeat the process every quarter. The calculation will never be perfectly precise, but doing it the same way each time keeps the trend honest, and the trend is what tells you whether the investment is working.

Step 4: Put an ad-equivalent value on your AI visibility (optional)

There is one more number executives love: what your AI visibility would have cost if you had paid for it. ChatGPT now sells ads by the impression at a reported $25 to $45 per 1,000 impressions, which gives you a benchmark.

AI ad-equivalent value = (mention-adjusted impressions ÷ 1,000) × CPM

Ahrefs ran this calculation on its own brand. It started with about 110 custom, bottom-of-funnel prompts, the questions people ask when they are actively shopping for an AI visibility tool. Using Brand Radar’s estimate of how often each prompt is asked on each AI platform, Ahrefs adjusted for how often it actually appeared in the answers. Over 30 days, it showed up in 46% of ChatGPT responses to those prompts.

That visibility worked out to roughly $77 to $139 in equivalent ad spend. That is small, and it should be. High-intent, bottom-of-funnel prompts are niche and do not get huge volume, whether paid or organic.

Running the same math across roughly 1,800 broader prompts from the Brand Radar index gave a better picture of top- and mid-funnel visibility. Ahrefs appeared in 45% of those ChatGPT answers, worth roughly $7,300 to $13,100 a month in equivalent ad spend.

It is a hypothetical number, not cash in the bank. But it shows leadership what it would cost to buy the same presence, and that makes a strong companion to your ROI figure.

The bottom line

You will not prove AI search ROI to the penny. You can build an estimate leadership trusts and can compare quarter to quarter:

1. Add up what you spend on AI search, counting pure costs in full and shared costs by share.
2. Estimate the revenue it brings in, using tracked conversions, survey answers or both.
3. Run the formula the same way every quarter, so when the number moves, you know performance moved with it.
4. If you have the data, add the ad-equivalent value of your visibility.

Remember that revenue is the smallest, most measurable slice of what AI search delivers. AI answers shape which brands buyers remember and put on their shortlist long before they ever visit a website. Ahrefs covers how to measure that in its follow-up guide to measuring AI search visibility.

If you want help setting up the tracking, the lead-form source questions and the CRM connections that make this math possible, book a strategy call. We’ll show you what you can measure today and what to fix first.

Source: Louise Linehan, Ahrefs, “AI Search ROI Is Messy. Here’s How to Measure It Anyway,” October 2026. Charts, screenshots and the ad-equivalent calculations come from that article.