Digital marketers have long been trying to solve a rather obvious task: where did this customer come from?
Once upon a time, it seemed like an obvious question. One clicked an ad on Google, visited a website, filled out a form, and finally became a customer. Someone else found a business on social media or via organic search results. Analytics tools helped track many of such actions and understand how it worked.
However, this image gets harder and harder to keep.
Search is changing, and people no longer use a predictable sequence of actions starting with a search query and ending up on a website. They can ask an AI service for some recommendations, read the answer without any clicks, look for a company separately, view their ratings, watch a video, and only after all of that visit the website.
When this person becomes a lead or a customer, it will be quite hard to determine the source of the influence.
Here is a new task for digital marketers: how do you measure the efficiency of marketing when the customer journey is unpredictable?
Table of Contents
Attribution Was Never as Simple as It Looked
Digital marketing attribution can be described as a system to identify what marketing actions led to a conversion.
Think of a customer who comes across a business via an Instagram ad. After a couple of days, the customer does some searches on Google. Two blogs later, the customer leaves the website, returns after a week via an organic search link, and submits an inquiry.
Whose credit is it?
Based on last-click attribution, it may go to the organic search since it was the last point of contact before the customer converted. The first-click attribution model will give the credit to Instagram. Multi-touch will attribute the credit to multiple points.
All the above models could be correct since they just look at the customer journey from different perspectives.
Now imagine adding AI to the customer journey.
This will especially be critical for a digital marketing agency since more clients are interested in knowing the origin of their leads as well as how their customers find them.
A Customer Can Be Influenced Without Clicking
One of the main differences that AI-driven search introduces is that not all interactions lead to visiting the website.
Imagine someone who wants to purchase accounting software.
This person asks an AI search engine which platforms are appropriate for use by a small business. Several brands are recommended. But the person doesn’t click on any link, as he or she only researches the product.
Two days later, this person searches for one of those brands on Google.
This person reads various reviews, visits the website, compares prices, and finally asks for a demo.
If you look at the report only, then the conversion looks like a result of branded organic search.
But maybe the brand became known to the person due to an AI search?
It cannot be said that AI made the person buy the product, but it might have influenced his or her decision in some way.
The Number of Discovery Channels Is Growing
The use of artificial intelligence search is not a replacement for all of the current marketing channels.
Instead, it is yet another addition to a growing list of choices that may present a potential customer with a brand name.
This may happen via Google Search, YouTube, Instagram, LinkedIn, review sites, industry publications, forums, emails, ad channels, or an AI-powered channel.
There may be some overlap between the channels used.
The buyer does not care about attribution models.
They simply accumulate information until they feel ready to make a decision.
But marketers do have to deal with the numbers.
That is why the traditional question, “Which channel converted the buyer?” does not always give the complete picture.
“What was the buyer’s journey before conversion?” would be a much more relevant one.
Zero-Click Searches Make the Picture Even Less Clear
A number of opinions have been expressed about zero-click searches, while AI search adds another aspect to the problem.
A person can receive a relevant answer without going to the site.
For the website owner, such a situation looks quite contradictory: a person found something helpful on his site but still has no visits registered by Google Analytics.
However, it doesn’t mean that the site did not benefit from being recognized.
Brand awareness can be formed without a direct visit.
A person can remember a brand name, share it with a colleague, search later, or even purchase from the site later.
AI Search Has Added Another Layer to the Customer Journey
Classic searches followed a predictable flow of events. One had a query, typed it into a search engine, scanned the results, clicked on a website, and then continued researching on that page.
AI-based search will turn it into a different story.
For example, one could ask a search assistant a specific question and get an answer compiled from several sources. The customer might find out what needs to be found out based on the answer received and not have to click on any of the websites just yet.
At some point, he or she may perform a search for one of the companies from the answer received previously.
It would be another isolated search in analytics reports.
And this is where attribution becomes complicated.
The first AI-powered search could contribute to the awareness creation process, but the visit to the website could be attributed to a different traffic channel altogether.
AI Citations Are Measurable, but They Don’t Tell the Whole Story
There is definitely one part of AI-related traffic that can be measured more precisely.
In some cases, AI searches return links to other websites, and people click on them. As OpenAI claims, for instance, publishers have the ability to get referral traffic from ChatGPT with utm_source=chatgpt.com used for referrals.
It gives the marketer something to track.
If a user visits the site via ChatGPT, it is possible to detect the source via analytics tools.
However, the problem remains even with this type of traffic.
For instance, a potential customer may be exposed to the brand via AI but never clicks through. He/she simply remembers the brand name and returns via some other channels.
The referral is measurable. But the initial impact is not necessarily so.
This is why it makes sense for the marketer not to confuse AI with regular referral traffic.
How Last-Click Attribution May Be False
Why Last-Click Attribution Can Be Misleading
If a client fills out a form as a result of clicking on an organic search listing, last-click attribution can attribute the success of the conversion to the last click.
But this click did not necessarily lead the client to be interested in the company.
Let’s imagine that the potential client interacts with the business in three ways:
- He hears about it in an AI response
- He watches a video on YouTube
- And then he searches the business name and fills out a form
Last-click attribution gives all the credit for the conversion to the last step.
This comes in handy when asking, “What led the person to the site before filling out the form?”
Multi-Touch Attribution Is Useful, But It’s Not Always Perfect
Multi-touch attribution attempts to recognize multiple touches instead of just one – either first or last touch.
This seems more logical and usually is.
However, multi-touch models still rely on the accuracy of the data gathered. If there is a touch that occurs outside the trackable journey, the model will not be able to give credit for it.
For instance, if someone finds out about your brand by chatting with AI, but then comes from direct traffic, it might be hard for the analytics system to connect these two touches.
Then there is the problem of assigning appropriate value to the touchpoints.
Was the first touch more valuable than the fifth? Did the customer come in for the last time just to confirm their previous decision? Was it a certain article that converted them?
Analytics can help to guess, but not always to be sure.
Data-Driven Attribution Is Advanced, Not Magic
Analytics tools today have gone past simple first-touch or last-click models.
For instance, Google Analytics has data-driven attribution, where the available customer journey data is used to make a prediction on the value of each touchpoint in achieving important events.
This can be more helpful than applying the same percentage to all channels.
However, marketers should know what these tools are really doing.
They are predicting values based on available data. They are not psychic.
If one step in the journey is not there, there is no magic way of including it.
This is particularly true when searching in the AI-driven search space, where discovery touchpoints may not involve website visits.
Direct Traffic Can Hide Part of the Story
Direct traffic is one more field that might lead to misunderstanding.
A direct visit means that a visitor entered the URL manually or clicked on a bookmark. However, direct visits happen when analytics does not know where the visitor came from.
Let’s suppose that a person finds out about a brand through the AI system, remembers its name, and types the website URL in their browser after several days.
The eventual visit might be marked as direct.
It would be wrong to analyze that particular visit and draw conclusions that AI had nothing to do with the customer’s journey.
On the other hand, it would be equally wrong to think that any growth in direct traffic is due to the AI search.
That is why attribution needs context.
AI Traffic Has Become Easier to Detect
There is some positive news regarding this issue: measurement is not going anywhere.
Google Analytics launched a new channel called AI Assistant, which can detect traffic from known AI assistants such as ChatGPT, Gemini, and Claude.
This will help marketers know if AI assistants are really sending visitors to their websites.
However, there is a caveat here: Google’s AI Assistant channel does not contain traffic from Google AI Overviews or AI Mode.
The difference between the two is subtle.
“AI traffic” is not just one thing.
Different AI experiences are different, and analytics platforms can classify them differently.
Thus, marketers should know what their reporting tool measures.
Search Console Adds Another Piece of the Puzzle
Search Console also keeps up with changes in search from Google.
Google explains that clicks on links inside AI Overviews and AI Mode will show up in Search Console statistics, and AI Mode can bring new queries into the picture if people follow up with additional questions.
This gives insight into the way websites show up in Google’s AI-enabled search results.
Once again, though, visibility is different from conversion. It might be seen, clicked on, visited, and eventually help to create a lead.
Or it might be seen, remembered, and found at a later date through some other channel.
These are separate events and should not all be treated as the same thing.
So What Should Marketers Measure?
It’s not to abandon attribution models. They do have their uses.
The right way to proceed is to measure several data types together.
1. Monitor AI Referral Traffic
Where your analytics system can measure AI referrals, track them separately. It will help you understand if the AI platforms actually drive traffic to your site.
2. Track Branded Searches
An increase in the number of branded searches can be considered a good indicator of brand awareness growth.
While it shouldn’t be immediately attributed to AI, it can be helpful to look at it along with other marketing efforts.
3. Monitor Assisted Conversions
Do not pay attention only to the source of the last interaction.
Check the customer journey data to understand if there were any other interactions before the conversion.
4. Link Analytics and CRM
Website analytics can tell you how the lead got to your site. The CRM system can tell you what happened next.
Linking the two can be more useful than just monitoring website conversions.
The lead that got to you through organic search may have had several other interactions before that, which won’t show up in the final conversion report.
5. Measure AI Visibility Separately
If AI search is a key source of traffic in your industry, you should monitor the visibility of your company, products, services, and expertise in such answers generated by AI.
This is separate from conventional traffic measurement and needs to be considered in addition to conventional conversion data.
Final Thoughts
Attribution in digital marketing is increasingly difficult to measure since the ways people find and assess companies are changing.
AI search provides yet another source of discovery; however, some of the engagements that happen through AI are not part of the standard online journey. A user may get to know a brand in an AI answer, remember it, learn more about it later on, and finally convert via another channel entirely.
It means that the last click is only one side of the story.
While marketers should keep using attribution, they need to go beyond it. AI referral traffic, branded search traffic, organic visibility, CRM data, assisted conversion, customer feedback, and revenue could offer a much clearer view of things.
Frequently Asked Questions
1. Why does AI search make it harder to attribute digital marketing?
It is possible for AI search to affect a customer even before they land on a website. In this case, they might have seen the company referenced by an AI response and decided to come back via Google, direct traffic, and other means. Therefore, it might be hard to attribute the initial influence to the conversion.
2. Can marketers attribute traffic from AI platforms?
Yes. It is possible to attribute some of the AI referral traffic. In particular, Google Analytics has an AI Assistant traffic channel to track traffic from known AI assistants such as ChatGPT, Gemini, and Claude. Different AI search experiences might be captured differently.
3. Is last click attribution valuable anymore?
Yes. Last click attribution is valuable in terms of identifying the last interaction that led to the conversion. However, it should not necessarily be taken as proof that this interaction was the only and the most influential one.
4. How can businesses assess AI search performance?
Businesses can assess the performance of their AI search efforts by tracking identifiable AI referrals, branded searches, organic search performance, AI visibility, assisted conversions, leads, and revenue. Connecting website analytics and CRM data could also help gauge lead quality.
5. Would the rise of AI search render the relevance of SEO obsolete?
Not likely. This is because SEO continues to be relevant as search engines still use websites as one of their sources of information. In addition to SEO, the importance of AI search lies in its role as another medium through which information can be researched by people.
Vandana is a Content Writer and Digital Marketing Executive with 1+ years of experience in SEO content writing, website content, and digital marketing. She specialises in creating engaging, SEO-friendly content, keyword research, and content planning to help brands improve their online presence and audience reach.
