A “direct” lead may have been influenced by ChatGPT, Google, reviews, LinkedIn and colleagues long before the enquiry appeared in analytics. Modern B2B attribution requires businesses to measure the wider journey, not just the final visible click.
Picture this.
A procurement manager first hears about your company in a ChatGPT answer.
They don’t click anything.
Two days later, they search your company name on their work laptop and read a case study. That evening, they check your reviews on their phone. The following week, a colleague sends them one of your LinkedIn posts.
A few days later, they return to your website from a bookmark and submit an enquiry.
Your analytics records the lead as direct.
Technically, that may be correct.
Commercially, it tells almost none of the story.
ChatGPT influenced the discovery. Google helped with validation. Reviews built confidence. LinkedIn reinforced the decision. A colleague may have added another layer of trust.
Yet the only interaction your analytics can confidently connect to the enquiry is the final website visit.
That is the problem with modern B2B attribution.
Attribution Was Easier When Journeys Were Easier
Traditional web analytics works best when the journey is relatively simple.
Someone searches Google, clicks your result, lands on your website and fills in a form during the same session.
There is a clear chain:
Search → Click → Website → Conversion
UTM parameters, referral information, cookies and analytics platforms can do a reasonable job of recording that journey.
But many B2B buying journeys no longer look like this.
Today, they may look more like:
ChatGPT → Google → website → reviews → LinkedIn → colleague → mobile search → bookmark → direct visit → enquiry
That journey may take place over several days or weeks.
It may involve multiple devices.
It may involve platforms that never send the buyer to your website.
And it may include conversations and recommendations that analytics software cannot observe at all.
By the time the person converts, much of the journey has disappeared from your reporting.
What Does “Direct Traffic” Actually Mean?
Direct traffic is often interpreted as someone typing your website address directly into their browser.
Sometimes that is exactly what happened.
But “direct” can also become the destination for visits where analytics cannot reliably identify the previous source.
For example, someone may:
use a bookmark;
follow a link copied into a private message;
move between devices;
return after an earlier research session;
encounter your brand through an AI answer and search for it later;
copy a URL from one application into a browser;
or visit after hearing about you from another person.
From an analytics perspective, the final session can genuinely be direct.
The mistake is assuming that the entire customer journey was direct.
Those are very different things.
AI Makes the Attribution Gap Bigger
Generative AI adds another complication because discovery can happen without producing a website visit.
A buyer can ask ChatGPT, Gemini, Perplexity or another AI system:
“What are the best companies for [service]?”
Your business might be mentioned.
The buyer reads the answer, remembers your name and closes the application.
There is no click.
No referral.
No session.
No UTM parameter.
Nothing appears in your analytics.
If that person searches your brand on Google two days later, analytics may record Google as the source.
If they subsequently return through a bookmark, the eventual enquiry could be reported as direct.
AI may have started the entire journey while receiving no attribution whatsoever.
This is one reason modern search visibility needs to be considered more broadly than website traffic alone.
Being understood and mentioned by AI systems can influence demand even when it does not immediately produce a measurable referral.
Search Is Increasingly Part of Validation, Not Just Discovery
Google attribution can be misleading for a similar reason.
Imagine someone discovers your company through:
ChatGPT;
a LinkedIn post;
a recommendation from a colleague;
a podcast;
an industry article;
or an online discussion.
They then search your company name.
Google Analytics may show an organic search visit.
But Google did not necessarily create the demand.
It may simply have helped the buyer validate something they already knew about.
That distinction matters.
Branded search is often influenced by marketing activity happening somewhere else.
If brand searches increase after your business gains more visibility across AI answers, social platforms, reviews and third-party websites, those channels may be contributing to organic traffic without receiving credit for it.
B2B Buyers Rarely Make Decisions in One Session
Attribution becomes even harder when purchases require significant consideration.
A buyer may need to:
research the problem;
identify potential suppliers;
compare alternatives;
gather evidence;
check reviews;
involve colleagues;
investigate pricing;
read case studies;
assess risk;
and obtain internal approval.
Those activities are unlikely to happen in one browser session.
And they may not happen on one device.
The procurement manager researching your company at work may later check reviews from their personal phone.
A director may receive your website link through Microsoft Teams.
A technical colleague may investigate your documentation separately.
Someone else may search your brand on Google after your name appears in an internal email.
Your analytics sees fragments.
The organisation experiences one connected buying journey.
This Is Why Last-Click Attribution Can Be Dangerous
Last-click attribution gives credit to the channel immediately preceding a conversion.
It is simple, convenient and easy to report.
It can also encourage bad decisions.
Suppose your enquiries look like this:
Direct: 40%
Organic search: 30%
Paid search: 15%
LinkedIn: 5%
Other: 10%
It would be tempting to conclude that LinkedIn contributes very little.
But what if LinkedIn content repeatedly introduces prospects to your expertise before they later search your company name?
What if AI systems are mentioning your business because of content and third-party references that prospects never click?
What if your reviews repeatedly influence the decision but are almost never the final interaction?
The channels creating and reinforcing demand may appear weaker precisely because they operate earlier in the journey.
Cutting investment based only on last-click conversions could therefore reduce the activity that was generating those “direct” and branded-search leads in the first place.
Analytics Is Not Lying. It Is Answering a Smaller Question
Your analytics platform is not necessarily wrong.
The problem is how the data is interpreted.
Analytics is very good at answering questions such as:
What happened on our website?
Which source immediately preceded this session?
Which pages did this visitor view?
Did this session produce a conversion?
It is less capable of answering:
Where did this person first hear about us?
Which AI answers mentioned us before they visited?
Which private messages influenced them?
Which colleague recommended us?
Which review changed their perception?
Which interactions happened on another device?
Which touchpoint caused them to include us on their shortlist?
Those are commercially important questions.
But they cannot always be answered through website analytics alone.
Better Attribution Starts With Accepting Imperfect Data
The answer is not to find a magical attribution platform that can track every buyer interaction.
That system does not exist.
Some parts of the buying journey are inherently private or disconnected.
A better approach is to combine multiple signals.
Website analytics should remain one of them, but not the only one.
Branded search demand
Are more people searching specifically for your company, products or people?
AI visibility
Does your organisation appear when relevant questions are asked across AI search and answer platforms?
Referral traffic
Which websites and platforms are sending identifiable visitors?
Direct traffic
Is direct traffic growing alongside wider brand activity?
Review activity
Are buyers increasingly checking, leaving or referencing reviews?
Case study engagement
Are prospects visiting proof-heavy pages during consideration?
CRM source information
What did the prospect say influenced their enquiry?
Sales conversations
What other companies, articles, reviews or platforms did the buyer consult?
No single metric provides perfect attribution.
Together, they provide a much better picture.
Ask Leads How They Found You
One of the simplest attribution tools is also one of the most overlooked.
Ask.
A short field on an enquiry form asking “How did you hear about us?” can reveal information analytics cannot.
Do not expect the answers to be perfectly precise.
Someone might write “Google” even though they initially heard about you on LinkedIn and later searched your name.
Another person may write “ChatGPT” even though they also read three case studies before contacting you.
That is fine.
Self-reported attribution is not a replacement for analytics.
It is another data point.
Over time, patterns become useful.
If more prospects begin mentioning AI platforms, referrals, colleagues, reviews or LinkedIn, you have evidence that those channels are influencing demand even when your analytics cannot connect them directly to conversions.
Measure Influence, Not Just Clicks
Modern website reporting needs a broader definition of success.
Clicks still matter.
Organic traffic still matters.
Conversion rates still matter.
But visibility can influence a buying decision before a click takes place.
That makes several previously “soft” signals more commercially significant.
Is your brand being found?
Is it being mentioned?
Is it being understood correctly?
Can buyers verify your claims?
Can AI systems identify what your company does?
Do third-party sources support those claims?
Can prospects find convincing evidence when they investigate you?
Does the website make the next step obvious once they arrive?
These questions connect SEO, GEO, reputation, content, website structure and conversion performance.
They also reflect the way buyers increasingly make decisions.
Your Website Still Matters — Possibly More Than Before
If discovery happens elsewhere, it might seem as though the website becomes less important.
The opposite is often true.
When a prospect finally reaches your website after encountering your business across multiple channels, they are usually trying to validate something.
They want evidence.
They may be asking:
Is this company credible?
Do they really offer what I need?
Have they done this before?
Can I trust them?
Do they understand my industry?
Who are the people behind the business?
What happens if I contact them?
Your website needs to answer those questions clearly.
That means strong service pages, useful case studies, clear company information, expert authorship, structured content, reviews or testimonials where appropriate, accessible design, fast performance and schema markup that helps machines understand the information.
GEO does not remove the need for a good website.
It increases the importance of making the website easy for both people and machines to understand.
Do Not Reallocate Budget Based on One Attribution Report
This is the commercial risk.
A marketing manager looks at the dashboard.
LinkedIn generated few last-click enquiries.
Organic traffic has flattened.
Direct leads increased.
AI referrals look tiny.
The obvious conclusion might be:
“Reduce LinkedIn.”
“Cut content.”
“SEO isn't working.”
“AI isn't sending enough traffic to matter.”
But the attribution data may only be showing the final visible part of a much longer journey.
Before changing strategy, ask a different question:
What could be influencing the leads that our analytics cannot currently explain?
Then investigate.
Look at branded search.
Ask customers how they discovered you.
Review CRM notes.
Monitor AI visibility.
Talk to sales.
Assess review platforms.
Look for increases in direct traffic alongside brand-building activity.
Examine which case studies and high-intent pages prospects consume before converting.
The goal is not to invent attribution where none exists.
It is to avoid pretending that incomplete attribution is complete.
The Dashboard Shows the Arrival. Not Always the Beginning.
Modern B2B buyers move between AI platforms, search engines, social networks, review sites, email, messaging apps, devices and conversations.
Some of those interactions can be measured.
Many cannot.
That does not make analytics useless.
It makes interpretation more important.
When a lead is recorded as “direct”, the correct conclusion is not automatically:
“Marketing had nothing to do with this.”
Sometimes the more accurate conclusion is:
“This is the last part of the journey we can see.”
And as AI-driven discovery grows, understanding that difference will become increasingly important.
Because your analytics may know where the buyer arrived from.
It will not always know where the decision began.