The B2B Attribution Reality Check: What We Find in 9 out of 10 Audits
Attribution has always been difficult in B2B marketing, but it’s been the changes to the marketing landscape over the last two years that have made it particularly more difficult. SEO has been at the centre of that shift.
What used to be a fairly reliable channel for direct, trackable leads is shifting dramatically closer to attention marketing: a way to build visibility, brand awareness and trust across the buying journey, rather than a tap you can turn on and watch the pipeline fill.
This is not a temporary dip, or a blip in the industry so long as AI is here to stay. It is the result of several major changes to how people search, stacked on top of each other in a short space of time. Which means the expectations most B2B business owners still hold about SEO are out of date.
Why has SEO stopped delivering the leads we used to get?
SEO hasn’t stopped working, we regularly speak to clients about the quality and quantity of leads coming from SEO. It’s simply stopped working in the way it used to two years ago. Search engines are increasingly answering queries directly on the search results page, or through AI tools entirely.
This isn’t something new to SEO, we’ve been dealing with zero click searches for a while now, but it’s had a doubling effect leading to less people clicking through to a website after a search. The click that used to convert into a lead and directly attribute to SEO or organic search now either happens through a direct branded search after further discovery in an AI tool or doesn’t happen at all.
Any SEO agency that truly knew their stuff never sold it as a performance or lead generation channel to B2B businesses. A prospect searched, landed on a page, filled in a form and that form fill was often the proof that the channel worked. That model relied on a search engine results page that no longer exists.
A stable page that meant ranking, clicking and converting. This relationship has uncoupled, some of the SEO used to capture as a click is now being captured as an impression. Meaning, someone sees the brand mentioned, cited or summarised in an AI Overview. They form an opinion on your business then and there, before they’ve even clicked on your site and read the copy you want them to read.
These users will often act on their opinion later on and often through a different channel, most commonly a direct search for the brand name. The influence and impact of SEO is still very real, the attribution is not.
What’s actually changed on Google in the last two years?
Having worked in SEO for the last five years, arguably the most turbulent time in the industry, I’d pin this down to 4 key moment; the rise of LLMs, ChatGPT in particular, Google rolling out Gemini, Google releasing AI Overviews and AI Mode becoming a recent, permanent part of a search result.
Each of these moments have compounded to where we are today, each one has reduced the number of clicks a traditional organic search ranking produces and changed how people search.
AI mode is the clearest signal of intent. For years Google resisted building a chat-style, conversational answer engine directly into Google search. This was largely because it competes with the ten blue links model that its advertising business is built on.
AI mode was that resistance giving way, largely due to the demand and pressure being placed on Google to improve their product and maintain market dominance. Now, it’s a fixed part of the SERP, and it has been designed to keep the user inside Google’s own interface, rather than sending them out to a website.
AI Overviews work the same for a huge split of information and even commercial queries. The answer is summarised, a handful of sources are cited, and a meaningful proportion of searches never scroll past it. Being cited in an AI Overview response is worth something, but its brand exposure often doesn’t lead to a direct conversion.
And at the same time, a growing number of buyers in B2B are starting their research in ChatGPT and Gemini rather than Google. That research happens almost entirely outside of any analytics platform a business owns. None of this means visibility stops mattering, it means visibility and lead generation have started to pull apart as two separate outcomes of the same activity. Just two years ago, they used to be treated as one and the same.
How do we work out what marketing is generating leads?
Attribution can be worked out in two layers, what your tracking can measure directly and what can only be inferred from a pattern of indirect signals. Most B2B businesses only have or build the first layer then treat its gaps as proof a channel isn’t working, when the truth is the channel’s effect was never being measured in the first place.
The direct layer is what most people mean by attribution. A lead lands on the site, fills in a form or books a call, and that action is traced back to the channel and even the specific page it was produced from. This is where Google Analytics 4, a CRM and a properly configured Tag Manager earn their keep, and it’s still very important. It just no longer tells the whole story.
The indirect layer is everything that shapes a buying decision without leaving a clean digital trail. Someone who saw the brand appear in an AI Overview weeks earlier, a colleague who mentioned it, a piece of content read on a previous visit that never converted at the time.
None of that shows up in last-click attribution in GA4, it shows up if you look for it, in the softer signals. A rise in branded search volume, an increase in direct traffic, sales conversations where a prospect already knows the business before the first call.
The practical answer is stop asking SEO for a single number and start asking for a clear pattern across several measures:
- Organic visibility
- Branded search growth
- Assisted conversions in GA4
- What sales are hearing
No one of those proves SEO generated a client on its own, instead they show whether it’s doing its job.
Why is GA4 attribution so unreliable in practice?
GA4 is essentially incapable of accurate attribution across the entire user journey, but in practice it is rarely set up well enough to deliver it. Of the prospects we speak to, roughly 1 in 20 already have GA4 set up correctly but just need help making sure it’s been done correctly. The rest have some form of broken setup that has to be fixed before attribution can be even remotely trusted.
The most common issues fall into a small number of categories:
| Issue | What it looks like | Effect on attribution |
| No CRM integration | Form fills tracked in GA4, but never matched to whether they became paying clients | SEO gets credit for leads, not revenue |
| Consent mode misconfigured | Cookiebot or another CMP blocking tags inconsistently | Data gaps that look like traffic drops |
| Cross-domain tracking missing | A booking tool like Plannr or a separate checkout on another domain | Sessions split in two, journeys look shorter than they are |
| Duplicate or conflicting GTM tags | Old Universal Analytics tags still firing alongside GA4 | Inflated or duplicated conversion counts |
| Default channel grouping untouched | Every AI referral or dark social visit dumped into “Direct” or “Unassigned” | Channels that are actually working look invisible |
The last row matters a lot more than it ever used to, specifically showing you the traffic that has come from users that started their journey in ChatGPT, Gemini and other AI tools frequently land in GA4 as direct or unassigned traffic rather than it being directly labelled.
Not because you can’t see traffic from AI Assistants, but because an average buyer in B2B doesn’t click the link in the AI platform, therefore, their session cannot be directly attributed to ChatGPT or Claude etc.
A business looking only at its channel report can end up concluding SEO and AI visibility are doing nothing, when in reality the tracking simply isn’t built to see the impact they’re actually having.
What most B2B businesses get wrong about attribution
Most businesses treat a symptom found in their reporting as a fact about their marketing. The actual cause is often sitting further back in how their tracking was set up. The pattern shows up the same way across nine out of ten audits we run.
| What it looks like in the reports | What’s usually actually happening |
| “Organic leads have dropped” | Leads are still arriving but landing in an unassigned or direct bucket |
| “SEO doesn’t produce sales-qualified leads” | Leads reach GA4 but are never matched back to CRM outcomes, so quality is invisible |
| “We get traffic but no conversions” | A booking or checkout tool on a different subdomain is breaking the session, not the traffic itself |
| “Marketing can’t prove ROI” | Sales-influencing content is being judged on last-click conversions alone |
The common thread is that attribution gets treated as something GA4 does for you, automatically, rather than something that has to be deliberately configured, continually maintained as tools, consent law and buyer behaviour change.
A GA4 property set up correctly eighteen months ago may be broken today. A new booking tool added to the site goes untracked, a new cookie banner policy knocks out all users or a site migrates without the necessary code needed to track analytics data.
How do you fix a broken attribution setup?
Honestly, unless you have a clued up marketing team, you need to speak to a specialist. The barrier to even getting started with correct attribution in tools like GA4 is getting higher and higher as data protection laws continue to evolve.
Taking it one step at a time, we’d recommend auditing not rebuilding. It’s tempting to get a clean start but before you change anything establish what your platforms are all currently doing, recording and collecting.
In practice, this usually means doing the following:
- Auditing the GA4 property and GTM container together, checking for duplicate tags, broken triggers, and consent tool conflicts, most commonly with Cookiebot.
- Correcting cross-domain tracking where a business uses a separate tool for booking or checkout, such as Plannr, so a journey isn’t split into two disconnected sessions.
- Connecting GA4 to the CRM the business actually runs on, whether that’s HubSpot, Pipedrive or Salesforce, so a lead can be traced through to whether it became a client, not just whether it filled in a form.
- Rebuilding channel groupings so referral traffic from ChatGPT, Gemini and other AI tools is labelled correctly instead of falling into direct or unassigned.
- Adding qualitative tools, such as Microsoft Clarity, to see how visitors actually behave on key pages, which matters more as fewer conversions happen on the first visit.
- For businesses with enough data volume, moving reporting into BigQuery so trends can be tracked over time without being limited by GA4’s own retention and sampling.
I’m assuming you have the right tools set up in the first place here.
None of this should be seen as a one-off part of a tracking project. Consent rules change, tools get sunset or swapped, sites get migrated. Attribution has to be checked periodically in the same way SEO performance does, before it quietly breaks without anyone noticing.
What should B2B business owners realistically expect from SEO now?
Expect SEO to keep driving, direct and trackable leads. It will be a smaller share than what could be attributed two years ago, not because it’s underperforming but because the search journey has shifted dramatically.
If the benchmark for success is still volume of form fills and calls booked like it’s 2024, that benchmark is out of date. Most importantly, chasing it is wasted effort and an evolution is needed.
The honest reframe is that SEO is now doing two jobs at once, but not as effective at either. Generating some leads from a traditional search and building authority, credibility, visibility, brand presence and share of voice.
Both matter commercially for B2B businesses, only one of them is easy to put a number to.
The businesses that adapt fast are the ones that fix their attribution setup first, so they can actually see what happened, rather than assuming a channel has just stopped working. Get your measurement right, and the conversation moves from ‘is SEO still worth it’ to a much more useful one, ‘which parts are earning their place and which ones need to change’.
How to get your attribution checked properly
If you’re not confident your reporting reflects what SEO is actually doing for the business, that’s worth sorting before anything else.
Speak to an SEO specialist at ROAR for a straightforward look at your GA4, GTM and CRM setup, and what it would take to see the full picture.
