Digital marketing · Growth
Analytics and Measurement
We put measurement in place that can see search, the website, and the AI tools your buyers now use, then help you spend against what is actually producing work.
The issue
The problem shows up differently depending on the firm. These are the questions we hear, in the words clients actually use, and the answer we would give in the room.
- Legal
New business mentions AI assistants. Nothing in Analytics matches. What are we not seeing?
- Accounting and finance
We have three dashboards and still cannot tell which spend produces new clients.
- Medical and scientific
A lot of the valuable reading never becomes a click. How do we count that?
- Travel and hospitality
The property system, the brand site and search do not talk. Where do we even start?
We put measurement in place that can see the parts of demand your dashboard cannot. AI assistants never run the tracking script, so they never appear in ordinary analytics. Google's AI Overviews send people who often never appear as a useful source. You can be getting recommended and never see it. We put a reading of the server next to the client-side view you already have, agree what a qualified enquiry means, and then look at which work is actually producing fees.
The point is not a more colourful report. The point is to stop spending against a picture that is missing the readers who now start the buying process. Client-side analytics still sees human visits, form fills and the campaigns you are paying for, and throwing it away would be a different kind of blindness. The two views sit together. Partners get a report that names what each view can see, and what it cannot.
How this looks in your firm
In a law firm, new-business meetings now mention AI assistants, and Search Console does not explain it, because the assistant never ran the tracking script.
In an accountancy, three dashboards still cannot say which spend produces new clients, because crediting the last click was never going to work for a six-month buying cycle.
In a medical or scientific practice, a lot of valuable reading never becomes a click, so the dashboard looks quiet while the reputation work is happening elsewhere.
In travel and hospitality, the property system, the brand site and search often do not talk, so the first job is to name the events that count as a booking enquiry.
The setup changes. The principle does not: if you cannot see it, you cannot build on it. If the logs show assistants reading pages that cannot convert, the next job is the site. If they show nothing being read, the next job is visibility.
A prettier dashboard on the same browser-based analytics will not show AI assistants. If measurement stays in the browser, AI assistants remain invisible, and we will not take that work.
What you get
These are the concrete things the engagement delivers, not only a report and an invoice.
A written measurement plan naming the events that count as progress, including a definition of a qualified enquiry the partners will accept.
A reading of the server for the tools that fetch the site, alongside the analytics you already use.
A monthly report that separates what humans did, what AI tools fetched, and what remains unknown.
Recommendations for spend only after the baseline exists, never from last-click alone.
MeasurementHow we know it is working
The service is working when partners can see three numbers they trust: volume of qualified enquiries, the share that can be attributed at all, and whether the pages you care about are actually being read.
We review monthly.
If the logs show AI tools reading pages that cannot convert, the next job is the site, not more media.
If they show nothing being read, the next job is visibility.
Measurement is the first move at this stage.
Spend comes second.
We will not attribute revenue to a channel on the basis of last-click alone, and we will say so in the report rather than letting a default model do the talking.
Who this is not for
If you want a better-looking dashboard on top of the same reports you already get, this will not help you. The work means looking at your website's own server records rather than the usual analytics, and if the firm is not prepared to do that, nothing changes.
An example of this work
Ranked on page one and invisible to every assistant
Disputes practice, north of England, around 40 partners. Anonymised.
Practice areas loaded on click, the case studies were PDFs, and the team pages were generated in the browser. Google could still read the site. The assistants their clients asked never quoted it.
Questions
Why can we not see AI assistants in Google Analytics?
Because analytics runs in the browser and AI assistants do not. The tools that fetch your pages on behalf of AI assistants take the words that are already there and never execute the tracking script, so they appear in the server logs and nowhere in the dashboard. A person who clicks through from an AI assistant does get counted as a referral. The larger signal, which pages those tools are reading and quoting, exists only in the logs. A firm that reports from the dashboard alone is measuring the small visible end of its own demand.
Do we have to replace our current analytics?
Not necessarily. The dashboard you already have still sees human visits, form fills and the campaigns you are paying for, and throwing it away would be a different kind of blindness. What we add is the server-side view: which tools fetch which pages, whether they get the words, and whether those pages are the ones you think matter. The two views sit together. Partners get a report that names what each view can see, and what it cannot.
How soon will this change what we spend?
The measurement itself can be in place within weeks. Changing spend waits until there is a baseline, because a single month of logs is a story, not a sample. We agree what a qualified enquiry means before we call anything a result. Three months is the minimum because professional services cycles are long, and because we would rather tell you we do not know yet than invent a trend from noise.