Three people wired the events over four years
Events were wired by three different people over four years. Two count the same thing, one fires twice, and the total has quietly been wrong for long enough that nobody wants to raise it.
Services
Measurement & Reporting
The numbers you take into the board meeting hold up when someone asks where they came from. We set up tracking a hard question cannot break, run the paid spend against it, and close every month with one document you can forward to your CEO without rewriting it.
Get in touch
Choosing a reporting system
What should a B2B website report include?
A useful B2B website reporting system connects search visibility, visitor behaviour and qualified enquiries to a dated record of work. Look for clear conversion definitions, comparable periods, source links, sample sizes and an owner for the next action. Keep paid, organic and AI referral traffic identifiable.
Ask to trace one reported enquiry back to its source. Check how the system handles consent, duplicate events and CRM stages, and whether it distinguishes a form submission from a qualified opportunity. A report should state when the sample is too small to support a conclusion.
Numbers That Hold
Why Marketing Struggles to Prove Its Number
The ProblemStandard Agency
The dashboard never answers whether it worked
The dashboard updates in real time and the same question goes unanswered every month: did what we did last month work. Real-time data is not the same as a decision.
You spend a morning translating the agency report
The agency report arrives full of platform screenshots and vocabulary nobody upstairs uses, so the marketing lead spends a morning rewriting it before it can go anywhere.
The Solution
The Legency Approach
The Legency ApproachA conversion means one thing everywhere
Events audited and rebuilt so a conversion means one thing across analytics, ads and the CRM. Where a platform overcounts, we say by how much rather than quoting it and hoping.
Budget decisions are made with the page in view
We manage the campaigns and build the pages the clicks land on, so budget decisions are made with the landing page in front of us rather than the click cost alone.
One document a month you forward as it is
One document a month: what changed, what it did, what happens next, with the number behind every claim. Plain language, no platform jargon, and nothing that needs translating before it goes upward.
The Monthly Engine
Four Numbers You Can Defend Upward
A Lead Followed From First Visit to Closed Deal
Analytics, conversion events, consent and CRM handoff, built so a lead can be followed from first visit to closed opportunity. This is the layer everything else depends on being right.
Paid Spend That Defends Itself
Campaigns run by the team that builds the landing pages, so the spend and the page it lands on answer to the same number. Budget moves toward what converts, search terms get pruned continuously, and creative is tested rather than refreshed on a feeling.
The AI Channel Counted on Its Own
Sessions arriving from AI assistants segmented and reported separately. It is a new channel in most analytics installs, and it is invisible unless somebody deliberately splits it out.
A Report That Travels Upward Unedited
One document covering rankings, citations, conversions and spend, with the method behind each figure. Written to go to a CEO or a board without a rewrite, which is the only test that matters.
Before Any Number Is Trusted
Most Analytics Installs Are Wrong in a Specific Direction
They are wrong in ways that flatter whatever is being measured, which is why nobody catches them. These four account for most of what we find on arrival, and every one of them ends up in a board deck if nobody looks.
One action counted several times
A form that fires its event on submit and again on the thank-you page reports twice the conversions. We have measured a single demo request firing two and three times on a live client install. Every downstream number inherits that error, including the ones in board decks.
Events counted as people
Event count and user count are different questions and most reports quote whichever is larger. If a form fires more than once per person, the honest figure is users, and the two have to be pulled side by side before either is quoted.
Consent silently removing traffic
Under a consent banner, a proportion of visitors are never measured at all. That is correct behaviour and it still means the totals are not the whole picture. It has to be stated rather than quietly absorbed.
AI traffic hidden inside direct
Sessions arriving from an AI assistant often carry no search term and get bucketed into direct or referral by default. A growing channel becomes invisible unless somebody deliberately splits it out, which is exactly when you most want to see it.
The Report
Written for the Meeting After Yours
The test of a marketing report is whether it can be forwarded without being rewritten. Most fail it, which is why marketing leads spend a morning translating before anything goes upward. Ours are built to pass it.
Numbers before adjectives
What changed, by how much, over what window, against what baseline. The narrative comes after the figure and explains it, rather than standing in for it.
A method beside every claim
Source, date range and denominator on each number, so a question about where a figure came from has an answer in the document instead of in somebody's inbox two days later.
The bad month said plainly
A report that only ever contains good news stops being read as information. When something fell, it is in there, with what we think caused it and what changes next.
Case Study
One report format, forwarded upward at Blueflame for eighteen months.
The same monthly document has gone to Blueflame since the engagement began. It covers organic, paid, AI citations and site performance in one place, and it has never needed a rewrite before it went upward.
View Case Study
Start With the Numbers You Have
Find Out What Your Analytics Is Actually Counting
Before anything is optimised, the counting has to be right. A call about how your tracking is set up now, and what it would take to get to numbers that survive a hard question.
- How we audit events for duplicates and separate people from event counts
- How AI-assistant traffic gets split out of direct, where most installs bury it
- One monthly report format you can forward without a rewrite
Common QuestionsFrequently Asked
Frequently Asked
Questions
What shipped, what it changed and what runs next, with the numbers behind each point: organic performance, AI citations, conversions and paid spend. Every figure names its source and its date range, so a question about where a number came from has an answer before it is asked.
Because most of it is wrong in a specific direction. Duplicate events, a form that fires twice, a thank-you page counted alongside the submit, consent blocking a tag on half the traffic. Optimising against numbers like that moves budget toward whatever is being overcounted, and the error ends up in a board deck.
Yes, Google Ads alongside the site. That closes the usual gap where one agency owns the spend and another owns the page the click lands on, and neither is accountable for the result.
Yes, and it is worth separating. Sessions from ChatGPT, Perplexity and similar arrive with no search term and get grouped with direct or referral traffic by default. We split the channel out and report it beside citation counts, so the visibility work and the traffic it produces sit on the same page.
It is a separate service, on its own page. Measurement makes the testing possible, and the testing programme is where the changes get made. Most clients run both, and they are scoped separately so it is clear what each one is doing.
GA4 and Search Console as the base, a search data provider for rankings and competitor data, our own citation panel for AI answers, and whatever CRM you already run. We do not resell a proprietary dashboard, because the numbers should stay checkable after we leave.