Use case · Attribution and reporting
Prove what AI search is actually doing to your pipeline.
AI assistants strip the referrer, so the demos they send land in Direct. We pair referrer forensics with what buyers tell you at intake, then report the corrected number next to the raw one, every month.
Attribution recovery · one funnel · 8,307 contacts
73×The gap between what the CRM logged and what buyers said. 11 contacts recorded as AI-sourced. 809 buyers named an AI assistant at intake.
Source · RTR forensic attribution audit · single client funnel
The problem you're living with
The number you can defend is a fraction of the number that is real.
Six of the last fifteen prospects raised this unprompted, and it is the reason good channels get cut. Budget follows whatever reports cleanly.
"Everything shows up as Direct."
Assistants send buyers without a referrer, so your analytics files them under Direct or Unassigned. The channel doing the work gets no credit, and the channel with clean tracking gets the budget.
"Reps hear it. The CRM never does."
Sales gets told "I found you through ChatGPT" on calls, and none of it reaches a field anyone reports on. So the story is anecdote versus dashboard, and the dashboard always wins.
"I cannot build the internal case."
You believe it, but the budget owner is a step above you. Without a defensible number and a stated method, you are asking a CFO to fund a feeling.
Why it is still broken
The obvious fixes do not make the number defensible.
The instinct is to instrument harder. A new UTM convention, a self-reported field, a different attribution model, a better dashboard. Every one of them still depends on a click that AI referrals never send.
Fix 01 · Tighten the UTMs
"We need better tagging."
Tags need a link with parameters on it. Answer engines send people who read a citation and then type your name, so the visit arrives as direct or branded and the tag never existed.
Fix 02 · Add a self-reported field
"Just ask them where they heard about us."
Self-reported gets you volume, not truth. We have seen 809 self-reported AI leads against 11 the CRM could corroborate. Both numbers are real, and neither one is defensible alone.
Fix 03 · Change the model
"Switch to data-driven attribution."
Any model divides credit across touchpoints it can see. If the AI touchpoint never enters the data, no model will invent it.
Fix 04 · Buy better reporting
"The dashboard is the problem."
Reporting renders the data you have. The problem is not presentation, it is that the first touch happened inside a chat window that sends no referrer.
The diagnosis
Two witnesses. Neither one can see the whole thing.
Your CRM is witness one: it sees clicks precisely and is structurally blind to AI, because the referrer is gone before the visit starts. Your intake form is witness two: it sees intent and channel but relies on human recall. Used alone, each one lies in a predictable direction. Used together, they resolve, and the rule for resolving them is fixed in advance so nobody is choosing the flattering answer after the fact. Whoever could actually see it wins: the machine takes paid clicks and branded direct, the human takes demand creation.
In one funnel of 8,307 contacts, the CRM logged 11 AI-sourced leads while 809 buyers named an assistant themselves. We report both numbers, always.
What we actually ship
Everything needed to make the number defensible.
No black box. You get the method, the raw data, and the corrected view side by side.
CRM and analytics instrumented
Fields, sources, and events set up so a demo carries its origin all the way to the opportunity record.
- CRM field and pipeline mapping
- GA4 and server-side checks
- UTM discipline and cleanup
- Lifecycle stage alignment
Moves: Data that survives the handoff
Intake capture that people answer
One well-worded question on the form, tested for completion, because a blank answer is worth nothing.
- "How did you hear about us?" wording
- Field placement and length testing
- Free-text normalisation
- Sales-call capture as backup
Moves: Completion rate
Two-Witness reconciliation
Every contact classified into one of five cases by a rule written before the data arrives.
- Confirmed, recovered, single signal
- Conflict resolution by rule
- Dark bucket kept honest
- Audit trail per contact
Moves: A number that holds up
Board-ready monthly
One page your CEO can read in three minutes, with raw next to corrected and the method stated.
- Raw and corrected side by side
- Channel view with dark shown
- Pipeline and revenue, not sessions
- Method note attached
Moves: Budget you can defend
Graded against a number set upfront
The forecast is built on your funnel maths before kickoff, then actuals are compared to it in writing.
- 3-scenario forecast
- Monthly actuals vs. forecast
- Misses named, not buried
- Re-forecast quarterly
Moves: Accountability
You own it when we leave
Documented, in your stack, reproducible without us. Attribution is not a retainer hostage.
- Written method doc
- Dashboards in your tools
- Team walkthrough
- Reproducible queries
Moves: Independence
How the 90 days run
Wired in week one, because everything else is graded on it.
Same engine as every other use case. What changes is the order of operations and what we grade first.
Week 1 · Wire it
CRM fields, intake question, and analytics instrumented before any content work starts, so there is a clean baseline to grade against.
Weeks 2 to 12 · Reconcile
Every contact classified weekly under the fixed rule, with the dark bucket tracked as its own line rather than allocated to a channel that looks good.
Always on · Report
Monthly board page with raw next to corrected, actuals against forecast, and a plain statement of what we are changing next.
The scoreboard
What we grade ourselves on.
Set before kickoff on your funnel data, graded monthly in writing. If the model cannot clear a 4× return, we tell you before you spend a dollar.
Counted in your CRM, not in analytics. Split by source so the channel stands on its own.
Graded monthly vs. forecast
For self-serve and sales-assist motions, tracked separately from demos so neither hides the other.
Graded monthly vs. forecast
Whether the demos are workable. Protects against a volume win that sales quietly ignores.
Reported with the raw count
Pipeline traced to a source under the Two-Witness Rule, with raw always shown next to corrected.
Reported to your board
Results · From the monthly reports
Named clients, published numbers.
Attribution is what makes every other number believable. These programs were graded on it.
AI search became a demand channel in one quarter
Organic sessions, AI referrals, category presence and attributed demand all moved together one quarter after launch.
Four months in, running ahead of forecast
Nearly half the demo pipeline was filed as Direct. The Two-Witness Rule recovered it, and the page program kept feeding it.
AI search and organic became the growth engine
A page set that did not exist twelve months ago now carries the growth and is the most-cited name in the category’s AI answers.
Fit · Before you fill anything in
Who this works for, and who it doesn't.
We say no often, and early. Attribution work only pays off if someone will act on the number. Reading this saves us both a call.
Great fit
- B2B SaaS, Series A to Series D$2-3M to $50M+ ARR, subscription or usage-based, sales-led or self-serve plus sales-assist.
- A category with a real competitor setVertical or category-defining: healthtech, fintech infrastructure, compliance, HR tech, legal tech, specialty lending.
- Marketing leadership in placeA CMO, VP Marketing, or Head of Growth who owns the number, usually with one under-resourced content or SEO person.
- Someone senior is asking where pipeline comes fromA CFO, CEO, or board is questioning the channel, and you need a number you can defend line by line.
- You can ship changesModern CMS and a willingness to make technical and on-page edits.
- Pipeline is the measureBudget $5,000+ for strategy, $7,500-$15,000+ for the full engine, and demos are how we're both graded.
Also a fit: a team whose AI-sourced leads are obvious in the CRM but invisible in the dashboard.
Not a fit
- Pre-seed or pre-PMFNo audience, no category clarity, and no runway for a compounding channel.
- B2C apps, DTC ecommerce, or local servicesDifferent playbook, different channels. Not our specialty and not our pricing model.
- Enterprise consultingThose cycles are relationship-driven, not search-driven.
- You want content volume without strategyWrong expectation. We decline these even when the budget is there.
- You can't make on-page or technical changesThe methodology doesn't work without them.
- Self-reported data is good enoughIf a form field is already treated as the source of truth, the Two-Witness Rule will feel like overhead.
- Not willing to let us driveIf every decision needs a committee and a month, the program stalls. You approve the strategy, we run it.
- Under $5K a monthNot enough investment to make the channel compound. You'd be paying for activity, not an outcome.
If that's you, we'll say so on the first call and point you somewhere better. No sequence, no follow-up.
In their words
The people who signed off on it.
"Justin and his team have been phenomenal from Day 0, and the consistency in quality and the measurable results they have delivered have been a true transformation to our business. We are extremely happy and we have already recommended them to our friends."
"The partnership with Rock The Rankings contributed to clear improvements in organic performance across Canada and Australia, particularly for MoonPay's high-value transactional keywords and pages."
"Rock The Rankings delivered great work for us. We came in with a massive index bloat issue, and Justin and team quickly figured out what needed to be done and helped us implement his suggestions on the site."
"Rock The Rankings have really gone above and beyond, in terms of exceeding our expectations in both their communications and their strategies around marketing. They're very specific at what they do. They've been able to help us get to positions 1 and 2 for our most valuable keywords."
"They simply get SEO: which levers to pull and when. During my search for the right partner I already had a pretty clear idea about what was needed, and they shared a very similar vision."
"The workflow was seamless between our teams. Overall, their process was very close to ours, so it was a great fit from the get-go. Most importantly, we learned a ton from their approach."
"Justin and his team helped us strengthen our SEO strategy, enabling us to adopt a more in-depth and strategic approach, and to better anticipate and organize our content operations."
"We set out to get valuable backlinks and they have just been a fantastic partner in making that happen. It's been an amazing experience working with Rock The Rankings, and I highly recommend that you try them out."
Straight answers
Asked on every call.
If yours isn't here, bring it to the call. Same unpolished answer either way.
HubSpot and Salesforce most often, Pipedrive and Close regularly. The method does not depend on the tool: two witnesses, five cases, one rule. If your stack can hold a field and export a list, it works.
Usually a couple of hours from whoever owns the CRM, plus a form change. No new tooling and no tracking script to install. If server-side work is genuinely needed we will scope it separately rather than smuggle it in.
Not on its own, which is exactly why it is one of two witnesses rather than the answer. People misremember channels but rarely invent an assistant they never used. Paired with referrer data and a fixed resolution rule, it turns an unmeasurable channel into a defensible range.
Contacts where neither witness names a source. Most reporting quietly redistributes those into whatever channel looks strongest. We report them at real size, because the first time a CFO catches an inflated attribution number, every other number you present loses its value.
If nobody in your organisation is asking where pipeline came from, this is overhead. It also will not help if your sales process does not record outcomes: we can recover the source of a lead, not invent the record of what happened to it.
Next step
Your GEO plan, built on your numbers.
Give us your domain and your pipeline goal. We map the prompts your buyers are asking, build your GEO marketing plan against them, and forecast the pipeline the next 90 days can realistically produce.
What we walk through