Put a number on the deals AI is sending to your competitor
Share of Model tells you that you are losing. It does not tell you what losing costs. Revenue at Risk prices the buying-intent prompts a competitor is winning, through your own funnel, so the conversation moves from a chart to a number your CFO can argue with.
Attribution proves what AI earned. This prices what it costs.
Every AI visibility tool can show you a gap. Almost none will say what the gap is worth, because saying so honestly requires numbers the tool does not have: your contract value, your close rate. We ask for them, and until you provide them we show no estimate at all.
What the estimate refuses to do
Use an industry-average contract value in place of yours.
Call a prompt lost after seeing one answer to it.
Price a prompt you lose half the time as though you lost it entirely.
Present a single confident figure where the honest answer is a range.
Five multiplications, each one shown
The estimate is a chain, so it is drawn as a chain. Anyone who disputes the total can point at the exact step they disagree with instead of rejecting the whole thing.
- Step 1
Searches you lose
Annual search volume behind the buying-intent prompts where AI names a competitor, weighted by how often you actually lose each one.
Measured
- Step 2
AI visits
What that demand would send you if the assistant recommended you instead.
Measured, or benchmark until your traffic can set it
- Step 3
Leads
Your visit-to-lead rate applied to those visits.
Measured or entered
- Step 4
Customers
Your lead-to-close rate applied to those leads.
You entered
- Step 5
Revenue at risk
Your average contract value applied to those customers.
You entered
A range, because the honest answer is a range
Nobody can know how many of those buyers would have picked you if the assistant had named you. Rather than bury that uncertainty, the product prices it: the same chain runs three times, varying only the step we model rather than measure.
Conservative
If few of those buyers would have switched to you anyway.
Most likely
Our central case, and the one to take into a planning conversation.
Optimistic
If the recommendation gap is costing you dearly.
A number whose assumptions are hidden is not defensible
This figure ends up in commercial conversations, so every input that produced it is on the page, labelled with where it came from.
Came from your connected analytics and your tracked prompt runs.
Came from your setup. Change it and the estimate moves with it.
Our starting assumption, used for one step and only until your own data replaces it.
The estimate names the prompts to fix first
A total is only useful if it points somewhere. The lost prompts are ranked by the search demand behind them, each showing how often you lose it, so the first week of work is already decided.
Three numbers switch it on: average contract value, lead-to-close rate, and visit-to-lead rate where we cannot measure it from your traffic.
See plansRevenue at Risk, explained
Revenue at Risk prices the buying-intent prompts where an AI assistant recommends a competitor instead of you. It takes the search demand behind those lost prompts and runs it through your own funnel, using your visit-to-lead rate, your lead-to-close rate and your average contract value, to estimate what the recommendation gap could be costing you in a year.
An estimate, and it is presented as a range rather than a single figure. We cannot know how many of those buyers would have chosen you if AI had recommended you, and that uncertainty is genuine, so the product shows a conservative, a most likely, and an optimistic case instead of hiding it behind one confident number.
Every input is tagged with its source. "Measured" means it came from your connected analytics. "You entered" means it came from your setup, such as contract value and close rate. "Benchmark" means we are using a starting assumption, which happens only for the rate at which search demand becomes a visit, and only until your own traffic can set it.
Three numbers only you have: average contract value, lead-to-close rate, and visit-to-lead rate if we cannot measure it from your traffic. Until those are filled in, Targetlytics deliberately shows no estimate at all. An industry-default contract value would produce a number that collapses the moment anyone asks where it came from.
A prompt is only called after it has been run more than once in the window. The same question can name different vendors on two consecutive runs, so a single answer cannot settle whether you win it. Prompts you lose only some of the time are priced at the share you actually lose, not in full, and prompts run once are left out entirely.
Attribution measures what AI traffic already earned you. Revenue at Risk estimates the other half: what the recommendations you are not winning could be worth. We never add the two together. One is measured, one is modelled, and mixing them would make both less trustworthy.
Pricing questions, head-to-head comparisons, shortlist and "best X for Y" queries. These are the questions someone asks when they are close to choosing a vendor. Awareness-stage questions are excluded, because a lost answer there does not map cleanly onto a lost deal.
Find out what the recommendation gap is costing you
Start with a free audit of where AI recommends you today, and where it recommends someone else.
Get a free visibility audit
