AI identity remediation
The assistants are not looking you up. They are repeating what they already absorbed.
And the sources they absorbed disagree with each other about your hours, your address, your phone number and what you actually do. A model meeting three versions of one business has to pick, and the safest thing it can do is name somebody it is more certain about.
This is not a ranking problem, and that is why it does not feel like one
Nothing about it shows up where you would look. Your website is fine. Your reviews are fine. Your phone still rings. The loss is a customer who asked an assistant for a business like yours, heard three names that were not yours, and called one of them. No impression was recorded and no report exists, because from your side nothing happened at all.
What the assistant needed was certainty about you, and it did not have any.
A listing with an old suite number. A directory that still has the phone line you moved off two years ago. A profile somebody at a former agency created and never handed over. Each one is small and none of them is wrong enough to notice. Together they are a business that cannot be described confidently, and confidence is the whole selection criterion.
What the work actually is
Mostly account recovery and administration, done carefully and in the right order. It is unglamorous and it is the part that changes the answer.
We take the disagreements in the order that matters
The analysis produced a list of sources that contradict each other about your business. They are not equally worth fixing. Some feed the answers directly, some are read occasionally, and some have not been looked at by anything in years. We start where a correction changes an answer, so nobody spends a fortnight on a directory that was never being read.
We get into the accounts
Almost none of this is fixed by writing anything. It is fixed by claiming a listing somebody set up in 2019 with an email address nobody still has, by proving to a platform that you are who you say you are, and by finding out that a category can only be changed on the phone. Each source has its own claim process and its own wait.
We make every source say the same thing
One name, one address, one phone number, one set of hours, one description of what you do, written once and repeated exactly. Not a better sentence per platform. When the sources describing a business disagree, the model resolves the ambiguity by naming somebody else, so identical beats improved.
We go back and confirm each one actually took
This is the step that separates the work from the submission. Platforms accept a change, say thank you, and apply nothing, with no notice to anybody. So we return weeks later to every source we touched and check what it says now, and we chase the ones that quietly reverted.
We re-ask the questions and date the answer
The same customer questions the analysis ran, run again in clean sessions, so you can see what changed against what it said before. A single run is reported as a single run, with its date, because assistant answers are personalised and a snapshot presented as a ranking is a lie with a decimal point in it.
AI identity remediation
$1,500 to $2,500
Scoped per client, and the $500 analysis is credited in full. If you bought the analysis, this runs $1,000 to $2,000 from where you stand.
Two things move it inside the band. How many sources are actually broken, which runs from three to a couple of dozen and has very little to do with how big your business is. And whether you hand over account access or sit at the keyboard with us, because several of these need your own verified login and that turns a task into a shared appointment.
Both are knowable after the analysis and neither is knowable before it, which is why the number is a band and not a guess with a decimal point.
Then there is keeping it that way
Sources revert on their own. A platform re-imports from an aggregator holding your old address. A listing you never knew existed resurfaces after a merger. Somebody at a directory decides your category was filed wrong. None of this generates a notification to you.
The quarterly watch re-asks the same customer questions and re-reads the same sources every three months, and tells you what moved.
$750, billed once every three months. Not monthly, and not $250 a month collected in threes. Quarterly because these answers move seasonally rather than weekly, and because eleven reports a year saying nothing changed is how a useful subscription gets cancelled by somebody who still needed it.
It follows the remediation rather than replacing it. Watching a set of sources that still contradict each other tells you only that they still do.
What this cannot do
It cannot put you in the answer. Nobody controls that. We remove the reasons you are being passed over and we measure what changes, which is a different promise from a place in a list.
It cannot get you into the third-party pages the models lean on most. Much of what an assistant repeats comes from other people writing about you: roundups, reviews, forum threads, local press. Moving those is slow, it looks more like public relations than marketing, and we do not sell it.
It cannot tell you how long a correction takes to show up in an answer. We have not measured it and we are not going to quote somebody else's number for it. What we do instead is date every change and re-ask the same questions later.
It cannot be done entirely without you. Some of these listings will only move for the person holding the verified login. Part of this work is an hour of your time or a set of access grants, and we scope which before quoting.
When not to buy this
If you have not had the analysis. Remediation without it is us guessing which sources matter to you, and guessing is the expensive way to do this. Start with the $500 analysis, which comes off this in full.
If the analysis found two things wrong. It happens, and when it does we say so and fix them for a fraction of this rather than scoping an engagement around a short list. A band exists for the businesses that need the band.
If your customers do not find you through search or maps at all. A trade running entirely on referral and repeat work has a different leak. Talk to us about that one instead.
Questions
Why is it a range instead of a price?
Two things move it, and neither is knowable before the analysis. How many sources are actually wrong, which is anywhere from three to a couple of dozen and is not related to how big your business is. And whether you hand over account access or sit at the keyboard with us, because a source that needs your verified login needs you present, and that is a scheduling problem more than a technical one.
Can you guarantee the assistants will name me afterwards?
No, and nobody who does is telling you the truth. What we can do is remove the reasons you are being passed over and measure what changes. That is a different promise from a place in a list, and we would rather draw the line here than after you have paid.
How is this different from a monitoring subscription?
A monitor watches and charts. What it hands you is a number that moved, with no way to tell you which of your listings caused it and no way to touch any of them. This is the work of touching them: the claims, the phone calls, the account recovery, the re-checks. We run monitoring too, and it is the smaller half of this by a long way.
I already paid for the analysis. Does that count toward this?
In full. The $500 comes off, so the band runs $1,000 to $2,000 from where you are standing. That was the arrangement when you bought the analysis and it does not expire.
What is the quarterly watch actually for?
Sources revert. A platform re-imports data from an aggregator, an old listing you never knew about resurfaces, somebody at a directory decides your category was wrong. The watch re-asks the same questions and re-reads the same sources every three months, so a reversal is caught within a quarter rather than being found next year by accident.
Why quarterly and not monthly?
Because AI answers move seasonally rather than weekly, and because eleven reports a year saying nothing changed is how a subscription gets cancelled by somebody who still needed it. $750 is billed once every three months, not monthly, and not $250 a month collected in threes.
Find out what is wrong, then we go and fix it.
The analysis names the sources and the order to take them in, and every dollar of it comes off this work. If you have already had one, from us or from anybody, send it over and we will scope from what it found.
Remediation $1,500 to $2,500, scoped per client. Quarterly watch $750 every three months. The $500 analysis is credited in full.