AI Connector
Connect your location data to Claude and ChatGPT so you can ask questions about performance in plain language.
The AI connector lets you interrogate your own measurement data in plain language through Claude or ChatGPT, which makes the questions that were previously too expensive to answer cheap enough to ask routinely.
What it does
Every measurement system has a shadow set of questions nobody asks. Not because the answers would be uninteresting, but because extracting them means an export, a join and half an hour of spreadsheet work, and the value rarely justifies that. The data exists; the answer never gets produced.
“Which locations declined for three consecutive periods and also have unanswered reviews?” is a good example. It is specific, it produces a work list, and almost nobody asks it. A dashboard cannot anticipate it, and building it by hand costs more than the answer is worth on any given Tuesday.
Connecting the data to an assistant collapses that cost to nothing. The question is asked in ordinary language and answered immediately, which changes not how the data is accessed but which questions get asked at all. That is a larger shift than it first appears.
It is deliberately read-only and deliberately not a reporting replacement. Recurring reports remain reports. What this covers is the long tail — the one-off, cross-domain, slightly awkward questions that a fixed dashboard was never going to answer and a spreadsheet made too tedious to pursue.
Everything AI Connector gives you
Natural-language interrogation
Ask about measurements, reviews, profiles and tasks without constructing a filter or building a view.
Cross-domain joins
Questions spanning rank series, review status and profile health answered together rather than in three separate places.
Read-only connection
Query access without write access, so convenience carries no operational risk.
Scoped visibility
Limit what the connection can see per user or per client, which matters in an agency context.
Comparative queries
Compare locations, periods or clients conversationally instead of assembling each comparison by hand.
Claude and ChatGPT
Connect through a standard connector to whichever assistant the team already uses.
Answer portability
Take a result straight into a client note without re-deriving it later.
No training on your data
Data is queried rather than used to train models.
Query log
A record of what was asked, which matters for governance across a client book.
From setup to first result
- 1
Authorise the connector
A one-time authorisation for Claude or ChatGPT. No development work at any point.
- 2
Set the visible scope
Decide which locations or clients the connection can read, before anyone starts asking.
- 3
Begin with a diagnostic
“Which locations declined this period?” produces a work list. “How are we doing?” produces a paragraph.
- 4
Ask comparatively
Comparison across locations or periods yields decisions; single-figure lookups mostly yield trivia.
- 5
Verify anything consequential
Check the underlying number before acting on something material. Fast is not the same as audited.
- 6
Move the answer into the workflow
Into a task list or client update immediately, rather than re-deriving it next week.
- 7
Leave recurring reporting alone
Scheduled reports still handle the recurring case. This is for the questions they cannot anticipate.
What changes with AI Connector
- Treating it as a reporting replacement
- Acting on an unverified figure
- Asking broad questions
- Leaving scope open across a client book
- Using it as the only interface
- The expensive questions are the useful ones
- Cost of asking determines what gets asked
- Cross-domain answers drive decisions
- Non-analysts stop being blocked
- Read-only makes it safe to open up
The expensive questions are the useful ones
Anything requiring a join across domains is exactly what a dashboard cannot anticipate and a spreadsheet makes too costly.
Cost of asking determines what gets asked
When a question takes thirty seconds instead of thirty minutes, people ask far more of them.
Cross-domain answers drive decisions
Rank alone or reviews alone rarely indicate action. The intersection usually does.
Non-analysts stop being blocked
Plain language removes the dependency on whoever normally builds the views.
Read-only makes it safe to open up
Query access can be given broadly precisely because it cannot change anything.
Reports and interrogation are different jobs
One is recurring and formatted; the other is ad-hoc and specific. Conflating them serves neither.
What it measures
- Questions asked per period
- Locations within scope
- Access scope per user
- Verification rate on material figures
- Query types most used
- Time saved against manual extraction
What you get out of it
Ask the awkward questions
Cross-domain, one-off, specific.
Answers in seconds
The cost of asking drops to nothing.
Open to the whole team
No query syntax required.
Safe by construction
Read-only and scoped.
Straight into the workflow
Answers become tasks or client notes.
Governed
A record of what was asked.
What AI Connector produces
Every output is exportable and white-label, with your branding and none of ours.
Direct answers
Responses to plain-language questions.
Comparisons
Locations, periods or clients side by side.
Condition lists
Everything matching a description you give.
Portable findings
Answers taken into notes or tasks.
Scoped views
Only what the connected user may see.
Query record
A log of questions asked.
Who uses AI Connector
Prove movement to clients with location-level evidence rather than a single national figure.
Compare visibility between sites and find the branches losing ground in their own city.
Give each franchisee their own numbers while head office keeps the estate view.
Find out where you actually appear before spending anything on improving it.
Get more out of it
- Ask diagnostic and comparative questions. Broad summary requests produce prose rather than decisions.
- Verify anything you will act on materially. Speed is the benefit; auditing is still your job.
- Scope access deliberately in an agency, since every client’s data being visible to everyone is a governance problem.
- Push answers into tasks immediately rather than re-deriving them in a fortnight.
- Leave recurring reporting in reports. Interrogation and reporting are different jobs.
- Open it to non-analysts. Most of the value comes from more people asking more questions.
Treating it as a reporting replacement
Recurring reporting needs a fixed format and a schedule. Conversational answers provide neither and were never meant to.
Acting on an unverified figure
Convenience is not audit. Anything with a material consequence attached should be checked against the source.
Asking broad questions
“How are we doing?” returns a summary. “Which locations declined three periods running?” returns a work list.
Leaving scope open across a client book
Convenience does not justify every member of staff being able to read every client’s data.
Using it as the only interface
Dashboards remain better for depth. This is for speed on a specific question.
Manually vs with AI Connector
| Doing it manually | With AI Connector |
|---|---|
| Questions need a view built first | Asked directly in plain language |
| Export, join, analyse | Answered in seconds |
| Only analysts can interrogate | Anyone on the team can ask |
| Domains queried separately | Cross-domain answers |
| Awkward questions never asked | Cost of asking removed |
| Write access exposure | Read-only by construction |
- Authorise Claude or ChatGPT
- Define the visible scope per user
- Start with a diagnostic question
- Prefer comparisons to lookups
- Verify material figures at source
- Move answers into tasks or notes
- Keep scheduled reporting in place
AI Connector is 1 of 42 tools you get
Every tool below is on the same plan at the same price. Nothing here is an add-on, an upgrade, or a separate subscription.
Rank & Visibility
Manage Profiles
Posts & Automation
Reviews & Reputation
Reports & White-Label
Others vs Local Rank Checker
How the usual pricing and packaging in this category compares with ours.
“Other tools” describes the common pattern across the category, not any one named product.
AI Connector is included on every plan
One flat price per location covers all 42 tools. There is no higher tier, no add-on, and no per-seat charge.
- All 42 tools on every plan
- No per-seat charges
- No setup fee and no contract
- Cancel or change locations any time
AI Connector questions
What problem does this actually solve?
The questions nobody asks because extracting the answer costs more than the answer is worth. It removes that cost.
Which assistants are supported?
Claude and ChatGPT, through a standard connector requiring no development work.
Is my data used for training?
No. The connection is read-only, scoped to your account, and data is queried rather than used to train models.
Can it change anything?
No. It reads. It cannot publish, edit profiles or alter settings, and that is deliberate.
What kind of question works best?
Diagnostic and comparative. Anything of the form “which locations match this condition” produces something you can act on.
Does it replace reports?
No. Recurring reporting needs a fixed format and schedule; this handles the one-off questions reports cannot anticipate.
Should I trust the numbers?
For orientation, yes. For anything material, verify at source. Fast retrieval is not the same as an audited figure.
Can clients query their own data?
Access is scoped, so you can expose exactly the locations a given person should see and nothing more.
Can it compare periods?
Yes, and comparison questions generally produce more actionable answers than single-value lookups.
How long does setup take?
Minutes. It is an authorisation rather than an integration project.
Can non-technical staff use it?
Yes, which is much of the point. Plain language removes the dependency on whoever normally builds views.
Can I export an answer?
Yes, straight into a client update or task list without re-deriving the figures.
Does it see historical data?
Yes, as far back as your series runs, which is another reason to establish baselines early.
Is there a query limit?
Practical limits come from the assistant rather than from the connector.
Can I ask about one location?
Yes, or a group, a client, or the whole estate depending on the scope you configured.
Does it cover reviews as well as rankings?
Yes, and questions spanning both are among the most useful it answers.
Is there an audit trail?
Yes, a log of what was asked, which matters for governance across a client book.
What if an answer looks wrong?
Check it at source. Report the discrepancy if the underlying figure disagrees, since that is worth knowing.
Should everyone have access?
Read-only access is safe to distribute; the constraint is data scope rather than risk of change.
What is the best first question?
“Which locations declined this period and have unanswered reviews?” It is specific, cross-domain and returns a work list.
Every tool. One price. No add-ons.
$5.33 per location per month gets you AI Connector and the other 41 tools.