Not every business wants to manage AI visibility itself.
Some business owners will be happy to test a few prompts manually. Others may decide to use software once the number of prompts and platforms becomes too large to manage comfortably.
But there is a third group:
businesses that understand AI visibility may matter, but have no interest in running the monitoring themselves.
They do not want another dashboard.
They do not want to decide which prompts should be tracked.
They do not want to interpret changing results across ChatGPT, Google AI Mode, Gemini, Perplexity and other systems.
They simply want someone else to handle it.
For those businesses, the right solution may not be AI visibility software at all.
It may be an agency, consultant or specialist service.
But choosing someone to manage AI visibility raises a different set of questions.
You are not simply buying access to software.
You are buying judgement, interpretation and action around the data.
That distinction matters.
What are you actually outsourcing?
At its simplest, AI visibility monitoring involves checking whether a business appears in AI-generated answers for relevant prompts.
But a useful service should normally involve more than running searches.
Someone managing AI visibility for you may need to:
- understand your business;
- identify important customer journeys;
- choose relevant prompts;
- decide which AI platforms matter;
- monitor your brand and competitors;
- track mentions, recommendations and citations;
- interpret changes;
- investigate unexpected results;
- identify potential visibility gaps;
- report what the findings actually mean;
- recommend sensible next steps.
Software can automate parts of that process.
It cannot automatically decide which parts of the process matter most to your business.
That is where the value of the person or agency should begin.
1. How do they choose the prompts?
This is probably one of the most important questions to ask.
A provider may be able to monitor hundreds or thousands of prompts.
That does not mean those prompts are useful.
If the wrong questions are being monitored, the resulting data may look impressive without telling you much about whether potential customers are likely to encounter your business.
Ask:
- How do you decide which prompts to monitor?
- Do you involve us in the process?
- Do you look at our customer journey?
- Do you consider our most important services or products?
- Do you distinguish informational queries from buyer-ready searches?
- How often is the prompt set reviewed?
A good answer should go beyond:
We use AI to generate a list of prompts.
AI-generated suggestions can be useful.
But they should normally be treated as ideas to evaluate rather than a complete monitoring strategy.
The people running the campaign should understand why each important group of prompts is being tracked.
2. Are the prompts connected to the customer journey?
Not all visibility has equal commercial value.
A business may appear frequently when people ask broad informational questions and rarely when someone asks which provider they should actually choose.
Those situations are very different.
A useful monitoring programme should ideally look at different stages of the customer journey.
For example:
Early research
What is the problem? What options exist?
Consideration
What approaches or providers should I compare?
Evaluation
Which option is suitable for my particular needs?
Buyer-ready questions
Which company should I contact or choose?
Ask the provider how its monitoring reflects those differences.
If everything is reduced to one overall visibility score, valuable context may be lost.
3. Which AI platforms do they monitor?
There is no single AI search system.
Depending on the service, a provider may monitor platforms such as:
- ChatGPT;
- Google AI Mode;
- Google AI Overviews;
- Gemini;
- Perplexity;
- Microsoft Copilot;
- other emerging AI answer systems.
More platforms are not automatically better.
What matters is whether the platforms being monitored are relevant to your customers and whether the provider can explain why they have been selected.
You should also understand whether every platform is included in the agreed service or whether additional coverage costs extra.
4. How frequently are searches monitored?
AI-generated results are dynamic.
The same prompt may return different recommendations, citations or ordering at different times.
That means monitoring should normally involve repeated observations rather than treating one search as a permanent result.
Ask:
- How often are prompts run?
- Are searches repeated over time?
- How are fluctuations treated?
- Does one unusual result trigger a report, or is it checked again first?
- How long does the provider normally wait before describing something as a meaningful trend?
The important point is not necessarily that every prompt should be checked constantly.
It is that the provider should recognise that one AI answer does not necessarily establish a stable pattern.
5. Do they distinguish mentions, recommendations and citations?
These terms are often mixed together.
A business can be mentioned without being recommended.
Its website can be cited without the company itself being presented as a provider.
And a business can be recommended even when its own website is not cited.
If the provider reports that your company has “40% AI visibility”, ask what that actually means.
Does it represent:
- brand mentions;
- recommendations;
- citations;
- position within the answer;
- some combination of several factors?
A metric can be useful.
But only if everyone understands what it measures.
6. Can you see the underlying results?
You should not necessarily expect to inspect hundreds of raw AI responses every month.
That is partly why you are outsourcing the work.
But important conclusions should be supported by evidence.
If the provider says:
Your visibility has improved significantly.
you should be able to understand what changed.
If it says:
Competitor A is repeatedly appearing where you are absent.
you should be able to see examples.
Ask whether reports allow you to inspect:
- individual prompts;
- AI responses;
- recommendations;
- citations;
- competitor appearances;
- changes over time.
You are paying for interpretation, but the interpretation should remain connected to the underlying observations.
7. Which competitors are being monitored?
Competitor visibility can provide important context.
Knowing that your business appears in 20% of monitored answers tells you relatively little on its own.
If your closest competitors appear in 5%, that may be encouraging.
If they appear in 70%, the picture is different.
Ask:
- How are competitors chosen?
- Can we nominate competitors ourselves?
- Does the software discover competitors automatically?
- Are unexpected brands recorded?
- Can competitor visibility be examined by individual prompt or customer-journey stage?
AI systems may sometimes surface companies you did not originally consider direct competitors.
That in itself can be useful information.
8. What software do they use?
If you are outsourcing the work, you do not necessarily need to become an expert in the provider’s software.
But you are entitled to understand the basic monitoring setup.
Ask:
- Which platform or platforms do you use?
- Why did you choose them?
- What does the software automate?
- What still requires human interpretation?
- Can important results be independently checked?
The name of the tool matters less than whether the agency understands its limitations.
A provider should not treat software output as unquestionable truth simply because it appears in a dashboard.
9. Do they validate important findings?
This is particularly important when results are surprising.
Suppose the software reports that your company has suddenly disappeared from several important prompts.
Or perhaps your visibility appears to have doubled in a week.
Before drawing major conclusions, it may be sensible to inspect some of those results manually or repeat the relevant checks.
Ask:
What do you do when the monitoring shows something unexpected?
A good answer might involve reviewing the underlying responses, checking whether the pattern continues and considering possible explanations before making recommendations.
Automation provides scale.
Human checking helps provide context.
10. How do they report the results?
A monthly report should do more than contain charts.
Useful reporting should answer questions such as:
- Where are we appearing?
- Where are we absent?
- Which competitors appear instead?
- Which parts of the customer journey are strongest or weakest?
- Has anything meaningfully changed?
- Are we being described accurately?
- Which findings deserve investigation?
- What should we consider doing next?
Be cautious of reports that are dominated by proprietary scores without explanation.
A graph can look sophisticated while telling you very little about what is happening commercially.
11. How do they interpret fluctuations?
AI visibility can change naturally.
That creates a risk of overreacting to normal variation.
A provider should be able to distinguish between:
a single unusual result
and
a pattern worth investigating.
Ask how they deal with volatility.
Do they compare multiple runs?
Do they look for recurring changes?
Do they explain uncertainty?
A useful service should help reduce confusion, not create alarm every time a graph moves.
12. What happens when they identify a visibility gap?
This may be the most important part of the service.
Suppose a provider finds that three competitors regularly appear for an important buyer-ready prompt and your business does not.
What happens next?
There are many possible explanations.
Perhaps:
- your website does not clearly address the subject;
- competitors provide more specific information;
- third-party sources frequently mention competitors;
- AI systems misunderstand what your company does;
- important evidence is scattered across several pages;
- your site lacks supporting detail;
- the pattern is simply unstable and needs more observation.
The correct response is not automatically:
Write another article.
The provider should investigate before recommending action.
Ask:
What is your process for moving from a visibility observation to a recommendation?
That answer may tell you more about the quality of the service than the software being used.
13. What do they actually mean by “optimisation”?
Terms such as AEO, GEO and AI search optimisation are now widely used.
The terminology matters less than the claims being made.
A provider may reasonably help with things such as:
- clearer website content;
- better coverage of customer questions;
- correcting inaccurate information;
- strengthening evidence around services;
- making important facts easier to understand;
- identifying missing topics;
- improving how information is structured.
What should make you cautious is certainty.
No agency can control exactly what ChatGPT, Google AI Mode or another AI system will recommend in every future response.
So claims such as:
“We guarantee first position in ChatGPT.”
or
“We can guarantee your business will be recommended.”
should raise serious questions.
The goal should be evidence-led improvement, not promises that cannot reasonably be controlled.
14. How transparent are they about limitations?
A credible provider should be comfortable saying:
- AI results fluctuate;
- different systems may produce different answers;
- visibility does not automatically equal traffic or sales;
- a citation is not the same as a recommendation;
- monitoring cannot prove causation;
- changes made today may not produce an identifiable result tomorrow;
- some findings may remain uncertain.
That does not weaken the service.
It makes the interpretation more trustworthy.
15. How do they measure success?
Ask this before the work begins.
Possible measures might include:
- mention rate;
- recommendation visibility;
- citation visibility;
- competitor share of voice;
- visibility across important prompt groups;
- accurate brand description;
- increased presence in buyer-ready searches;
- reduced gaps against important competitors.
But the provider should also understand the commercial context.
A business could improve its overall visibility score while becoming more visible only for low-value informational queries.
That may be less useful than improving visibility for a small group of prompts closely connected to potential customers choosing a provider.
A good measurement approach should therefore consider where the visibility occurs, not merely how much visibility exists.
16. How is the service priced?
Different providers may charge in very different ways.
You may encounter:
- fixed monthly fees;
- project-based fees;
- pricing based on the number of prompts;
- pricing based on the number of platforms;
- consultancy retainers;
- monitoring-only packages;
- broader packages including content or strategy work.
Ask what is actually included.
For example:
- How many prompts?
- How many AI platforms?
- How many competitors?
- How frequently is monitoring performed?
- Is analysis included?
- Are meetings included?
- Are content recommendations included?
- Is implementation included?
- Are there additional charges when the prompt set expands?
A cheap monitoring package may be poor value if it consists mainly of forwarding an automated dashboard.
A more expensive service may be worthwhile if it includes meaningful analysis and investigation.
17. Who actually does the work?
This can matter, particularly with larger agencies.
You may speak initially to a senior specialist and then discover that most of the ongoing work is carried out elsewhere.
Ask:
- Who selects the prompts?
- Who reviews the results?
- Who writes the reports?
- Who investigates unusual findings?
- Who will we speak to when we have questions?
There is nothing wrong with different team members doing different parts of the work.
But you should understand how quality and interpretation are controlled.
18. Can they explain the process in plain English?
AI visibility is already full of new terminology.
A provider should make the subject easier to understand rather than more confusing.
If you ask:
Why did our visibility fall this month?
you should receive an explanation you can follow.
If every answer relies on proprietary terminology, unexplained scores or vague references to “AI optimisation”, it may be difficult to assess whether the service is actually useful.
Complex technology does not require deliberately complex explanations.
19. Are they willing to show what they actually do?
Before committing, ask for an example of:
- a report;
- a dashboard;
- an anonymised client workflow;
- the types of prompts monitored;
- how recommendations are presented;
- how a visibility gap is investigated.
You are not asking for confidential client information.
You are trying to understand the service you are buying.
A provider should be able to explain its process more clearly than simply saying:
We improve your AI visibility.
20. Don’t outsource understanding completely
There is a final point worth making.
Hiring someone to manage AI visibility does not mean the business itself should understand nothing about the process.
You do not need to know how every tool works.
But you should have a basic understanding of:
- which customer questions matter;
- what is being monitored;
- what the main metrics mean;
- which competitors are appearing;
- what conclusions are being drawn.
That makes it much easier to judge whether the service is delivering anything useful.
The agency provides specialist expertise.
The business still provides essential knowledge about its customers, priorities and commercial objectives.
The strongest arrangement is likely to involve both.
A practical checklist before hiring someone
Before appointing an AI visibility agency or consultant, I would want answers to questions such as:
How will you learn about our business and customers?
How will you choose the prompts?
How are prompts connected to our customer journey?
Which AI platforms will you monitor?
How frequently will you test them?
How do you handle changing AI results?
How do you distinguish mentions, recommendations and citations?
Which competitors will be monitored?
Can we see the evidence behind important findings?
Which software do you use and why?
Do you manually check important or unusual results?
How will the findings be reported?
How do you decide when a visibility gap requires action?
What happens after a gap is identified?
How do you measure success?
What exactly is included in the fee?
Who will actually manage our work?
What claims are you prepared to make—and what will you not guarantee?
Those answers should tell you far more than a homepage promising to “dominate AI search.”
Red flags worth thinking about
I would be cautious if a provider:
- guarantees recommendations from AI systems;
- cannot explain how prompts are selected;
- reports only a proprietary visibility score;
- cannot show the underlying evidence;
- treats every mention as equally valuable;
- never discusses fluctuations or uncertainty;
- cannot explain the difference between a citation and a recommendation;
- automatically recommends creating content for every visibility gap;
- cannot explain what happens after the software produces its report.
None of these automatically proves that a service is poor.
But they are sensible areas to question before committing.
A good service should add more than a dashboard
AI visibility monitoring software can automate a substantial amount of repetitive work.
That is valuable.
But if you are hiring someone else to manage the process, the real value should come from what happens around the software.
The provider should help determine:
what is worth monitoring;
what the results actually mean;
which changes matter;
which gaps deserve investigation;
and what sensible action, if any, should follow.
If all you receive is a monthly automated report containing information you do not understand, you may simply be paying someone else to own the software subscription.
A useful agency or consultant should provide something different:
judgement.
Outsourcing can make sense—but know what you are buying
For some businesses, learning to run AI visibility monitoring internally will make sense.
For others, self-service software will provide an efficient way to automate the work.
And some businesses will reasonably decide that they have neither the time nor the interest to do any of it themselves.
There is nothing wrong with that.
But if you outsource the work, don’t evaluate the provider solely on the software they use or the number of prompts they promise to monitor.
Look at the quality of the process around it.
The most useful question may not be:
“Which AI visibility tool do you use?”
It may be:
“What will you do with the information once the tool has found it?”
That is where the real value of an outsourced AI visibility service should begin.