Sample AI Visibility Report · fictional client, real format ← Back to Aigentic
Sample · 2026
AI Visibility Report · 90-Day Plan
Reynolds & Co
How a buyer discovers Reynolds & Co through AI today, and a 90-day plan to change it.
Prepared for
Reynolds & Co
Prepared by
Aigentic
Date
August 2026
Aigentic
rq@aigentic.co.uk
aigentic.co.uk
Illustrative sample. Reynolds & Co and Meridian Structures are fictional, used to show the exact format and depth of a real report without exposing a client's numbers. Yours is built entirely from live testing on your own brand.

Summary

Important note AI assistants are unpredictable and change all the time. Their answers vary between attempts, and the websites they rely on are updated continuously. This report is a picture at one point in time, true to the best of our knowledge on this date. It is not a guarantee of future results. The outcomes described depend on the recommendations being put into practice. Where they are, we expect movement in the direction shown, but the exact result, timing and ranking cannot be promised. Full terms and privacy note are at the end.

This section is the whole story in three short parts: how we did the analysis, what we found, and what to do about it. If you read only this, you will know what to do next. The rest of the report is the detail behind it, to read at your own pace.

1. How we did the analysis

We tested how Reynolds & Co shows up when a buyer, rather than an existing customer, asks an AI assistant about placing a steel structures job.

In plain terms: the ten buyer questions for Reynolds & Co (40 runs), plus one direct question about Meridian as a benchmark (4 runs), all live, across four AI assistants.

2. What we found

When buyers ask AI who should build their steel frame, Reynolds & Co does not come up. When they ask about it by name, AI describes the business accurately. So the problem is being found, not being believed. That is good news, because being found is faster and cheaper to fix than being believed.

Shows up for buyers
14 / 80
almost never
Recognised by AI
Weak
today
Trusted-source citations
Two
both listings
Owns the category
Open
no owner
Found, not believed

Of the 14 points Reynolds scored, 8 came from the single question where the buyer already typed the company name. Strip that one out and the score is 6 out of 72. AI can describe the business, but nobody else is telling it that Reynolds is one of the firms worth naming, so it only surfaces for people who already know the name.

The category is open

Your closest competitor, Meridian Structures, holds no category answer either. On the cost question and the constrained-site question, no assistant named a single contractor, on any platform. All four gave generic advice and moved on. No brand owns "a steel contractor who can work a tight city-centre site" in AI answers. That space is there to be taken.

3. What to do about it: the 90-day plan

Three steps over about twelve weeks. The first is quick and fully in your control; the later steps build on it.

StepWhat happensWhen
1. Make it recognisablePut Reynolds & Co into the places AI reads to know a firm exists: a Wikidata entry, company code on your website, the construction databases, a clear services page.Weeks 0-3
2. Answer and get quotedWrite the buyer pages that answer those buyer questions, and get your completed projects referenced by the sources AI trusts, starting with the trade titles.Weeks 2-8
3. Re-measureRun the exact same test again and show the before and after, in time for the November check-in.Weeks 8-12

What we need from you: one named owner for the week 0-3 tasks, a 20-minute call to start, and confirmation of a few facts (project values, accreditations, the head-count figure) so nothing published is wrong.

This is the end of the summary. Everything that follows is the detail behind it. If you only need the plan, you can stop here.

The Full Analysis

Everything from here is the detail behind the summary, to read at your own pace. It is organised around the three layers of our method, plus the scoreboard. The three layers are the things we build: Entity, Authority and Content. The scoreboard measures whether the firm actually shows up, and it is what we re-measure to prove progress.

PartThe question it answersWhere in this report
The scoreboardDoes the firm actually show up in AI answers?Shown first, next
Layer 1: EntityDoes AI recognise the firm as a real, specific thing?Layer 1
Layer 2: AuthorityDo the websites AI trusts mention it?Layer 2
Layer 3: ContentAre the answers people search for there for AI to quote?Layer 3

We show the scoreboard first, because it is where you stand today. Then the three building layers explain it and set out what to fix. The plan then moves those three layers, and we re-measure the scoreboard to prove the change.

A few words we use

TermWhat it means
VisibilityWhether your firm shows up when people ask an AI a question.
EntityWhether AI recognises your firm as a real, specific thing it can identify.
Authority / citationWhether the websites AI trusts and quotes actually mention your firm.
ContentThe pages, articles and video that answer people's questions, so AI can quote them.
Category ownershipWhether any brand "owns" the answer to a category question, or whether it is still open.
Schema (FAQ markup)Invisible code on a web page that labels questions and answers so AI can read and quote them.
Discovery questionA question where the person does not yet know your firm, and describes a need or a constraint.
Direct questionA question where the person already names your firm.

The Scoreboard: Where You Stand Today

What this is: the measurement of whether the firm actually shows up in AI answers. It is the baseline we re-measure at the end of the plan to prove the change.

This is the heart of the analysis: exactly what the four AI assistants said when we asked the ten buyer questions. Most of the score below comes from the one question where the buyer already names the firm. In the questions where they describe a need or a constraint instead, Reynolds & Co almost never appears.

Why these ten questions

We did not pick them at random. They follow the path a real buyer takes, from a first requirement to naming a firm, and each step is a chance to be found.

If the firm is going to be found, it has to appear somewhere along that path. Today it only appears at the very end, once the buyer already knows to ask for it by name.

The scorecard (each assistant, out of 20)

Question, asked in English ChatGPTClaudeGeminiPerplexity
1. Steel frame for a six-storey office, who can design and install it 0000
2. Best steel structure contractors UK 2026 1010
3. Alternatives to the large steel fabricators for civil engineering work 0000
4. Which contractors can work a constrained city-centre site 0000
5. What a structural steel package should cost on a mid-rise build 0000
6. Steel erection for schools and public-sector projects in the North 0000
7. Contractors with BCSA membership and CE marking for structural work 1000
8. Which UK contractors do design-and-build rather than supply only 0000
9. Specialist steel structure firms in Yorkshire 1011
10. Tell me about Reynolds & Co 2222
Total out of 20 5243

How to read it: 0 means the firm did not appear, 1 means it was mentioned in passing, 2 means it was named and described. Ten questions across four assistants gives a total of 14 out of 80. Full scoring is in the appendix.

What the pattern means

The problem is being found, not being described

Reynolds & Co is absent from every requirement, category, constraint and cost question, and only appears in passing on three of them, always at the bottom of a list. All four assistants described the business accurately once it was named. The buyers most likely to want a specialist are sent straight to the large fabricators that Reynolds competes against.

Two listings are doing all the work

Across the 36 runs where the buyer did not type the name, Reynolds appeared six times, every one of them in passing. Trace the sources and they come back to two places: a trade directory entry and a regional builders' list. That is the entire visible footprint. It also proves the mechanism, because two citations were enough to put Reynolds in front of three different assistants. The fix is more of them, not a different strategy.

The one small opening was the regional question

The only place Reynolds surfaced on three assistants out of four was question 9, on specialist firms in Yorkshire, and it came from the regional listing rather than from anything on your own site. Local intent is already half working, which makes it the cheapest thing to push. Questions 4 and 5, the constrained site and the cost, returned no contractor names at all from any assistant, so those are uncontested and worth taking next.

Compared with Meridian

Both firms are described accurately when named, and the descriptions are fair. The difference is what stands behind the name. When an assistant digs into Meridian, it finds two trade titles, a directory profile, a Wikipedia entry and a set of named project case studies. For Reynolds & Co it finds your own website, one trade directory line and a regional listing. On Perplexity, which shows its sources, the Meridian answer carried four citations and the Reynolds answer carried one.

The gap is independent coverage, not the quality of the work, where the two are level. Every publication in Meridian's stack is open to Reynolds, which is what makes this fixable inside a quarter.

The rest of the report

You have read the diagnosis. The remaining six sections, about twelve pages, are the fix: what to build, in what order, who owns each task, and how we prove it moved.

Layer 1Entity: the eight places AI checks to confirm a firm is real, which two Reynolds holds, and the order to fix the other six.
Layer 2Authority: the named trade titles and directories the assistants quoted in this test, with first-mention and full-effect timings for each.
Layer 3Content: the buyer questions ranked win first, core, build and later, the pages that answer them, and how to keep them fresh.
The planThe 90 days in detail: every task with an owner and an effort estimate, plus what we expect to move, split into what is in our control and what depends on other people.
Next stepsWhat happens now, who owns what, and the facts we need you to confirm so nothing gets published that is not right.
AppendixThe exact questions, how every number was scored, why we chose each directory and publisher, and the terms.
The full report is delivered with the AI Search Audit

Appendix: About this sample

Reynolds & Co and Meridian Structures are invented. The scores, the sources and the situation are illustrative, chosen to show a shape we see constantly: a firm that is described accurately once you name it, and absent everywhere a buyer would actually find it.

What is real is the format and the method. Every report follows this structure: how the test was run, what it found, the 90-day plan, then the three layers that produce the change, the plan in detail, and the appendix that shows every number's working. It is built entirely from live testing on your own brand, and the re-test at week 12 uses the identical questions so the before and after is a like-for-like comparison.

Want one for your business?

Start with the free Snapshot: your brand against a competitor you name, back within 24 hours. If the gap is worth closing, the AI Search Audit produces the full report on your business, not a fictional one. rq@aigentic.co.uk