A woman working at her desk late in the evening
A woman working at her desk late in the evening
A woman working at her desk late in the evening

Vendors

Get selected, not just mentioned, when AI builds the shortlist.

Your buyers bring their own list to AI tools like ChatGPT, Claude, Gemini or Perplexity and ask them to cut it down. I measure whether you make the cut, why, and what to fix first.

For CMOs, marketing directors, founders and revenue leaders at B2B vendors whose buyers now compare suppliers in AI tools.

£1,995 · seven days · 60-minute walkthrough · money-back guarantee

TRACK RECORD

Ran B2B marketing at Samsung and Rackspace

Built the Spiceworks lead-gen platform that sold intent-based programmes to the likes of Dell, HPE, Microsoft and Oracle

10 B2B categories measured since September 2026

Page last updated: November 2026

Buyer research

PUBLISHED DATA

51%

of B2B software buyers start their research with an AI chatbot more often than with Google.

Share of B2B software buyers surveyed

Start with an AI chatbot more often than with Google: 51%, up from 29% in G2’s 2025 report

The rest of the buyers surveyed

Source: G2, The Answer Economy, April 2026 · 1,076 B2B software buyers

The shift

Being in the answer isn’t the same as being on the shortlist.

Buyers still start with their own list. Then they take it to AI tools to compare it and cut it down: list in, shortlist out. The answer gets built one question at a time, and the buyer’s wording changes who’s in it. In September I tracked the questions UK banks ask when they’re choosing a partner to modernise their payments systems. One of the world’s largest IT services firms was named in 67% of answers to “Who can help meet the ISO 20022 deadline?” But when you ask “Best partner for ISO 20022 migration”, it disappears: the same engines didn’t name it once (my analysis: The Inclusion Problem). Same firm, same need - different words, different answer.


When I show vendor marketers where they stand, the first reaction is usually surprise. They think they’re doing better than they are, because that’s what their tools and their agencies have been showing them. What those reports rarely separate is the broad category question from the specific ones buyers ask as they get closer to a decision - the comparisons, the alternatives, “who suits a business like ours”. That’s where the shortlist gets made, and where names drop out.


You can’t see any of it from your own analytics. There’s no Search Console for AI answers. So measure it before you pay anyone to fix it.

TRY IT BEFORE YOU BUY ANYTHING

Pick three questions a buyer in your category asks near the end of their search. One that compares you with your biggest competitor, one that describes the problem you solve without naming the category, and one that asks who suits a business like your best customer. Ask each in ChatGPT and Gemini, and note whether you’re named at all. If you turn up for the first and go missing on the other two, that’s the gap the Scan measures.

Three questions in two engines on one day will show you a gap. They won’t tell you how big it is, why it’s there, who takes your place or what to fix first. That takes around thirty questions across five platforms, run repeatedly against the competitors you name, which is what the Scan does.

The model

Four layers between a mention and a shortlist

THE SHIFT

AI mediates the buying decision

Being recommended, not just mentioned, decides the shortlist

THE MEASUREMENT

Selection Index and sector sets

Who gets named, how, and which sources the answers draw on

OWNERS AND MANAGEMENT TEAMS

Measure

Source Snapshot

Diagnose

Audience Asset Review

Build

90-day phases

Hold

Board advisory

New revenue · Exit value · Commercial capability

VENDORS

Diagnose

AI Shortlist Scan

Engineer

Selection Engineering

Optimise

Quarterly re-run

Getting selected

The Selection Layer is where AI shortlists get built. I measure it in four layers:

  • Inclusion: are you in the answer at all?

  • Positioning: when you’re there, how are you described?

  • Preference: are you recommended, or merely listed, and on what grounds?

  • Selection: when the buyer checks what AI said, on your site and with their peers, does it hold up?


The first three follow the AI answer, where the buyer compares and cuts the list. The fourth follows the buyer coming out of it, checking what it said before they commit. It needs your pages and your claims, which is why only a Scan scores it.

Underneath all four sit the sources the answers are built from: your pages, your competitors’ pages, review and analyst sites, communities and trade media. Most of the fixing happens there.

The evidence base is the measurement: the Selection Index, one sector each quarter, plus the sector sets I keep between editions. If you own or run a media, events or information business, the owners page is here.

How it works

Three stages, starting with a seven-day measurement

You start with the AI Shortlist Scan, and only go further if the numbers say you should. The next step is quoted from what the Scan finds.

Select a stage to see how it works.

01

Start here

Diagnose

AI Shortlist Scan

Seven days · £1,995 · money-back guarantee

You leave with

Your position on all four layers against four competitors you name, and the five actions ranked by impact

Showing below ↓

02

Engineer

Selection Engineering

Six weeks · fixed scope, quoted from £15,000 at your walkthrough

You leave with

The top actions done by your team with me directing, and a re-run showing what moved

See the detail ↓

03

Optimise

A re-run each quarter

Discussed at your walkthrough

You leave with

Whether your position is holding as the answers change

See the detail ↓

01 · Diagnose

AI Shortlist Scan

Where you stand when your buyers ask AI to compare, and why.

I put around thirty buyer questions to five AI platforms - ChatGPT, Google’s AI Overviews, Gemini, Perplexity and Claude - and run them repeatedly over a window, because one run tells you very little. The questions come from your buyers: the category questions, the briefs a buyer in your target segment would write, comparisons with the four competitors you name, the problem you solve described without the category, and the checks buyers make near the end. Every answer is scored on all four layers.


You also get the Shortlist Rate: how often you’re in the recommended set on the decision-stage questions, reported beside how often you’re named at all. That gap is usually where the surprise is. It starts with a short call to agree the competitors and buyer segments, and ends with a 60-minute walkthrough of the five actions that would move your position most, ranked by impact.

£1,995 · seven days · 60-minute walkthrough · money-back guarantee

You leave with

A measured position, the reasons behind it, and the five things to do first. If, by the end of the walkthrough, you don’t think the Scan has shown you where you stand and what to fix first, say so and I’ll refund the fee in full.

02 · Engineer

Selection Engineering

The top actions from your Scan, done properly, with a re-run to show what moved.

Six weeks, fixed scope, quoted from the Scan’s action list at your walkthrough, from £15,000. The work goes where the Scan points: pages that answer the specific questions your buyers ask, the comparisons you’re missing, the third-party sources the answers lean on, and the claims about you that don’t match what you sell. I direct it. Your team and your existing agency do the execution, with specialist content creators from me where you need them.

Six weeks · fixed scope, quoted from £15,000 · your team executes, I direct · one re-run and movement report

You leave with

The top actions in place and a measured answer on whether they worked. The movement is measured. It isn’t guaranteed, and I won’t pretend otherwise.

03 · Optimise

A re-run each quarter

Whether your position is holding.

AI answers move when models update and sources shift. A vendor shortlisted in March can be gone by June, and the only way to know is to measure again, the same way. After a Scan or Selection Engineering, I can re-run your questions against your baseline each quarter. We’ll talk about it at your walkthrough if the numbers suggest it’s worth doing.

Discussed at your walkthrough

You leave with

Whether your position is holding as the answers change.

£1,995 · seven days · money-back guarantee

Outcomes

What the three stages work towards

The measurement and the deliverables are guaranteed. The outcome isn’t, and nobody who’s honest about how these answers work can promise it.

Getting selected

Being on the shortlist when your buyers ask AI to compare.

THE PROOF

I’ve sat on your side of this. I ran B2B marketing at Samsung and Rackspace, and at Spiceworks I built the lead-generation platform that sold intent-based programmes to the likes of Dell, HPE, Microsoft and Oracle. Spiceworks also taught me what AI now does at speed. IT buyers didn’t trust what vendors said on their own sites, so they asked each other. AI runs that same check before it writes the answer, which is why your own website is only part of the picture.

Sample

What a Scan reports

A Scan comes back as one report for one brand against four competitors you name, read through with you in a 60-minute walkthrough.

The sample here is my own business. When a question names Decoding Growth, AI includes it in 158 of 179 answers. When the question describes what I do without the name, it turns up in 26 of 2,328, about one answer in 90, and Google’s AI Overviews named it twice in 768. That’s a Recognition problem, the first of the conditions for Inclusion, and it’s the first thing I’m fixing. The other rows show what a live Scan fills in from your answers.

FOR CONTEXT

Every Scan reports the Shortlist Rate beside the Inclusion rate: how often you’re recommended on the decision-stage questions, against how often you’re named at all. It can be far lower.

AI SHORTLIST SCAN · SAMPLE

Decoding Growth · B2B AI selection

ChatGPT, Perplexity, Google AI Overviews · 8 October to 6 November 2026

INCLUSION · REAL DATA

88%

of answers to questions that name Decoding Growth include it (158 of 179)

1 in 90

unbranded answers name it (26 of 2,328)

2 in 768

unbranded Google AI Overviews answers name it

Positioning, Preference and Selection

Scored in a live Scan against your four named competitors: how you’re described, whether you’re recommended, and whether your site and profiles back it up, each with the evidence behind the score.

Top five actions

Ranked by impact at your walkthrough

Recognition

Comparison content

Third-party sources

Profile hygiene

Site claims

Illustrative, to show the format. A live Scan ranks your five from your own results.

The read

Named when asked for by name, almost invisible when buyers describe the problem. The first fix is Recognition: being known as a player before anyone asks for you.

Source: Decoding Growth’s own tracking across ChatGPT, Perplexity and Google AI Overviews, 8 October to 6 November 2026. Method as in The Inclusion Problem. Rows other than Inclusion are illustrative.

Fit

Who this is for

Marketing and revenue leaders at B2B vendors: CMOs, marketing directors and VPs of marketing, founders and CEOs of smaller businesses, heads of demand generation, and CROs or sales leaders. Usually with a marketing team of somewhere between three and twenty-five people, in any sector where buyers compare suppliers in AI. The Scan is priced so a marketing lead can commission it on their own authority.

You’re likely to recognise this:

Your tools or your agency say you’re showing up in AI answers, and nobody can tell you whether that’s turning into shortlists

You’re named when buyers ask for you, and missing when they describe the problem you solve

A competitor keeps turning up in AI answers you’d expect to win, and nobody can say why

The board has asked what AI is doing to how buyers find you, and the honest answer is “we’re not sure”

THIS ISN’T THE RIGHT FIT IF

You want content produced on retainer.

I direct the work; your team or your existing agency does it.

You’re after more traffic.

I measure whether AI recommends you, which is a different number.

You want a monitoring tool or a dashboard.

A Scan is a measurement with a diagnosis attached, read through with you.

Your buyers don’t research with AI yet.

A few regulated categories still don’t, and if the Scan shows that, I’ll say so.

If you recognised yourself above, seven days is the quickest way to find out.

Working with your team

I work beside the agencies you already have

I don’t sell execution, so nothing on a Scan’s action list is there to keep me busy.

I sit beside the people already doing the work: your content and SEO team, your agency, your PR and analyst-relations people. I measure the one thing none of them can see from where they sit - which questions you’re losing at the shortlist, and why - and I direct the fixes in that order. They keep their roles, and most of the work in Selection Engineering is theirs.

Questions

Frequently asked questions

Do I have to start with the Scan?

For Selection Engineering, yes. It’s quoted from the Scan’s action list, so without the measurement there’s nothing to scope. If you’d rather talk first, book a call.

What does the Scan measure?

All four layers: whether you’re in the answer, how you’re described, whether you’re recommended, and whether your site and the sources behind you hold up when the buyer checks. Around thirty buyer questions, five AI platforms, four competitors you name. It also reports the Shortlist Rate, how often you’re in the recommended set on the decision-stage questions, which can be far lower than how often you’re named at all.

Which AI platforms?

ChatGPT, Google’s AI Overviews, Gemini, Perplexity and Claude. Each question is run repeatedly over a window, because the answers move between runs.

Is this GEO?

It overlaps. GEO work usually aims to get you mentioned more often. The Scan asks whether you’re chosen when a buyer compares, which is a later and harder test, and tells you which fixes matter for that. If your agency is doing GEO, the Scan gives them a precise target.

Who does the work?

Your team and your existing agency. In Selection Engineering I direct the order and the briefs, and check what’s moving. Where you don’t have the people, I can bring in specialist content creators, scoped and quoted at your walkthrough. There’s no ongoing content retainer.

We already have an agency. Do we need you?

Keep them. The Scan tells them where to point the work, and in Selection Engineering they do most of it.

How do you know it worked?

Selection Engineering ends with a re-run of your Scan questions and a report on what moved against your baseline. AI answers move for reasons nobody controls, so the report shows how much they move between runs as well as what changed.

What if the Scan shows we’re already well positioned?

Then you know, with evidence, and you know what to defend.

What does it cost?

The Scan is £1,995, fixed, with a money-back guarantee. Selection Engineering is a six-week fixed scope, quoted at your walkthrough from £15,000. No day rates and no retainers.

Do you work outside the UK?

Yes, with vendors in the UK, the US and Europe. Most of my measured sets so far are UK-framed, and sets for other markets are on the way. A Scan is built around your buyers, including the market they buy in.

What’s the Selection Index?

My quarterly measurement of one sector’s buying questions across the AI engines: who gets named, how, and which sources the answers draw on. The first edition, ecommerce fraud prevention and payment security, publishes in November 2026. If your category is covered, the Index shows where you sit. The Scan shows why, and what to fix first.

Next step

Start with the Scan

Seven days to a measured answer: where you stand when your buyers ask AI to compare, against four competitors you name, why, and the five things to fix first. Then a 60-minute walkthrough, and a quote for Selection Engineering if the numbers say it’s worth doing. If they say you’re already well placed, I’ll tell you that too.

£1,995 · seven days · 60-minute walkthrough · money-back guarantee

Or book a call if you’d rather talk first.

FREE NEWSLETTER · EVERY OTHER MONDAY

The Selection Layer

What AI is doing to B2B buying, and what to do about it. Each issue brings new findings from my own tracking, plus the outside research and commentary worth your time, read for vendors and for media and events owners.