John Webb - AI selection consultant
I help B2B vendors get selected, not just mentioned, when AI systems build buyer shortlists.
I'm the founder of Decoding Growth and the author of The Selection Layer: the model, the benchmark methodology, and the diagnostics built on it. I work one-to-one with B2B SaaS and technology vendors, from the UK, with clients across the UK, US and Europe.
○ WHY THIS
I've been on the other side of the shortlist
For two years I ran Autovia - Auto Express, Carbuyer, evo, DrivingElectric - where two in every three in-market car buyers passed through our titles during their purchase decision. My job was the machinery between car brands and car buyers: the content, the first-party data and the intent signals that shaped which vehicles made the shortlist. Before that I led demand generation across Dennis Publishing's IT, automotive and financial services brands, connecting future and in-market buyers to vendors one-to-one.
I spent those years operating the layer that decided who got considered. Then that layer moved into a machine, and almost nobody could see inside it.
That's what Decoding Growth measures.
○ THE WORK
What I do
Three things, in sequence, all built on the same benchmark methodology:
- Diagnose. The AI Shortlist Scan establishes where a brand stands across all four layers of The Selection Layer - Inclusion, Positioning, Preference, Selection - using real buyer prompts across five AI platforms, repeat-tested over a window.
- Engineer. The Selection Engineering Sprint changes it: the top five actions from the Scan, the sources that decide the category, the coverage that gets a brand cited.
- Optimise. A quarterly benchmark re-runs the prompt set and reports movement, because AI answers move and a position has to be held.
I direct; client teams execute. I don't run an execution retainer, which means there's nothing in the diagnosis I'm protecting.
○ THE MODEL
The Selection Layer
The Selection Layer is the space between a brand and its buyer where B2B shortlists now get built. I developed it because the existing vocabulary - visibility, rankings, mentions - measures the wrong thing. Being visible isn't being chosen.
It has four layers, each measurable and each engineerable: whether you're in the answer, how you're framed when you appear, whether you're recommended, and whether the buyer's check on what AI said confirms it.
The model is free. Everything I publish gives it away. What's paid is the prompts, the scoring, the interpretation and the engineering.
- Read the argument in full: How B2B Buyers Actually Buy in 2026: The Evidence
- The category, defined: What Is The Selection Layer?
○ BACKGROUND
Twenty-five years, one pattern
Current benchmark work spans ecommerce fraud prevention and business technology, with the first sector Selection Index publishing this autumn.
Before Decoding Growth I spent twenty-five years in commercial and leadership roles across technology, media and consumer brands: European B2B Marketing Director at Samsung Electronics, EMEA & APAC lead at Spiceworks, marketing and startup programme director at Rackspace, International Marketing Director at Rockstar Games, and UK & Ireland Marketing Director at THQ, alongside roles at Yahoo! and InfoSpace Mobile. Managing Director at Dennis Publishing and then Autovia.
Six different markets taught me the same lesson: buyers form a view of who's worth considering long before anyone in sales hears about it. What's changed is where that view now forms.
○ HOW I WORK
How I work with clients
Fixed scope, published prices, evidence for every score. No day rates, no retainers, no engagement without a benchmark component.
Every engagement starts with a Scan, because I'd rather show you where you stand than tell you what I think. In first conversations I'll usually run three of your real buyer prompts live, so you can see the method before you buy anything.
○ SPEAKING
Speaking and podcasts
I talk about one thing: how B2B buyers now build their shortlists inside AI conversations, and what it costs vendors who can't see it happening.
Every talk is built on work I've run rather than a summary of other people's research - real buyer prompts, repeat-tested across ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini and Claude, scored across four layers. In most rooms I'll run three prompts live on a brand in the audience. The argument lands differently when the answer comes back in front of you.
Three formats: a 20-minute conference session, a 45 to 60 minute briefing for a single team or an invited room, or a podcast conversation. I'll also chair panels, which is often the more useful version - the people on stage with me usually have the problem I'm describing.
Recent stages include Advertising Week New York 2025, where I chaired "B2B's New Power Play: Lead Gen Rewired", and bauma 2025 in Munich, on turning data into deals. Earlier, B2B Ignite in London.
If you're programming an event or booking a show, get in touch.
○ QUESTIONS
Frequently asked questions
Visibility work asks whether you're mentioned. I ask whether you're selected. The distinction is measurable: your Shortlist Rate - how often you're in the recommended set rather than merely in the answer - is usually far lower than your mention rate, and the gap between the two is where deals are lost. I also don't sell execution retainers, so there's nothing in the findings I have an interest in.
B2B SaaS and technology vendors whose buyers research the category with AI before contacting anyone. Typically the person who engages me owns the number - a CMO, marketing director, CRO or founder - and has noticed demand softening without an obvious cause. I'm based in the UK and work with clients across the UK, US and Europe.
Yes, on how AI-mediated shortlists get built and what vendors can do about it. Conference sessions, private briefings for a single team, panels I chair, and podcasts. I bring benchmark data from my own runs rather than a slide of other people's statistics, and I'll usually run live prompts on a brand in the room. Event programmers and podcast producers can reach me through the contact page.
The AI Shortlist Scan is £1,995, delivered in seven days with a 60-minute walkthrough and a money-back guarantee. The Selection Engineering Sprint is fixed scope from £15,000 for six weeks, scoped from Scan findings. Prices are published because a diagnostic you can't price isn't a diagnostic.
Twenty-five years in commercial and leadership roles across technology, media and consumer brands, including Samsung Electronics, Rackspace, Rockstar Games, Spiceworks and THQ, then Managing Director at Dennis Publishing and Autovia. At Autovia I ran the content, data and intent machinery that shaped which cars made buyers' shortlists - which is the same problem I now solve for B2B vendors, except the shortlist is built by a machine. Decoding Growth is that specialisation.
Yes. The Selection Layer newsletter goes out fortnightly with the thinking, the benchmark findings and what they mean. The library on this site carries the full articles. The model is free by design.
○ NEXT STEP
Start with the evidence
If you want to know where your brand stands when AI builds the shortlist, the Scan answers it in seven days.
£1,995 · seven days · 60-minute walkthrough · money-back guarantee.
Or follow the work: The Selection Layer, fortnightly
Connect on LinkedIn
Page last updated: 7 September 2026