Walk into the pitch with the deck already written.

A deck. Ten slides with a beginning and an end, built from measured data about the prospect, carrying your agency instead of ours.

The measurement ends in a dashboard you screen-share. That works while the call is running. The person who signs off the retainer is often not on the call, and a login does not get forwarded.

The honest version of the claim: the deck does not win the pitch. Agencies told AgencyAnalytics that industry expertise (59%) is what wins pitches, with clear ROI and outcomes and case studies tied at 46% (n=494, surveyed February to April 2026). The deck supplies one of those three: the ROI and outcomes, measured on the prospect before the pitch. What it does is put you three slides ahead of whoever pitches after you.

In order: the cover, where the prospect stands on the questions that win new customers, how AI answers about their own name, how that splits across the engines measured, the gap to the goal, three levers (topics to create, pages to reach out to, pages to optimise), a dated 30-day plan, and the next step.

The headline figure carries its provenance on the slide: 99 prompts, two engines, 12 runs each, 1 to 14 September.

The sample earlier on this page is the full deck. Read all ten before you decide whether it is worth a prospect's time.

The sample deck was built from a fourteen-day window, because frequency across repeated runs is the whole method. A one-day snapshot is the thing the method is built to avoid.

A shorter window is possible, with fewer runs behind each figure. The provenance line under the headline figure states the window, so the prospect can see which one you used.

Yes. Treat it as a draft you own rather than a locked export: edit what needs changing and cut the slides that do not fit the room.

We generate the half you cannot write without data. The half that actually wins the account, which is why your agency and why this market, is still yours to write.

Both are good products, and both do the measurement half. The difference is what you leave the room with.

Checked September 2026: Profound ships Agency Mode as on-demand pitch environments that expire with the contract, which is a workspace the prospect visits. Peec AI publishes an MCP recipe called "Build Live Data Pitch Slides", pitched as "Win pitches with data, not slides", and it meters pitch projects by tier.

So the recipe exists — and you still assemble the deck. That is the gap, and it is a narrower and more defensible one than pretending nobody has thought about slides.

We are arguing against that headline, and not because slides beat data. The deck is the data. It is the data in a form that survives being forwarded to the two people who were not on the call.

Judge all three on the same tests: does the output leave the room, is it in your branding rather than the vendor's, does it end in a plan with the effort scoped in hours, and is it metered per pitch. Vendor tiers and prices move, so check theirs yourself rather than taking ours.

You point at the line under the headline figure on slide two — 12 runs each, over fourteen days — and you are the only person in the room who saw it coming.

SparkToro and Gumshoe ran 12 brand-recommendation prompts 2,961 times across ChatGPT, Claude and Google AI Overviews and AI Mode, published 28 January 2026. The odds of getting the same list of brands twice came back under 1 in 100, and the same list in the same order closer to 1 in 1,000.

So the deck never prints a single run. Every figure is a frequency across repeated runs over a stated window, with the prompt set, the engines, the run count and the date range printed under the headline figure. A live run that comes back different is one more sample, not a contradiction, and the deck already told the room to expect it.

Once, on slide two, and never from a single run. The sample reads 6.9% citation share, ranked #1 of 6 on the non-brand questions, and that rank orders shares measured across 12 runs over fourteen days.

What we will not print is the rank one answer happened to give. That is the number the variance study above shows to be noise, and it is the one a prospect can contradict from their phone in the meeting.

Whichever ones the project is set to measure. The sample measured two, Google AI Overview and ChatGPT, and says so on the cover.

The cover also shows three engines buyers use that were not measured, and says so directly beneath them. Slide 4 repeats it: no figure in the deck describes those three.

Google is right, and we are not going to argue with the source. From its generative-AI guidance, published 15 May 2026: "From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO." The same guide says you do not need llms.txt, you do not need AI-specific structured data, and you do not need to chop your content into pieces.

Every recommendation in the deck stays inside that. The tactics are the ones you already sell.

What Google does not hand you is the measurement: which answers name this company, how often, over what window, and who gets named instead when they do not. That is what the deck is, and it is the part your prospect cannot check for themselves.

Not by design, and we would treat it as a bug if it did. Google updated its Search spam policies on 15 May 2026 to cover attempting to manipulate generative AI responses, naming bought or altered citations, biased listicles and recommendation poisoning. Penalties run from demotion to removal from Search.

None of that appears in a generated deck, and no deck guarantees a citation, a placement or a mention. A plan that promises an outcome nobody controls is the fastest way to lose the client in month three.

Yours. Your logo and name on the cover, and your logo and contact details on the closing slide. The slides between carry the prospect's domain in the footer.

No client-facing Wellows branding appears anywhere. You stay the agency in the room; we stay behind it. That holds on every plan from day one rather than as a tier unlock.

It is not supposed to, anywhere, including the places nobody checks. File metadata is where a rebranded export usually gives itself away, so on your first one read whatever properties your export carries as well as the slides.

If our name is in a deck you are about to send, that is a bug and not a setting. Tell us and we will fix it.

That is the point of it. White-labelled prospect audits exist; SEO audit generators have done them for years. What none of them do is measure the answer layer and end on a dated plan with the effort scoped. That is the version this is.

Nothing in the deck is private data. It is measured from AI answers anyone can generate about a public website.

Yes. Plenty of agencies have a template their new-business meetings run on, and a ten-slide document that arrives with its own layout is not going to displace it.

Take the figures and the provenance line into your own template. That provenance line is what makes the number defensible when someone in the room asks where it came from, so it has to travel with the figure rather than be left behind with the layout.

No. It is included on every plan at no extra cost, with no per-pitch fee and no pitch counter. It is not an add-on and it is not a higher tier.

What you need is a Wellows workspace. What you do not need is a bigger one because this month has three pitches in it rather than one.

Not self-serve yet. We are opening it up through demos first, so book one and we will walk you through the deck slide by slide.

When it opens, it will be included on every Wellows plan at no extra cost. Until then, the sample above is the full ten-slide deck.

Because the artefact is the demo. Nothing we could write on a website makes the case as well as ten slides about a company you are already chasing.

And because the money is in what happens next. If the deck helps you win the account, the client project runs in here afterwards. Charging per pitch would trade that for a small fee at the exact moment you are least sure about us.

Use it. Clarity Citations went generally available on 13 May 2026, free with no traffic cap, reporting page-level citations, share of authority and AI referral traffic.

Two limits for this particular job. It covers Microsoft AI surfaces, Copilot and Bing generative search, rather than ChatGPT, Gemini, Perplexity, or Google AI Overviews and AI Mode. And it reads a site you have installed a tag on. A prospect is not going to install your tag before the first meeting.

Pre-sale measurement has to work on a company that has given you nothing, which is why the deck is built from answers rather than from analytics.

It becomes the first month. The plan on the second-to-last slide is already dated and ordered, so kickoff is a conversation about start dates rather than about what to do.

The prompt set you measured them on before they signed is the one you keep measuring them on afterwards, so month one reports against a baseline you took while they were still shopping. Very few agencies can show a client where they stood the week before the contract.

That is what it is shaped for. Every recommendation names a page or a topic and is ordered by what it is likely to move, and the plan slide totals the effort in hours: 64 across the four weeks in the sample.

The hours are the input you price. Apply your own rate and the plan becomes a quote; hand the same hours to your team and it becomes a sprint.

You still have a measured read on a company in your target market, and a document worth sending the week after the meeting, when the winning agency has gone quiet.

Losing a pitch usually leaves you with nothing to follow up with. This leaves you with a dated baseline and a reason to send the next one.

No. It sits on top of it. The work happens in the answer layer, which is whether AI systems cite your client, alongside the ranking work already on your service menu.

Sold as its own line, it is the thing clients are already asking about. Sold as a replacement, it is a harder conversation with no upside.