Sponsored by

daydream

When I was an analyst at IDC, vendors took briefings seriously. They'd bring their product leads, their CMO, sometimes the CEO, and they'd rehearse. They knew that how an analyst understood their product could decide whether they made a buyer's shortlist, because that's where a lot of enterprise buying started.

That briefing still happens. It just has a new audience.

Forrester's 2025 Buyers' Journey Survey found that 94% of business buyers now use AI somewhere in their buying process, and buyers named generative AI and conversational search as a more meaningful source of information twice as often as any other source. Before your buyer ever fills out a form, there's a good chance they've asked ChatGPT, Gemini, Perplexity or a Google AI Overview who they should be looking at.

Nobody from your company was in the room for that one. Whatever the model said about you was assembled from whatever it could find.

daydream AI Visibility Leaderboard

Is AI recommending your competitor instead of you?

Are you in the top 3? Do you know who is? The AI Visibility Leaderboard from daydream will tell you for free. Built on 60,726 AI answers and over half a million citations in the last 3 months alone, it ranks 145 brands across 18 categories. See who's winning your category.

BONUS: Get a personalized AI visibility dashboard based on what real buyers ask AI. See how often AI recommends you versus competitors and which sources it cites, ready to share with your team once built.

The Shortlist Gets Built Before You Show Up

I've written a lot here about what happens after the first meeting. This edition is about what happens before it, because the two are more connected than most of us treat them.

When a competitor reaches a buying committee first, their narrative becomes the standard. Their features become the requirements. Their pricing becomes the benchmark. By the time you show up, you're defending against someone else's definition of the problem, and that's how you end up in a feature bake-off, discounting to win, or watching the deal stall out in no decision.

AI assistants speed that up. A buyer asks "what's the best customer data platform for a 500-person B2B company that runs on Salesforce?" and gets one tidy answer with three or four vendors and a few reasons for each. That answer becomes the starting point for the evaluation. If you're not in it, you may never get a chance to change their mind. If you are in it but the model got your pricing or deployment model wrong, you'll spend the first call correcting it.

The model quotes whichever sources answer the question most clearly. In a lot of categories, that's your competitor's pricing page, a two-year-old Reddit thread and a YouTube video somebody's customer recorded.

What the Data Says AI Actually Reads

Most AI visibility advice is still anecdotes. The most useful data I've found this fall comes from daydream's free AI Visibility Leaderboard, which tracks which brands and websites ChatGPT, Claude, Gemini, Perplexity and Google's AI Overviews recommend across 18 software categories, using real buyer questions pulled from Google searches, Reddit, G2 and Capterra. Full disclosure, daydream is this week's sponsor. The leaderboard is free and doesn't ask for your email.

A few findings from Houston Barnett-Gearhart, who runs their agency, stuck with me because they're so fixable:

A cited page with no answer is a missed shot. In legal AI, ChatGPT cited LegalOn's pricing page 12 times over two months on questions from in-house teams. On a budget-focused question, LegalOn didn't show up in a single answer. The page gave no price, and ChatGPT recommended the vendors who published one.

Buried facts don't count. In workflow automation, ChatGPT cited Make's pricing page in every answer to a "lowest cost" question. Boomi's per-message rate was sitting in its help docs, and ChatGPT didn't cite a single Boomi page.

Sources churn, and not evenly. From October to November, 45% of the citations in Google's AI Overviews turned over, and 39% of Perplexity's. ChatGPT was the steady one, with 90% of its citations going to the same sites month over month. In Google's AI Overviews, the share of answers citing YouTube jumped from 27.9% to 36.8%, while Reddit's fell from 27.0% to 22.3%.

Long pages may not get read all the way through. In Houston's early tests, Gemini found a hidden code on a 4,000-word page but missed it on a 5,000-word page and reported the text "cuts off." ChatGPT and Claude both found it on a 6,000-word page. It's one model and a small test, but if your most important detail lives at the bottom of a giant guide, I'd move it up.

None of that requires a big budget. Most of it is the content team doing the basics on pages that already exist.

B2B Tech Hides the Answers Buyers Need

This one's personal, because I've done it.

B2B tech companies are some of the worst offenders at hiding the information buyers need. Pricing is "contact sales." Implementation details live in a deck. Security and compliance docs sit in a trust center behind an NDA. Integrations are a logo wall with no explanation of what actually connects to what. In cybersecurity, where I've spent most of my career, it's even more extreme.

Some of that is for good reasons. Some of it is habit. Either way, an AI model can't quote what it can't read, and B2B buyers ask very specific questions. How is it priced at 200 seats? Does it integrate with Salesforce or ServiceNow? Is it SOC 2 compliant? How long does implementation take? What does it replace? If your competitor answers those in public and you don't, the model is going to tell your buyer about your competitor.

A friend of mine, Daniel Michan, writes about PKI and secrets management. A few days ago he published a LinkedIn article comparing Akeyless and HashiCorp Vault for secrets and PKI. Within three days it was on the first page of Google for that comparison, right under a dedicated comparison site. As someone who ran marketing at Akeyless, I'll admit that one made me look twice. A single clear, specific article from someone who knows the space can show up in the exact place a buyer starts their evaluation.

You don't have to publish everything. But you do have to publish enough for a model, and a buyer, to understand who you're for, roughly what it costs, how it's deployed and what it works with.

Eight Things to Do This Quarter

1. Build a prompt set from real buyer questions. Skip the keyword tool for this one. Pull questions from sales call recordings, RFPs, G2 reviews, Reddit threads and your own support tickets. Aim for 20 to 30, grouped by persona and stage. "Best [category] for a mid-market healthcare company" and "[you] vs. [competitor] pricing" are very different questions, and you'll want both.

2. Run them across the assistants your buyers actually use. ChatGPT, Gemini, Perplexity, Claude and Google's AI Overviews. For each answer, write down who's named, who's cited and what facts get quoted about you. Do it monthly with the same set so you can see the trend. A single snapshot will mislead you because sources change so much month to month. If you don't want to run it by hand every month, this is exactly what daydream's free dashboard does.

3. Fix the pages AI already cites before writing new ones. Pricing, security and trust, docs, integrations and comparison pages get cited constantly. Make sure each one plainly answers the question a buyer would ask: who it's for, roughly what it costs, team size, deployment options, compliance, and what it replaces.

4. Put the answer at the top. Lead with the fact, then explain it. If you have a 6,000-word ultimate guide, consider breaking it into focused pages that each answer one question well.

5. Show up where the models look outside your site. YouTube is gaining fast in AI Overviews, so clip your webinars, podcasts and founder talks and post them with real titles and descriptions. Ask happy customers for reviews that name the use case and company size. Get your experts into industry publications, communities and podcasts. LinkedIn articles get indexed quickly, too.

6. Write the comparison page before your competitor does. Make it honest, including who should pick the other vendor. Skeptical buyers trust it more, and models favor content that clearly explains when each option fits.

7. Don't flood the zone. It's tempting to spin up 500 AI-written pages. That usually creates indexing problems, pages competing against each other and a cleanup project. Fifty pages that answer real buyer questions will do more than 500 that Google ignores.

8. Connect it to pipeline, even imperfectly. Add "ChatGPT or another AI assistant" as an option on your "how did you hear about us" field. Ask on discovery calls what buyers already searched or asked AI before the meeting. Watch branded search and direct traffic over time. Most AI influence will never show up in your attribution report, which is one more reason to keep a floor of budget on brand work you can't perfectly measure.

My Take

The vendors who did best in my analyst briefings made it easy for me to understand what they did, who it was for and why it was different, and they backed it up with proof. Some of them didn't have the best product. But I could repeat their story accurately to a buyer because they'd given me something clear to repeat.

An AI model needs the same things. The difference is that it doesn't take a briefing. It reads what you've published, compares it with what everyone else has published and what your customers say about you, and it repeats whatever is clearest.

I watched ServiceNow and VMware win markets that analysts had considered settled for decades. They got out early and showed the market a new way to do the work. Buyers took that new set of requirements back to the incumbents, who couldn't meet them. The companies that write the clearest narrative in AI answers today are setting the requirements for tomorrow's deals in the same way.

If you want somewhere to start this week, pull ten questions from last month's sales calls. Run them through ChatGPT and Google, screenshot what comes back, and share it with your sales leader and your CEO. I'd bet at least one answer gets your pricing, your positioning or your competition wrong.

Then reply and tell me what you found.

P.S. If you'd rather start with data than screenshots, daydream's AI Visibility Leaderboard shows who's winning your category right now. It's free and doesn't ask for your email.

Compare Notes With Other Marketers

AI search and how to measure it come up in the MarketingHQ community every week. It's a free, private group of 120+ in-house B2B marketers comparing notes on exactly this kind of work, with a Leaders Circle for heads of marketing and the jobs list every Friday before it hits this newsletter.

No vendors. No pitches. Just marketers helping marketers.

Career Up: Marketing Jobs This Week

Hand-picked roles from the MarketingHQ Job Board. Community members get them first, every Friday.

Leadership

VP, Head of Marketing at Hannover Re (Orlando / $321–434K)

VP of Marketing at CXtec (Phoenix / $180–210K)

Product marketing, ops & growth

Senior Marketing Operations Manager at Drata (cyber) (California / $135–207K)

Event Marketing Manager at Triomics (New York / $100–170K)

Marketing Operations Specialist at Talkdesk (San Francisco / $86–144K)

Early career

Find many more roles on our MarketingHQ Job Board

Let's keep the conversation going on LinkedIn

I share more of my thinking on B2B marketing, GTM strategy, and the challenges of leading marketing between newsletter editions. If you enjoy the newsletter, follow me on LinkedIn and join the conversation.

If this email was forwarded to you, sign up here to get our weekly newsletter directly in your inbox.