Forrester: two-thirds of B2B providers will lose deals over claims buyers can't verify

Forrester puts proof at the center of 2027

On October 21, Forrester published its 2027 B2B go-to-market predictions. It expects business buyers to deploy AI agents across the buying process, from supplier discovery to contract-renewal analysis, while truly autonomous buying stays concentrated in high-volume, low-risk, repeatable purchases. Two predictions sit closest to marketing. First, two-thirds of providers will lose deals because buyers can't verify their claims. Second, buyers will double their human interactions to validate AI recommendations, going to provider teams, analysts, customer references and partners.

For a GTM team, this turns proof into shared infrastructure. The case study, the ROI figure, the reference customer's account and the rep's talk track are now checked against each other by the same buyer, often in the same week. Forrester's prescription is credible, consistent proof points that AI tools can index and validate. A claim that lives only in the deck, or reads differently on the website than on a call, is the kind Forrester expects to cost deals. The human touchpoints matter more, too: they're where buyers go to confirm what the AI told them.

Forrester also attaches a staffing warning. It predicts CMOs will cut as many as 25% of roles expecting productivity gains that fall short, then face a choice between cutting programs to rebuild capacity or accepting reduced execution. The people who keep proof current are often the ones those cuts reach first.

Demandbase and Peec ship agents for what AI assistants say about you

Demandbase announced three agents at its GO conference in New York on Oct 21: AEO, Campaign and Audience. The AEO Agent compares how major AI assistants represent a brand, shows which sources influence those answers, and flags outdated information and content gaps. A day later, Peec released Peec Agent, which ranks AI-search visibility fixes by impact. Peec's agent reads data freely but shows any change for approval first, unless a team turns on auto-approve.

The useful output for product marketing is the source list. An assistant's description of a product draws on review sites, analyst notes, partner pages and old launch posts as well as the company's own site, and these tools now show which ones it leans on. That makes third-party coverage part of how a product is positioned. MarTech Edge adds a caution: visibility monitoring doesn't guarantee a recommendation, since results vary by model, query and sources.

This is the other half of Forrester's prediction. Buyers verify with AI, and these tools show what the AI found.

Optimove puts QA agents in front of every send

Optimove launched Content Studio, a single workspace for campaign content across email, SMS, mobile, web, WhatsApp and ad networks. Before launch, a Content Advisor Agent checks content for brand alignment and a Content QA Agent catches broken links, missing images and invalid promotional details.

Pre-launch review is becoming automated, and these agents check against brand guidelines and approved assets that a person still defines and maintains. The agent can confirm a message matches the guide. Whether the guide reflects what the company sells today is outside its view. It's the same pattern as the AEO tools, pointed inward.

Anthropic opens Claude Duet, an agent that works while you're away

Anthropic opened a waitlist research preview of Claude Duet for US Pro and Max subscribers on Oct 21. Duet sends a morning brief, checks connected apps every few hours, and keeps working on assigned tasks in a virtual machine while the user's devices are off. Before sending a message, making a payment, signing a document or deleting anything, it shows the proposed action and waits. Team and Enterprise plans aren't included yet.

For GTM teams, agents that draft follow-ups in a seller's name will arrive on individual subscriptions before any team policy covers them. Pulling in Forrester's findings, it's a reminder that both sides of a deal now have agents in the loop: one researching vendors, one drafting the replies.

My take

Every item this week has a machine checking or repeating what a company says. The buyer's AI fact-checks claims. Answer engines summarize them. QA agents compare drafts to the brand guide. An agent drafts the follow-up. But what should the claim be?

I watched a version of this before generative AI was so widespread. A team brought in a sales content management system to fix content effectiveness. It solved versioning, a real problem, and leaders were right to celebrate that. It also locked in the wrong lessons: an asset referenced in one big win became the one that works, whether or not it had anything to do with what the rest of the market needed to hear. The system made sure content got delivered. It had no way to tell anyone whether the content was right.

Today's tools are far better at checking, and that's good news. They still check against whatever version they're given. Forrester's two-thirds will be the companies whose versions disagree with each other: the site, the deck, the reference call. That's GTM debt surfacing in a buyer's search results instead of a pipeline review. The story and the proof behind it come first, and this week's launches will carry whatever version a company has settled on, consistent or not.