Prospects keep arriving at sales calls already knowing things you don't publish. Here's why.
Prospects keep arriving at sales calls already knowing things you don't publish. Here's why.
Prospects keep arriving at sales calls already knowing things you don't publish. Here's why.
If your sales team has started saying things like "the buyer already knew our pricing tiers and we don't publish them" or "the prospect referenced a feature we deprecated two years ago and never wrote about," they are not imagining it. Something specific is happening, and it's now happening at scale.
The symptom
Three patterns we hear repeatedly from sales leaders in 2026:
- Buyers reference unpublished pricing in the first call. Sometimes accurate, sometimes anchored on a year-old number nobody on the team recognizes.
- Buyers cite specific feature comparisons against competitors that nobody at your company has ever written or said.
- Buyers ask compliance questions ("how do you handle PHI in EU regions?") that your website doesn't actually answer.
The information is wrong as often as it's right. But the buyer is acting on it as if it's authoritative. By the time you're on the call, the anchor is set.
The diagnosis
Your buyer is no longer doing the research. Their buyer agent is. And the buyer agent doesn't limit itself to the surface you control.
When a buyer asks ChatGPT, Claude, or Perplexity to research your company, the agent fans out to wherever your information might live. That includes places you forgot you exist:
- Old marketplace listings. A pricing tier you posted on AWS Marketplace three years ago is still indexed. The agent treats it as a primary source.
- Third-party review sites. G2, Capterra, TrustRadius. Reviews from 2023 referencing features that no longer exist. User complaints about pricing you've since changed. The agent reads them as current.
- Job postings. Your engineering team posted a role mentioning a deprecated product line. The agent finds it and assumes you still ship it.
- Conference talks and podcasts. A founder mentioned an internal pricing benchmark on a 2024 podcast. The transcript is indexed. The number gets quoted back as canonical.
- GitHub, Reddit, Stack Overflow. Engineers, support reps, and customers discuss your product in places you don't moderate. The agent treats it all as evidence.
- Old press releases. A funding announcement from 2022 is more findable than your current product page. The agent uses it as the canonical "what does this company do."
The published surface you spent a decade optimizing is no longer the only signal. It's one signal among a dozen, and most of the others are out of date.
Why this is new in 2026 (and not before)
Three years ago, a buyer doing research read your website. Maybe a G2 page. Maybe a press release. They read selectively, and they read what was current.
A buyer agent in 2026 reads everything in parallel and treats it all as evidence with roughly equal weight. Recency is not strongly weighted by default. Source authority is not strongly weighted by default. The agent synthesizes whatever it finds and reports back as if the synthesis is the truth.
This is structural, not a bug. It will not be fixed by waiting for the LLMs to "get smarter." Source weighting and recency are model-level decisions that vary across providers and update unpredictably. The only durable fix is making the right answer easier to find and harder to mistake.
What to do about it
Three moves, in order:
- Get visibility into what agents are actually pulling. Most analytics tools don't separate buyer agent traffic from human traffic, let alone classify which agent. You need a view of which questions agents are asking about your company and what they're getting back. Without this, you are debugging blind.
- Become agent-ready. Audit the indirect surfaces. Update old marketplace listings to redirect to current pricing. Reply to outdated G2 reviews with corrections. Update job postings that reference deprecated products. The goal is not to control everything; it's to make your current truth more findable than the stale alternatives.
- Run Dynamic Agent Optimization. When an agent visits your site, serve it a clean, current, governed answer instead of letting it stitch together inferences from old marketing copy. The static-page surface is the one channel you fully control. Make sure it overpowers the noise.
The leak isn't going to stop. The buyer's agent is going to fan out and read whatever it can find. The question is whether the loudest, most current, most correct answer about you comes from your stack, or from an old marketplace listing nobody on your team remembers.
What this changes for sales
A short note on the conversation that changes inside your sales org. The "discovery call" as it existed in 2022 is mostly gone. The buyer arrives anchored. The first 10 minutes are no longer about uncovering needs. They're about correcting agent-formed misimpressions and re-establishing context.
Sales reps who try to run the old discovery script come across as wasting the buyer's time. Reps who lead with "let me clarify what you've probably already read" win the room. The change is structural, and the smart sales leaders are retraining for it now.
