BOFU is the new TOFU: Why your help center is now a first impression

Every company website has a page like this. A support engineer writes a clear, careful article explaining how to fix an uncommon error. It does its job. The handful of customers who hit the problem find the fix, and the ticket queue gets a little shorter.

Months later, a prospective buyer asks an AI assistant whether your product is reliable. The answer cites that support article as evidence of a known issue. It doesn't mention how rare the problem is, or that it was fixed two releases ago. The buyer moves on to the next vendor on their list.

Nothing in the article was wrong. It was simply read by someone it was never written for.

In marketing shorthand, bottom-of-funnel (BOFU) content serves people who are close to a decision or are already customers, while top-of-funnel (TOFU) content reaches people who are just starting their research.

Answer engines are blurring that line. They draw on whatever content seems most relevant and credible for a question, and that's often your most detailed, operational content. Support articles, documentation, pricing pages, and legal terms can now shape a prospect's first impression. BOFU, in other words, is the new TOFU.

This post looks at why that happens, why marketing now has to account for content it didn't traditionally own, and how to run your content so it holds up wherever it's read.

Get in touch

Palmata by Contentful

Manage AI discovery and reputation at scale.

The funnel used to mirror the org chart

For years, content types mapped neatly onto audiences and the teams that served them. Marketing pages were written for prospects. Documentation was written for developers. Support articles were written for customers with a specific problem. Legal pages were written for procurement and compliance reviewers.

Answer engines don't see those boundaries. When someone asks ChatGPT, Gemini, or Perplexity about a product, the engine can draw on public information from across the web, including a company's own pages and third-party sources, and combine it into a single, synthesized answer. It doesn't know which team owns a page, or which stage of the journey it was meant for.

That's the shift answer engine optimization (AEO) is concerned with: making sure your brand is cited, and represented accurately, when AI systems answer questions about your category. It doesn't replace search engine optimization. Palmata's Head of Product & Engineering, Jason McGhee, dug into this in our recent webinar:

"Today, SEO [is] absolutely critical to even be part of the conversation."

Strong SEO is what gets your content into the pool that answer engines draw from in the first place. AEO is about what happens next: whether you're included in the answer, and how you're described.

Content now has to win twice

In the webinar, Taylor Wagner, who leads product marketing for Palmata, described this as content that has to "win twice." Every page has to work for the answer engine that selects and interprets it, and then for the person who reads the resulting answer.

The two audiences look for different things. People seek out relevance, credibility, and a clear reason to care. Answer engines look for structure, completeness, context, and evidence they can connect to specific claims.

That doesn't mean writing two versions of everything, or writing for robots. Jason noted that some of the earliest advice on writing for answer engines, such as rigid question-and-answer formats and tightly limited context, has aged quickly. Over the past year especially, models have become much better at understanding content and can read far more of it. What matters more is a cohesive, consistent story across everything you publish.

In practice, winning twice is mostly about context. Take a product limitation. A bare statement of the limit is accurate, but out of context, it can read as a weakness. The same entry is much harder to misread when it explains when the limit applies, who it affects, what alternatives exist, and where to find the current specification. People benefit from that context too.

AI reputation makes governance a marketing priority

Here's the uncomfortable part: marketing doesn't own most of the content that can shape an AI-generated answer about your company, but it now has to account for all of it. Details that once sat quietly in documentation or a support article, with little effect on how the brand was perceived, can now surface when a buyer asks their first question.

Ownership of that content spans the business:

• Marketing owns positioning and campaign content.

• Product defines capabilities and limitations.

• Support records known issues and their resolutions.

• Legal manages terms, privacy, and compliance language.

• Developer relations and documentation teams publish implementation details.

• Communications handles public events and company changes.

Each team is doing its job. None of them was asked to consider how its pages might read when pulled into a buyer's first question about your product. Jason described seeing support docs, API references, and even terms of service cited in answers in ways that misrepresent the product.

This is also why ownership matters more than it first appears. In a live poll during the webinar, few attendees picked "deciding who's responsible for AEO" as their biggest challenge. Yet Jason noted that it's a common gap for organizations getting started with AEO.

The answer isn't central approval of every sentence. That would slow everyone down, and teams would still miss things. What teams need is shared visibility into which sources influence important buyer questions, and a clear owner to act when a page turns out to be outdated, incomplete, or easy to misread.

What a connected content operation looks like

If any public page can shape a first impression, content has to be managed as one connected system rather than a set of separate libraries. A few practices make that much easier:

• One source of truth for repeated facts. Pricing tiers, usage limits, specifications, and availability tend to appear across marketing pages, docs, and support articles. When those facts live in one place and are reused everywhere, a single update reaches every page, and old versions don't linger.

• Fields for context. A content model can give context its own structure: who a page applies to, when it was last verified, what scope it covers, and which resolution relates to a known issue. Context that's built into the model is less likely to be dropped.

• Deliberate connections. Linking a limitation to its workaround, or a known issue to the release that fixed it, helps readers and answer engines see how the pieces fit the current product story.

• Review cycles for customer content. Campaign pages tend to get regular attention. High-exposure support articles and docs often don't, and they deserve a review cadence too.

• Ownership of the content itself. When the responsible team is recorded with the content, the right people can be found quickly when a problem surfaces.

These are the same principles behind structured content, applied to a new kind of risk. They also make fixes faster. In the webinar Q&A, Jason described using the Palmata MCP server alongside the Contentful MCP server, so a team can go from asking what needs to change to updating the content in the same workflow.

Where to start

No team can review every public page at once, and no team should try. Start with content that appears repeatedly in answers to strategically important buyer questions, or that contributes to a meaningful misunderstanding about your product.

From there, most fixes fall into three categories:

• Clarify existing content by adding scope, dates, conditions, and resolution details.

• Connect related content so limitations, updates, and policies link back to the wider product story.

• Create missing explanations where competitors, review sites, or scattered pages are currently answering important questions for you.

Finding those priority pages takes evidence: which questions buyers are asking, how answer engines describe you, and which sources they rely on. That's the work Palmata was built for. Our companion post on the Palmata blog goes deeper into how to diagnose and fix these problems.

Manage public content as one environment

Content no longer stays with the audience or funnel stage it was written for. A support article, a pricing page, or a line in your terms of service can now be part of a buyer's first impression.

That doesn't mean turning every page into marketing copy. Each page still has to serve its primary reader. It also has to hold up when it's quoted somewhere else, without its surrounding context. Teams that manage public content as a single connected environment are far better placed to ensure it does.

Want to see real examples? Watch the webinar on demand to hear Taylor Wagner and Jason McGhee walk through how answer engines interpret brands, and what teams can do about it.

Take a tour

Chat with our team

Your experiences should wow, not wait. Let's talk.