Your support tickets answer the buyer you already won


Your support tickets answer the buyer you already won

HubSpot’s homepage now introduces “the Customer Relationship Harness.” In AI, a harness is the setup around a model that lets it do real work: the information it can access, the tools it can use and the guardrails on what it’s allowed to do. As strategy, I think it’s sharp. As a headline, I’m not sure most of the millions of people HubSpot wants to help know the word yet. But it’s a message still being worked out in public, and the team building it knows things I don’t.

It does raise a useful question for any message: who do you check it against? Most people who build with AI would say “harness” is obvious. Someone still deciding whether HubSpot is for them might not.

Your homepage faces the same question. If your team reads support tickets and customer success notes to decide what the homepage should answer, you’re checking it against customers who have already bought. I keep seeing teams do exactly that.

Tickets show what customers ask after they buy

The questions in tickets are real, but customers ask them while setting the product up, fixing something or trying to get the result they paid for. Someone who hasn’t bought yet wants to know different things: is this for a company like mine, what would it replace, and can I believe this page? Those are the questions your homepage has to answer.

Bob Moesta and Chris Spiek’s Jobs to be Done Timeline shows where each kind of question comes from. It maps how someone switches to a new product: a first thought that something has to change, then casual browsing, active comparing, the decision, and finally using what they bought. As the Timeline puts it, “The act of shopping is how they figure out what they want.” While prospects work out what they want, they also work out what worries them. That’s where homepage objections come from. Tickets come from the other end of the Timeline, once customers are using the product.

Sort your sources by when you collected them

List every place your team collects information about customers and prospects, and sort each one by when it’s captured:

  • Before the decision: sales calls, lost and stalled deals, reviews of the alternatives people compare you with. Use these for objections and comparisons.
  • At the decision: short interviews or surveys in the first days after purchase. Use these for what started the search, the pain behind it and what made them choose.
  • After the decision: support tickets and customer success notes. Use these for onboarding, help content, retention and the gap between what customers expected and what they got.
  • Leaving: churn reasons. Use these to find promises the product didn’t keep.

This prompt does a first pass in ChatGPT, Claude or whichever AI assistant you use:

I'm a marketer at [company]. Below is every place we collect information about customers and prospects:
[paste your list, e.g. support tickets, CS call notes, sales call recordings, win/loss notes, onboarding survey, G2 reviews, churn survey]
Use the Jobs to be Done Timeline by Bob Moesta and Chris Spiek as the map. It follows a buyer from the first thought that something needs to change, through passive and then active looking at options, to the decision, and then actual use of the product. Add one more stage after that: leaving.
For each source, tell me:
1. Where on the buyer's timeline it comes from: before they decided (first noticing the problem, browsing, actively comparing), at the decision, after purchase (onboarding and ongoing use), or leaving.
2. Which marketing question it can answer reliably, and which it can't.
3. How fresh the buyer's memory is when we capture it.
Then list the stages where we have no source at all, and suggest one practical way to fill each gap.
Don't invent sources I didn't list. If a source is ambiguous, ask me instead of guessing.

Check the assistant’s sorting against how each source is really collected inside your company.

Fill the before-the-decision gap with interviews

If you have little from before the decision, start with people who bought in the last few days. In a switch interview, you walk back through the decision with them: what first made them look, what they compared, what worried them and what tipped it. Ask soon, because memory fades. The Timeline notes that “The first thought is almost always further back in time than people report,” so when someone tells you where it started, ask what was happening before that.

Then talk to prospects who didn’t buy. Customers can only describe the doubts they got past. Lost deals show you the other side: the doubts that stopped a sale.

Test a page with simulated prospects

You can’t run new interviews for every headline you want to try. Simulated prospects help here: AI stand-ins built from research with people still deciding, like sales-call notes and lost-deal interviews. You show them your page and ask what’s unclear or hard to believe.

For my session at the Scaling SaaS Summit on 18 Nov, I had no research on the company I used as the example, so I worked from 30 public reviews of its product. I rewrote the above-the-fold value proposition of its homepage in customers’ own words, then tested the old and new versions in 32 simulated interviews for under $1. The test showed where the new version was clearer and what new hesitations it raised.

Those reviews were fine for a quick demo, but they share the blind spot of support tickets: customers who already bought wrote them. Build simulated prospects from either, and they’ll ask what your customers already ask, while the doubts that stop prospects stay hidden. Normally, I’d build them from full research with people still deciding. Even then, the simulation proves nothing on its own. It tells you which doubts to take into real conversations, to confirm or kill.

So before you rewrite your homepage, check when each of your sources was captured. If most of them come after the sale (support tickets, reviews of your own product, customer success notes), you’re answering people who already said yes. Your homepage has to answer what worries people before they say yes, so get that from prospects still deciding, people who just decided and the deals you lost.

Discovery

Your context will outlast your AI tools

On the a16z Show, Ken Liu, co-founder of First Person Labs, introduces Second to a16z’s Martin Casado and Joel de la Garza. The problem they discuss: every new AI agent still has to learn who you are. Second is a “personal context layer” that understands what you’re doing across your digital life and makes that context available to the agents you choose.

It’s part of a pattern I keep seeing. Better knowledge work with AI comes from the context an agent can draw on: your priorities, your past decisions, how your company describes what it does and for whom. Agents will keep changing. The context you curate for yourself and your company carries over to whichever one you use next. If you haven’t started building that personal or company “brain,” this episode is a good nudge. Listen to the episode.

Watch HubSpot rethink its message in public

Dharmesh Shah, HubSpot’s co-founder, is posting a note every day about what he calls HubSpot’s “refounding,” his public rethink of the company’s mission and message. In the Day 3 note, he cut two words from the mission, from “Help millions of organizations grow better” to “Help millions grow better,” because a lot of the people HubSpot wants to help aren’t full-on organizations.

Whatever you make of “harness,” it’s rare to watch a company HubSpot’s size rework its message out loud, one post at a time. Worth following while it lasts. Read Dharmesh’s Day 3 note.

Resonance

“...most think of buying and onboarding as the beginning, but we always think of it as the end of part one in a journey to make progress.” — Bob Moesta and Greg Engle, Demand-Side Sales 101

Hi, I'm Chris, The Conversion Alchemist

I'm the founder and chief conversion copywriter at Conversion Alchemy. We help 7 and 8 figure SaaS and Ecommerce businesses convert more website visitors into happy customers. Unpacking Meaning is the only newsletter B2B SaaS leaders need to sharpen messaging and shorten sales cycles. A weekly email with one field-tested idea you can use to boost conversions without raising ad spend, make value obvious and friction low, and align teams with clear, scalable messaging.

Read more from Hi, I'm Chris, The Conversion Alchemist
Unpacking Meaning cover: why this at all

Are you winning “why us” and losing “why this at all”? Picture someone at a company who’s found your product and likes it. They get why it’s better than the other options. But in B2B, the person who likes you is rarely the person who signs off on the money. Before anything gets bought, they have to go to their boss, their finance team or the rest of the team and answer a much more basic question: “Why should we spend money on this kind of thing at all?” If they can’t answer that, it doesn’t...

When your next trial user is not human... AI assistants such as Claude Code, Codex and Cursor now write a lot of software. And more and more, they also choose the tools that software runs on. A developer asks for a place to store data or a way to send emails, and the agent compares the options, picks one and moves on. So who’s reading your pricing page first? More and more often, it’s an agent. Armature, who helps developer-tool companies get recommended by these assistants, ran nearly 17,000...

I built an AI coach that makes me do the work A big dilemma for every marketer using AI to write copy. Actually, to revise copy: You still approve it, but if AI chooses the problem, picks the evidence and writes the argument before you get involved, you’ve skipped several opportunities to practice making those decisions yourself. Reviewing the result takes judgment too. But recognizing a decent answer and arriving at one are different kinds of work. I don’t want my writing practice to become...