Baseline: the Product Brief and its coverage
Baseline, under Identity, is what the rest is built on. Start runs Find, then confirm; the other tabs follow.
Nothing runs on its own. When HubSpot is connected and your website is on file, Start says Ready for analysis; otherwise it names what is missing. Analysis begins when you press Start analysis.
The tab row stays in place while the page under it scrolls. On Product Brief, Save and Cancel sit at the right of the tab row; Cancel puts the brief back as last saved.
Product Brief
What the company actually provides. The positioning is built on it.
- What you offer: each offer, who buys it, what they get.
- Who it is not for: who should not buy and what it does not do. Saying this plainly is one of the strongest trust signals a reader gets.
- What changed in the world, What it costs to ignore, What good looks like: the stakes. When a Campaign Brief in the Advisor leaves any of these empty, the plan uses the line from here. A line written in the Campaign Brief always wins.
One Product Brief per positioning. Pick the positioning at the top when there is more than one.
The Account Brief lives here too: the facts, approved claims and never-say rules sent with every AI generation on the account. Edit opens the same editor it always had.
Competitors
Add up to six competitors per positioning: a name and their website. Read reads their site with the same crawler as yours and lists what they claim (promises, proof, who they say it is for, how they say they differ), each kept only if the words are on their page. None of it can become a claim your drafts cite.
Find competitors lists up to eight sites like yours: a name, what they do (from their own page), the website, and why it came up. Press This one or Not one on each. This one accepts it and runs the research as one job: their claims, their news, and where they will be. Not one is remembered: that site never comes back, and the next search skips it. Accepted competitors seed the next search beside your site, so the second round is closer than the first. Add takes a name and a website for the ones a buyer names that the search missed; research runs the same way.
News lists their launches, pricing changes, funding, hires, partnerships and awards from the last 90 days, each with its page. Where they will be lists your confirmed events (Identity, Events) whose pages name them, with their part. A fact is kept only with the page that carries it. Nothing from your competitor list leaves your account.
Compare reads their claims against your pillars. Each pillar is Yours alone, Shared (a buyer cannot tell you apart there; say what only you can prove), Contested, or Not proven yet. It also names the old way: what the alternatives have in common. That is what your copy argues against. Copy never names a competitor; the Composer flags a draft that does, because naming one makes the email about you, not the reader. The Product Brief tab shows the latest comparison beside what you offer.
Market
Two read-only views of the market outside the business. Both search the web (Exa) and need EXA_API_KEY in the vault for mxa-identity.
Market brief. Write the brief searches from your pillars, your ICPs and up to three tracked topics, published in the last 90 days, and reads the results once. It names three to six shifts (what is moving and why it matters to a pillar), angles worth composing now, and what to keep an eye on. Every shift and angle carries numbered citations to the pages it rests on; one with no page behind it is left out. Briefs are kept per positioning; writing it again keeps the old one.
Topic radar. Track up to ten topics: a regulation, a cost your buyers watch, a change in how they buy. Check topics now finds pages published since each topic was last checked (the last 30 days the first time), with no AI call. Add as evidence puts a page in Hubs, Evidence as a third-party source to read for claims; Dismiss clears it.
Nothing on Market becomes a claim a draft can cite until it is added as evidence and its claims are approved. Radar's Ask reads the latest brief and the open pages.
Coverage
Your approved claims from Hubs, Evidence, laid out pillar by kind of proof: mechanism, number, named customer, testimonial, third party, perspective. Pick an ICP to see only the claims that speak to it; a claim tagged to no ICP counts for every ICP. Evidence proves results; perspective proves judgment, and a pillar needs both.
Gaps are ranked by what they block: the pools in approved or building campaign plans that write on that pillar. Each gap says what it blocks and what would close it:
- No proof on a pillar: rows on it have nothing to stand on at the prove beat, so they say less.
- One piece of proof: every row leans on the same claim.
- No stated position: teach and reframe beats have nothing of yours to carry.
Ask a customer lists what only a customer can give: permission to name one you already have, or a result on a pillar they will put their name to. Customer proof cannot be written into existence.
Approved claims with no pillar are not counted. Tag them in Hubs, Evidence.
The positioning starts here
AI Assist on a positioning (each section, Fill, Fix and the intake questions) starts from Baseline before it reads your sites and past messages:
- the Product Brief of that positioning: what you offer, who it is not for, the stakes;
- your approved claims from Hubs, Evidence, as they stand, customers unnamed until they agree;
- the latest comparison: which pillars are yours alone, which are shared or contested, and the old way. Competitor names are left out.
Where Baseline and the site disagree, Baseline wins. The site fills in what Baseline does not say. After a draft, a note says what it started from. A positioning not saved yet starts from your approved claims only.
Over time
Baseline is read again every night with HubSpot, and whenever you press Re-baseline now on the Over time tab. Each reading keeps the HubSpot record (rich, messy or blank), your approved claims per pillar and the open coverage gaps. The tab shows:
- when the HubSpot record changed;
- library depth per pillar, against 30 days ago;
- gaps closed and opened per month;
- the time from each Find, then confirm start to its first approved campaign.
Exemplars
Once there are 60 days of stage movement on record, the nightly read picks this account's own messages that came before contacts moved lifecycle stage, at a rate above the account's usual (at least 20 contacts reached and 3 moved). Up to three per move and audience. The Composer reads them when it drafts for the same move and audience: what they did, never their sentences, and never a claim that is not approved. Moved after a message is not proof it moved them.
To see what exemplars add, run the Quality test once with them and once with Without exemplars ticked, then compare the two runs.