Your playbook and scoring

Everything the analyzer knows about your business lives on the Playbook tab, behind two tabs named by the question each one answers. Everyone in the workspace can read both.

That matters: a rep is graded against this document, so they need to open it mid-deal.

• What we sell. Pick Company-wide for the company overview, ideal customer profile, required disclosures and banned phrases, red flags and disqualifiers, and tone of voice. Pick an offer for its details, core pillars, buyer personas, objections and handlings, FAQs, discovery questions, differentiators, and required talking points.
• How we grade calls. One card per call type, holding its name, colour, order, archive control, and two answers: Which calls count as this type? and What does a strong call of this type look like? The second is where "never present pricing on a discovery call" lives, and the grader holds reps to it.

A call type exists in exactly one place. The name and the classifier answer used to sit under Settings › Sales while the grading rules sat here, which meant the same four call types were listed twice in near-identical layouts. Settings now carries a pointer and a button instead.

Writing all of that from a blank page is the reason a playbook stays empty, so there is a Generate with AI button at the top of the Playbook tab, above both tabs. Owner and admin only, because it writes to the parts only they can edit.

Give it any combination of three things:

• A website address. Your homepage or an offer page. It reads what is on the page. It cannot sign in and it does not crawl the rest of the site.
• PDFs. Sales decks, one-pagers, offer docs, FAQs. They are read in your browser rather than uploaded, so file size is not a limit, but a scanned document has no text in it to read. Paste its key points into the notes box instead.
• Anything else, typed or pasted. Pricing, your ideal customer, common objections, how you run a discovery call against a closing call. The more you give it, the better the draft.

Generation runs in the background and takes a minute or two. You can leave the page and come back to it.

Nothing is written until you say so. The review screen groups the proposal into Add, Update and Remove:

• Additions and updates arrive ticked, and you can edit the wording before applying.
• Removals arrive unticked. You tick the ones you actually want deleted.
• Couldn't generate yet lists what there was not enough material to write. Those are left blank rather than invented. Fill them in later by generating again with more about each, without resubmitting everything else.

It can propose grading rules too, the ones under How we grade calls, and it matches the call type it names against your own list, allowing for punctuation and shorthand: a rule proposed for "Demo" lands on your Demo / Presentation type. Where it cannot tell which of your call types was meant, the proposal is dropped rather than offered. Anything you can tick is something that will actually land.

Every call is marked in eight categories, each out of ten, each with a quote from the transcript behind the mark.

A discovery call that never asks for the sale is a good discovery call. A closing call that never asks for the sale is a failure.

So each call type carries its own weighting.

• On a Discovery call, discovery, qualification and listening are weighted heavily. Closing barely counts.
• On a Closing call, objection handling and closing carry the most weight.

That is why the type shown on a call matters, and why you can correct it when the sorter gets it wrong.