01
Prepare every first call in minutes
Also called Discovery call preparation
Before a first call, an agent pulls the full history from your CRM, researches what the company is working on now and builds a one-screen brief: who they are, what changed, who else is involved, three guesses about their priorities and five questions worth asking.
Why it matters. Reps who arrive prepared ask better questions and earn a second meeting. Research by hand takes too long to do for every call.
How to do it
- Confirm the company and the people on the call.
- Pull everything you already know from the CRM: emails, meetings, past deals.
- Research current priorities from news, reports, posts and job ads.
- Draft three hypotheses about what matters to them and five questions.
- List what the agent could not verify, so nobody repeats a guess as fact.
What goes wrong
- Briefs that run to three pages. Keep it to one screen.
- Unsourced claims. Every fact needs a source and a date, or is marked as a guess.
Run it with the First call prep agent
02
Keep a live picture of your key accounts
Also called Account intelligence brief
Each priority account gets one pinned note with two parts: what you already know from the CRM, and what is happening at the company now, with a source for every point. An agent refreshes it on a schedule and flags anything that looks wrong.
Why it matters. Reps change, notes scatter and history is lost. One trusted page per account keeps everyone, including AI, working from the same facts.
How to do it
- Part one from the CRM only: who you know, how engaged they are, past deals and why they closed.
- Part two from research: priorities, recent changes, what leaders said publicly, why now.
- Tie each suggested angle to a specific finding and person.
- Replace the old brief each time instead of stacking new ones.
- Run a few accounts at a time, highest priority first.
What goes wrong
- A brief that sounds impressive but cannot be checked. Cite or mark it as a guess.
- Attributing quotes to the wrong person. Only quote budget holders, and say so when nothing public exists.
Run it with the Account intelligence agent
03
Turn every call into next steps and CRM updates
Also called Post-call analysis (SPICED) and CRM write-back
After a call, an agent reads the transcript, writes up the situation, the problem, its impact, any deadline and how they decide, lists the buyer’s questions and proposes CRM updates. Nothing is saved until the rep approves.
Why it matters. Call notes are the first thing to slip when a week gets busy, and pipeline reports are only as good as the notes behind them.
How to do it
- Pull the transcript once.
- Write up situation, problem, impact, deadline and decision process, each with evidence from the call.
- Where something was not discussed, say "not covered, ask next time" instead of guessing.
- Log the buyer’s questions and who owns each answer.
- Propose CRM changes as a before-and-after table and wait for approval.
What goes wrong
- Writing to the CRM without a person checking it.
- Overwriting fields that already hold information.
Run it with the Call follow-up agent
04
Make sure no deal depends on one person
Also called Stakeholder map and multithreading
For each deal, map the people involved by role, influence and attitude, show who is missing, and plan who to reach next and with what message. A deal with one contact is a deal at risk.
Why it matters. When your only contact goes quiet, changes jobs or loses the internal debate, the deal stops.
How to do it
- List everyone you have met or emailed, from the CRM, calendar and transcripts.
- Mark the missing roles: budget holder, champion, possible blocker, procurement.
- Plan the next contact per missing role and the angle for each.
What goes wrong
- Counting attendees on a call as relationships. Contact means a two-way conversation.
05
Review pipeline honestly every week
Also called Pipeline, deal and forecast reviews
Use prompted reviews over CRM data: which deals are aging, which close dates keep slipping, which deals rely on one person and what changed since last week. The forecast becomes a short, honest story instead of a guess.
Why it matters. Most forecast surprises were visible weeks earlier in the data. Nobody looked.
How to do it
- Check coverage, deal age, stage conversion and past-due close dates.
- Flag deals that went quiet, lost a champion or keep slipping.
- Write the forecast as committed, likely and possible, with what changed.
- Model what happens if the biggest deal slips.
What goes wrong
- Letting AI edit the CRM during a review. Reviews should be read-only.
06
Learn from every sales call
Also called Conversation intelligence
Pull every meeting transcript, sort meetings into sales, customer, product feedback and internal, link each one to its account, and ask questions across many calls at once to find objections, requests and competitor mentions.
Why it matters. Your calls hold the best market research you have. Most of it is never read twice.
How to do it
- Collect transcripts from your note-taker.
- Sort meetings with clear rules first and AI only for unclear cases.
- Link each meeting to its company.
- Run themed questions across calls: objections, feature requests, competitors, pricing.
- Quote buyers word for word, with who said it.
What goes wrong
- Presenting a summary as a direct quote.
- Recording without consent.
07
Coach from evidence, not impressions
Also called Call scoring and sales coaching
Score each rep’s calls every week on the same set of skills, remove duplicate calls so counts are right, and send managers a short report so coaching starts from real examples.
Why it matters. Managers cannot listen to every call. A consistent score shows who needs help with what.
How to do it
- Fetch the week’s calls from your CRM and dialer.
- Remove duplicates so totals match the dialer.
- Score each call on a fixed rubric, such as opening, questions, listening and next step.
- Send a weekly summary with examples to each manager.
What goes wrong
- Using scores to rank people instead of to coach them.
Questions people ask
What should a discovery call prep brief include?
Who the company is, what changed recently, who else is involved in the decision, your past history with them, three guesses about their priorities, five questions and a list of anything you could not verify. Keep it to one screen.
What is SPICED?
SPICED is a framework from Winning by Design for understanding a buyer: Situation, Pain, Impact, Critical event (a deadline) and Decision process. It works well for writing up discovery calls.
Can AI update my CRM after a sales call?
Yes, and it should propose the changes first. A good setup shows the current and proposed values side by side and saves only after the rep approves.