Choreography: account and ICP research in minutes
Account research that used to eat a whole afternoon becomes minutes: AI puts together the prospect's dossier, and you read, prioritize, and find the hook. The caution is checking the fact before citing it to the customer.
You have a call tomorrow at 10am with a prospect you barely know. Tonight, you open the company's site, read the "about us" page, hunt for news on Google, try to find who the decision-maker is on LinkedIn, and take notes in a notebook. An hour and a half later you have half a dozen scattered notes and still don't know what the hook is. You walk into the call generically, asking "tell me a bit about your operation" like someone who hasn't done the homework. With AI putting that dossier together in minutes, there's time left to do what only you can do: choose the right angle and check the fact before citing it.
Tomorrow you're presenting a credit proposal to a company that requested a quote yesterday. Tonight you spend the evening gathering the estimated revenue, the sector, whether there's been recent fundraising, who signs off on the decision. Two hours later, scattered notes and no clear angle. You walk into the meeting without knowing which financial pain that company has that your product solves best. With AI putting that dossier together in minutes, there's time left to read the numbers calmly and confirm the fundraising figure against a reliable source before citing it, because a made-up revenue number in front of the client costs the whole proposal its credibility.
Tomorrow you have a meeting with a potential corporate client. Today you scan the site, the business registry, news of lawsuits, expansion, ownership changes, to understand where your firm fits in. It takes the whole afternoon. You show up with a vague picture and a generic question, instead of showing up knowing which legal risk of that company is your hook. With AI putting that dossier together in minutes, there's time left to read the ownership history calmly and confirm any lawsuit with the business registry before mentioning it, because citing a lawsuit that doesn't exist in front of the client burns the firm's credibility in seconds.
Tomorrow you're pitching a brand that wants to outsource content. Today you open their Instagram, the site, look at competitors, try to understand positioning and who decides marketing there. An afternoon disappears. You walk into the meeting with a loose impression, with no hook like "I saw you launched X and channel Y has been stalled". With AI putting that dossier together in minutes, there's time left to choose the right hook and confirm the launch date before citing it, because getting the launched product's name wrong in front of a marketing client sounds amateur.
Tomorrow you have a final interview with a senior candidate who could switch careers for your job opening. Tonight, you open their LinkedIn, read the history, dig into where they've worked, try to understand what they're looking for. An hour later you have scattered notes and no clear angle. You walk into the conversation generically, asking "tell me about your journey", instead of showing up knowing which career pain your job opening solves best for them. With AI putting the professional dossier together in minutes, there's time left to choose the right question and confirm their current title and company before citing it, because getting their last employer's name wrong in the final interview sounds like you never read the résumé.
Tomorrow you have a discovery call with a key customer to understand if that roadmap feature makes sense for them. Today you scan their product usage, the tickets they've opened, their segment, trying to find where it hurts. Two hours later, scattered notes and no clear hypothesis. You walk into the conversation without knowing which of that customer's problems your discovery should attack. With AI putting the usage dossier together in minutes, there's time left to choose the right hypothesis and confirm in the dashboard the ticket AI cited before bringing it to the call, because citing a ticket the customer never opened breaks trust right at the start.
Tomorrow you have a prospecting call with an account that entered the pipeline yesterday. Tonight, you open the site, check the size, hunt for news, try to find who signs off on the purchase on LinkedIn. An hour and a half later you have half a dozen notes and still don't know what the hook is for the proposal. You walk into the call generically, asking "tell me a bit about your operation", like someone who hasn't done the homework.
Tomorrow you have a meeting with a new supplier who could take over part of your logistics. Today you gather their SLA history, capacity, whether there's been a quality problem, who decides on their end. Two hours later, scattered notes and no clear angle. You walk into the conversation without knowing which of your operation's bottlenecks that supplier solves best. With AI putting the supplier's dossier together in minutes, there's time left to choose the right angle and confirm the quality history against a source before citing it, because accusing the supplier of a problem they never had starts the negotiation on the wrong foot.
Tomorrow you have a meeting with an area about to go through internal audit. Today you scan its controls, incident history, applicable policies, where the risk lives. It takes the whole afternoon. You show up with a vague picture and a generic question, instead of showing up knowing which control of that area is your entry point. With AI putting the risk dossier together in minutes, there's time left to choose the right control and confirm the incident history against the original record before citing it, because attributing an incident that never happened to the wrong area during an audit is the kind of mistake that turns into a formal complaint.
Tomorrow you have a meeting with a team that wants to adopt your platform service. Today you scan their stack, repositories, how they deploy, where the architecture strains. An afternoon disappears. You walk into the conversation with a loose impression, with no hook like "I saw your deploy is still manual and the build queue has been stalling". With AI putting the technical dossier together in minutes, there's time left to choose the right hook and confirm the stack detail in the public repository itself before citing it, because claiming the team uses a tool they already swapped out last month sounds like you never did your homework.
Tomorrow you have a session with a product team that wants to rethink the onboarding flow. Today you open the current prototype, look at abandonment metrics, hunt for where the user gets stuck in the journey. An hour and a half later you have scattered notes and no clear angle. You walk into the conversation generically, asking "tell me how you think about the flow", instead of showing up knowing which stage of the journey is your hook. With AI putting the journey dossier together in minutes, there's time left to choose the right stage and confirm the abandonment metric against the original source before citing it, because pointing out the wrong drop-off number in front of the product team tanks your credibility right at the start.
Tomorrow you're presenting a recommendation to the board on entering an adjacent market. Today you comb through analyst reports, competitor-movement news, try to estimate the market size by hand. Two hours later, scattered notes and no number that survives a tough question. You walk into the room with a weak estimate, defending a market sizing that can't withstand the CFO's first challenge. With AI putting the competitive dossier together in minutes, there's time left to choose the right thesis and check every market number against a source before bringing it to the board, because an inflated TAM that AI made up becomes the first shot that brings down your entire recommendation.
Think with me: have you ever walked into a sales call knowing you should have researched the company more, but account research eats time you don't have, so you winged it? It's a feeling you can relate to, isn't it? The customer notices when you haven't done the homework. The good news is that afternoon of digging becomes a few minutes. The part you don't outsource, and that I treat seriously here, is judgment: what matters for this sale and what the hook is. That stays yours.
The core idea of this lesson. Researching an account before approaching it is a two-role choreography. AI does the manual labor: it assembles the account dossier in minutes (who the company is, the sector, recent signals, likely pains, who decides) and summarizes what used to take an afternoon. You do the head-work: you read, prioritize what's relevant to this sale, and find the hook (the credible reason to talk now). AI gathers; you judge. And there's a caution this lesson treats seriously: AI can invent a "fact" about the client's company (a news item that doesn't exist, a wrong number), and citing that in your approach burns your credibility instantly. Verify before mentioning.
You've already seen AI connecting sources in lesson 2.3 (MCP). Here that connection becomes a sales choreography: many sources go in, a dossier comes out, and you find the angle. Fair enough?
01Account research is manual labor disguised as strategy
Researching an account sounds noble, but most of the time it's digging. Opening tabs, reading "about us", hunting for news, piecing together the puzzle of who the company is. It's the boring part that precedes the part that matters: deciding the approach's angle.
The problem is economic. Every well-researched account costs one to two hours. Across a portfolio of twenty prospects, that's a week of work just digging, before a single conversation. So what happens in practice? You research the three big accounts and wing it on the other seventeen. Account research becomes a luxury, and most of your calls start out generic.
It's exactly that digging, the gathering and summarizing, that AI does well. Not the judgment. The digging.
02The choreography: many sources go in, a dossier comes out, you find the hook
The work split is the heart of this lesson. AI has fast hands for gathering; you have the judgment to decide. Look at the flow:
Notice where the line falls. Everything to the left of the dossier is digging: AI gathers the sources and summarizes. Everything to the right is yours: reading, prioritizing what matters for this sale, and finding the hook. AI doesn't know which of that company's pains your product solves best, nor which recent news is the credible reason you're calling today. That's a salesperson's reading. It hands you the picture; you choose where to enter.
02What goes into the dossier and what becomes a hook
A good account dossier has five pieces, and you can ask for all of them at once:
- Who the company is: what it does, size, market it's in.
- The sector: how it's doing, what pressures and trends surround whoever's in it.
- Recent signals: launched a product, opened a role, raised funding, expanded, changed leadership.
- Likely pains: given the sector and size, what usually hurts there.
- Who decides: the role that signs off on buying what you sell.
AI gathers this in minutes and hands it to you summarized. Then your work begins. You read and ask: of these pieces, which one connects with what I sell? A recent signal is usually the best hook, because it gives a credible reason to talk now. "I saw you opened a new operation in the Northeast" is a conversation starter; "tell me about the company" isn't. The hook is the piece that turns research into approach, and that choice is yours, not AI's.
Learn more: why the hook is yours and not AI's
AI gives you ten facts about the account. It doesn't know which one is the hook, because the hook depends on two things that live in your head: what you sell and the moment of the relationship. The same fact (the company opened a tech role) is a golden hook for someone selling software and noise for someone selling fleet insurance. Asking AI to "suggest the hook" helps open options, but the final decision is your reading, because you know your offer and your pipeline. Use AI so you don't miss a signal; use your judgment to choose which signal drives the conversation forward. Whoever outsources that choice hands over calls that sound exactly like everyone else's.
03The caution: AI can invent a fact about the client's company
Here's the part I treat seriously, because the damage is immediate. AI, when assembling the dossier, can invent. A news item that never ran. A revenue figure it estimated and presented as certain. A funding round that never happened. To it, all of this comes out looking exactly like fact.
The problem isn't the error in the document. It's you citing that error in front of the customer. Imagine opening the call with "congratulations on the twenty-million-dollar round you raised" and the decision-maker replying, dryly, "we didn't raise any round." Done. In the first thirty seconds you've proven you don't know what you're talking about, and the conversation died. The invented data didn't leave you unprepared; it left you worse than unprepared, because it looked like preparation.
The rule is simple and cheap. Any specific fact about the customer you're going to cite (a number, a news item, an event, a name), verify it against a source before taking it into the call. The dossier is your starting point, not your certainty. For most facts, verifying takes seconds: the company's site, the original news item, the LinkedIn profile. If you can't find the source, don't cite it as fact. Turn it into a question: instead of stating the funding round, ask "I saw a movement toward expansion at your company, is that right?". The question protects you and still opens the conversation. This connects with lesson N.vnd.7, on auditing: the discipline of checking before using is the same, here applied to your credibility in front of the customer.
04The frame: research shrinks from hours to minutes, and you show up knowing
Look at what changes in your day. Before, properly researching an account cost one to two hours, so you'd only research the big accounts. Now the dossier comes out in minutes, and that changes the whole economic equation: you can show up prepared for every account, not just the three that justified the afternoon of digging.
The gain isn't just time. It's the quality of the opening. You walk into the call knowing who the company is, what's been stirring over there, and what your hook is. Instead of spending the first ten minutes extracting basic context from the customer, you use that time on what matters: their pain and your solution. The customer feels you've done the homework, and that changes the tone of the conversation right from the start.
What doesn't change: AI doesn't sell for you. It hands you the dossier and saves you the digging time. Reading what matters, choosing the hook, and checking the fact before citing it remain your work. AI set the table; you're the one who drives the conversation.
Do it now
Take your real task (a real account or prospect you'll be approaching soon) and run the whole choreography:
- AI gathers: ask for the dossier in five pieces (who the company is, the sector, recent signals, likely pains, who decides). Also ask it to list the sources for each fact.
- You read and prioritize: mark, in the dossier, what actually matters for this sale. What connects with what you sell?
- You find the hook: choose a credible reason to talk now, ideally a recent signal. Write the first sentence of the approach.
- You verify: take the specific fact you're going to cite (the number, the news item, the event) and confirm it against a source. If you can't find it, don't cite it: turn it into a question.
You come out with a personalized, verified approach, in the time that used to only get you through three tabs. That's the rhythm of someone who shows up to the call already knowing the customer.
Practice
1. In the AI-powered account-research choreography, what's the division of labor?
2. AI, while building the dossier, states the client's company raised a funding round. What should you do before citing this on the call?
3. What's the economic frame that justifies using AI for account research?
For the board
On what the research ismanual work disguised as strategy. The digging is the AI's, the hook is yours.
On the invented factany specific claim about the company gets checked before you cite it. With no source, it becomes a question instead of a statement.
On the maths that changedwhen research cost an afternoon, only the big account was worth it. In minutes, every prospect gets a prepared call.
Thanks for the feedback. It helps sharpen the next lesson.