A Claude skill that sits alongside a seller as a living intelligence layer on the accounts they are pursuing, fielding, and trying to close. It reads each account through the bad CRM data instead of trusting it, surfaces the few signals that should change how the seller works the deal, and points at the next move. The same layer rolls up for a manager into a read across the whole patch: where the team is strong, where it is exposed, and where management support would change an outcome.
The seller is the main user. They are the one who actions into the account. The manager view is the wider lens on top, not a way to catch a seller without the latest news.
For the seller, on the deals they own:
- Builds the context on an account before they work it: what is really happening, what changed since last time, and the one move most likely to advance the deal.
- Pulls three classes of signal: account signals from usage and product data, personnel and business-news flags from outside the CRM, and pipeline-health alerts from the shape of the deal.
- Reads every signal through the data, weighting what the buyer actually did over what got typed into a record months ago.
- Checks what is happening on the account against what the business is clearly better at, so the brief points at where the account's movement meets a real differentiator.
- Catches the silent deal-killers, like a champion who just changed jobs or an account whose usage cratered, the week they happen rather than the month the deal slips.
- Names a recognized, openable source on every signal, so a flag is one step from its evidence and built on something a buyer will not argue with: the product's own usage data, the call platforms, LinkedIn, company filings and earnings releases, and the research firms your industry trusts (Gartner, Nielsen, Newzoo, Coresight, and the like, named for your vertical at setup).
For the manager, across the patch:
- Rolls the same signal up into a read on the whole book: where the team is strong, where the number is exposed, and which deals need help now.
- Points at where a management move matters, an exec-to-exec touch, air cover, a resource, so support is aimed instead of generic.
- Hands a deal into coaching as a pointer into
deal-action-planning, naming the gap to work instead of supplying the answer.
For both, over time:
- Sharpens with use. Each cycle it folds back what was right and what was noise, so the layer gets smarter instead of repeating itself.
Most sales orgs do some version of this already. The problem is how. It runs once a quarter tied to planning. Its quality swings on who does it. It often lands with someone in revops or a support seat who is not close to the accounts. It misses the reality of the business because the CRM is half-entered by hand. So it becomes shelfware, or worse, a single point-in-time slice used for bottoms-up forecasting.
This skill is the version that was missing. Something that sits alongside the person working the deal, adjusts as they go, and gets better over time. Something that leads to action, rather than getting used superficially once a period.
A seller uses it on their own accounts. Build the context on a deal before a buyer meeting. Deep-read an account that has gone quiet. Drop in a signal they just heard and get back whether it changes the play. The brief becomes the thing they bring to a deal call already knowing where their account stands and what they want to do about it.
A manager uses the same layer one level up. The patch view shows where the book is strong, where the number is exposed, and which deals need a management move this week. It turns a one-on-one or a deal call into aimed support rather than a status check, because the seller and the manager are reading the same picture.
Behind it is a running account watch file the skill reads before each brief and updates after. That file is how it remembers what is true on each account, what has been acted on, and whose CRM record to trust. The brief is a short slice of that file aimed at one moment. The file is the whole picture, carried forward. A seller keeps it on their book, and it rolls up to the manager's patch.
It is built to run with deal-action-planning. This skill builds the intelligence on the account. That skill runs the coaching conversation. Each one works alone, and they share the same grid idiom so a signal here lands cleanly as a coaching item there.
Observed problem, trial solution. The problem is one I lived. Across previous sales, sales support, and sales leadership roles I ran account and deal intelligence, and I still watch teams I work with hit the same limits. The solution is what we aspired to and never had the resources to build. This skill is that aspiration turned into a tool, in trial. Run it, and tell me where it bends.
The pattern is everywhere. Teams do this by hand. They do it at the most macro level, a book-wide pull that never gets specific enough to act on. Or they have outputs and misuse them, feeding a frozen slice into bottoms-up forecasting, the one job it is worst at. It is rarely usable. The work happens, the artifact exists, and almost no seller does anything different on Monday because of it.
We wanted something usable. A more actionable way to leverage account research, one that stays tied to what the business is clearly better at. Account intelligence read in a vacuum lists facts about the buyer. Account intelligence read against a real differentiator turns those facts into a move, because the signal that matters is the one that lines up with a strength you can sell on. That is the thread running through this skill. It checks what is happening on the account against the differentiators you name at setup, and surfaces the match as the place to coach.
The layer also has to sit next to the team and stay current, adjusting as the quarter moves and getting sharper the longer it runs. The signal classes and the health checks are borrowed and adapted from the qualification and revenue-intelligence schools, because they already do the diagnostic part well:
- MEDDPICC for the deal-gap structure the flags route into.
- The Gong and Clari revenue-intelligence school for reading buyer-side behavior over CRM self-report.
- Account-planning practice (the discipline behind a real account plan, not the annual binder) for the account-signal classes.
When I ran these businesses, the account intelligence we produced was an event, not a habit. It came alive around planning season. Someone would pull a book-wide read, often a person a layer removed from the accounts, and the quality tracked exactly how close that person was to the deals. The output looked authoritative. A clean export reads as truth. Underneath, it inherited every optimistic stage and stale close date in the CRM, so the tidy dashboard was quietly laundering deals nobody had actually worked.
The failure that taught me the most was the silent kind. A champion would change roles and the deal would keep showing green for weeks, because the record had no way to know the person who carried it was gone. By the time it surfaced as a slip, we had lost the month we needed to rebuild the relationship. The signal existed the day it happened. It just lived outside the CRM, in a place no quarterly pull was looking.
The adjustment I always wanted was to stop treating intelligence as a planning artifact and start running it as a layer that sits alongside the team. Read the book through the data rather than around it. Weight what the buyer did over what the rep typed. Check each signal against what we are clearly better at, so the brief points at where the account's movement meets a real differentiator, which is the only intelligence a seller can act on. Catch the personnel and account signals the week they move. Surface only the two or three places attention pays off, and hand each one to the coaching room as a question, not an answer. Then close the loop, so the read that was wrong last cycle is the read that is sharper this one.
That is the skill. It is the upgrade to a thing every sales org already does badly, built so the intelligence guides sellers instead of sitting on a shelf.
Each moment has a thing to say and a thing you get back. This is the whole map of when, why, and how.
If you are a seller, working your own accounts:
| When | Why | Say | You get |
|---|---|---|---|
| You are starting out | The layer needs to know your accounts and what you are clearly better at before it can read anything | "Set up the brief for my accounts." |
A watch file and a list of your differentiators, set up once |
| You are about to work a deal | You want the real picture and the next move before a buyer meeting | "Build me the brief on the Acme deal." |
What is really happening, what changed, and the move most likely to advance it |
| A deal has gone quiet | You want the honest read on one account, not the optimistic CRM version | "Deep-read the Acme account." |
The real stage, what is actually true, the darkest gap |
| You just heard something | A champion moved, they raised, usage jumped, and you want to know if it changes the play | "Their champion just left, does this change anything?" |
A fast read and a call to act now or hold it |
If you are a manager, across the patch:
| When | Why | Say | You get |
|---|---|---|---|
| Before a one-on-one or deal call | You want to see the whole book and where your help matters | "Build the patch brief for this week." |
A ranked, short read on where the team is strong, exposed, and where to step in |
| A deal needs a management move | You want to know where an exec touch or air cover changes the outcome | "Where do I need to lean in this week?" |
The few deals where a management move pays off, and what the move is |
| A cycle wraps | You want the layer to learn from what actually happened | "Update the watch, here's how the deals went." |
The loop tunes: the brief gets sharper next cycle |
This skill builds the intelligence on the account. Its sibling, deal-action-planning, runs the coaching conversation once you are in the room. Use this one to get ready, that one during.
A tool that promises to be usable has to prove it does not end up on a shelf. This one keeps its own score in the watch file, and you should hold it to these:
- How often a flag led to action. Of the flags the brief raises, the share that turned into a real next step in the deal call. If this is low, the brief is flagging the wrong things, and the layer is built to correct it.
- Slips we saw coming. Of the deals that slipped or were lost, the share the brief had flagged ahead of time. This is the one that matters most. When it climbs, the layer is catching the quiet deal-killers before they cost you the month.
- How many flags, kept small. Fewer, better flags beat a long list. A brief that flags everything flags nothing.
The brief gets better over time on purpose. Each cycle it works out which kinds of signal actually paid off for your team, then raises those sooner and holds back the noise. That feedback is the difference between a layer that learns and a report that repeats.
It is a Claude skill. Drop the folder where Claude loads skills, and keep one running account watch file per book of business. For signal, connect whatever you have: a CRM for structure (Salesforce, HubSpot), product or usage data for the account signals, call transcripts (Gong, Granola, Fathom), news and email for the personnel flags, and the research firms your vertical treats as authority (Gartner and Forrester in tech, Nielsen and Comscore in media, Newzoo in gaming, Coresight in retail, named at setup). The skill runs on pasted input too, so a missing connector only adds a step, it does not break the brief.
No product-usage feed? The layer still runs strong. Two of the three signal classes need no telemetry at all: personnel and business-news flags come from news, email, and LinkedIn, and account signals come from call transcripts and a signal capture inbox where reps drop what they hear in the field. A team that captures field signal deliberately gets most of the value with nothing but a CRM and a habit. The setup walks you through it.
- MEDDPICC and MEDDIC qualification methodology (Jack Napoli, Dick Dunkel)
- Gong and Clari, revenue-intelligence and buyer-signal practice
- Strategic account-planning practice (the working discipline, not the annual binder)
Open an issue or start a discussion. If you run this and the signal taxonomy misses a class that matters in your motion, or the brief over-flags, tell me. That is the most useful thing you can send.