Summit53 skills for Claude and Codex

The question

I have to walk into the forecast call, a deal review or a board meeting. Can I get the prep done by asking for it?

Each recurring revenue meeting needs the same kind of prep every time: the right numbers, the right window, the evidence behind them and the limits. The Summit53 skills each do one of those jobs, end to end, and stop at the edge of what the data supports.

What the skills are

The Summit53 plugin (version 1.1.3) adds a set of skills to your AI assistant and connects it to your Summit53 workspace. It works in Claude (claude.ai, Claude Desktop, Cowork and Claude Code) and in Codex. Each skill is a guided workflow: you ask in plain language, it reads your Summit53 data through the connector and produces one finished thing, such as a one-slide brief, a deck or a one-page sheet.

To install the plugin and sign in, see the MCP setup guide. If you’re not sure which skill fits, ask What can Summit53 do? or Which Summit53 skill should I use?: the summit53 skill checks your connection, tells you which organisation you’re signed in to, and routes you to the right one.

How every skill behaves

• It checks whose data first. Before reading anything, it confirms which Summit53 organisation it’s connected to, and stops if that can’t be confirmed.
• Missing is shown as missing. Evidence it can’t find is reported as unavailable or not yet measured, never as zero.
• Causal words are earned. Deals and signals are described as consistent with a movement, not as its cause.
• Demo data is labelled. On a demo workspace, every output carries a label such as Demo data — not for customer or commercial decisions (or, for coaching and HR outputs, not for real coaching or not for employment decisions).
• It says what it changed. Most skills only read. Each entry below says whether that skill saves anything.

Find the right skill

summit53: am I connected, and which skill do I need?

The job: the starting point. It checks your connection, tells you which Summit53 organisation you’re working in, and hands your request to the right skill.

Ask it like this

• Which Summit53 organisation am I connected to?
• Set up Summit53 for me — is it connected?
• What can Summit53 do?, or which Summit53 skill should I use?
• Any revenue question that mentions Summit53 without naming a workflow

What it reads

• Your connection and the organisation, user and role it’s signed in as.

What you get

• A one-line connection check: the organisation, your user and that the connection is live, plus a note when it’s a demo workspace.
• A route to the matching skill (see Find the right skill). If two skills fit, it asks one short question naming both; if none fits, it says what the skills cover and answers from the connector directly.
• Connection steps for your assistant when you’re not connected: in Claude Code, run /mcp, select summit53 and choose Authenticate; in claude.ai, Claude Desktop or Cowork, connect Summit53 on the plugin’s Connectorstab; in Codex, add and sign in to the connector. Then start a new conversation.
• It saves nothing.

What it won’t do

• Produce a brief or deck of its own; it hands that to the matching skill.
• Answer a Summit53 question from general knowledge when the connector isn’t attached.

When the data is thin

• Not signed in. It asks you to sign in again rather than retrying.
• Membership not approved. It tells you a Summit53 admin needs to approve you.
• Two connectors. Two for the same organisation is harmless and it uses one; two for different organisations, and it asks which you mean. It never merges results across them.

Installation and sign-in for each assistant are in the MCP setup guide.

forecast-delta-brief: why is the number different from last week?

The job: the weekly forecast or commit call, where the hard question isn’t “what’s the number?” but “why is it different from what you told me last week?”

Ask it like this

• Why is the forecast number different from last week? Prep me for the commit call.
• What moved the forecast this week?
• Forecast confidence dropped — why?
• Weekly forecast delta, or commit call brief

What it reads

• Your organisation, to confirm whose data it is.
• Forecast confidence today and exactly seven days ago, and how much each part of it moved: pipeline shape, execution health, data quality, account and renewal health, and forecast accuracy.
• Today’s forecast levels (projected landing, commit, coverage, attainment, revenue at risk) for the whole organisation, unless you ask for your own team.
• Your riskiest open deals, and the first thing blocking up to three of them.

What you get

• One 16:9 slide (PowerPoint), read top to bottom: the window (previous date → current date); one line you can say in the room (level, signed movement, the main contributor); a chart of what moved, adding up to the total or showing the remainder; up to three deals consistent with the movement, or a statement that no single deal explains it; today’s levels; and the limits. Sources go in the speaker notes.
• A five-line summary in the chat, readable on a phone on the way into the room.
• It saves nothing. It doesn’t write to your CRM, send anything or force a recalculation.

Where your assistant supports scheduled tasks, you can run it every Monday morning before the call.

What it won’t do

• Daily, monthly or “since the board meeting” windows: it says so and offers the weekly brief instead of quietly changing the window.
• A board or investor pack (use board-revenue-pack), an internal quarterly review (qbr-generator) or a renewal outlook (cs-renewal-forecast).
• Rank reps, or show week-on-week changes in individual deal health scores.

When the data is thin

• Whose data and which window. It stops if the organisation in the answer doesn’t match the one you’re signed in to, or if today’s reading isn’t from today. If there’s no reading from seven days ago, you get a short readiness note instead of an invented movement.
• Does it add up? The parts are shown summing to the headline; any part of the move that no measured driver explains is shown as a remainder.
• What it can’t see. A driver Summit53 doesn’t measure yet is listed as not yet measured, never plotted as zero. Last week’s levels aren’t available from the same source, so those cells are left blank and the slide says why.
• A narrower view. If you can only see your own team, the slide labels the levels as a team view while confidence and movement stay organisation-wide.

By hand, this is a Monday diff of two exports. The method is in Revenue Engine masthead: will we make the number?

board-revenue-pack: is it a market problem or an execution problem?

The job: the quarterly board or investor meeting, where two questions decide the room: “you told us last quarter this would be fixed — what’s different now?” and “is this a market problem or an execution problem?”

Ask it like this

• I have a board meeting in three weeks. Build my board revenue pack for this quarter.
• Is our revenue problem a market problem or an execution problem? The board will ask.
• Prep the numbers for the board, investor update or quarterly revenue deck for investors

What it reads

• Your organisation, to confirm whose data it is.
• How forecast confidence has moved since the prior quarter, and which parts of it account for the move.
• ARR, net and gross revenue retention, logo churn and time to value, with reported and constant-currency growth where you sell in more than one currency.
• Performance by segment (win rate with its sample, coverage, sales cycle, deal size), and by region where your workspace has regions set up.
• This quarter’s and last quarter’s forecast levels, the renewal book, account health trends, data quality and how much history the forecast model rests on.

What you get

• An eight-slide deck (PowerPoint, nine with an optional regional slide), ordered by the board’s questions: what changed since last quarter and what accounts for it; ARR and retention; market or execution, with one sentence saying which way the evidence leans and how strongly; pipeline and forecast; renewals and customer health; and What we cannot yet answer, a slide of its own rather than an appendix.
• Segment-, band- and portfolio-level only: no rep, deal or account names.
• It saves nothing. It doesn’t write to your CRM, send anything or force a recalculation.

What it won’t do

• The weekly forecast call (use forecast-delta-brief), an internal quarterly review (qbr-generator) or a customer-facing review (customer-qbr).
• Compare against last quarter’s deck. Summit53 doesn’t store past packs, so it compares last quarter’s data, not what was claimed, and tells you so when it delivers.
• Fall back to a weekly or daily movement when there isn’t a quarter of history.

When the data is thin

• Which window. It needs a genuine prior-quarter comparison (75–105 days of history). Without one it stops before building any slides and tells you how much history there is and what it needs.
• Like for like. If last quarter’s closed deals or pipeline are a small fraction of this quarter’s, it says the quarters aren’t comparable instead of presenting the change as growth.
• What it can’t see. Segment-level retention reads not yet computable, never 0%; ramp figures carry a note that they leave out reps who left during ramp, and no median is shown for fewer than three reps; estimated net retention carries its limitations.
• Currency. Exchange-rate movement is shown as its own line, never as market or execution performance.
• Demo data. On a demo workspace every slide reads Demo data — not for customer or commercial decisions, and demo history never counts as a real prior quarter.

See Board pack and quarterly review.

qbr-generator: how did the quarter go, and what do we do next?

The job: the internal quarterly business review with your leadership team: one headline story, the honest risks, quarter-on-quarter movement and the few decisions for next quarter.

Ask it like this

• Prep for my QBR — pull together this quarter’s numbers into our internal quarterly revenue review.
• Build our quarterly business review deck for the exec team.
• Quarterly sales review, executive revenue review or build my quarterly deck

What it reads

• Your organisation, to confirm whose data it is.
• This quarter’s and last quarter’s forecast, quota, coverage, win rate and revenue per rep, with the projected landing and its basis.
• ARR, net and gross retention, logo churn, time to value and, where you’re allowed to see it, renewals coming due.
• Pipeline risk, stalled deals, quiet accounts, account health and its movement, and your biggest wins and losses with their notes and qualification evidence.
• Team action-plan completion and your own weekly plan, and, where available, regions and the first failing stage of the revenue loop.

What you get

• An executive deck (PowerPoint, about 12 slides, more when optional slides apply): executive summary, revenue performance, ARR growth and retention, pipeline health, risk and concentration, account portfolio, engagement gaps, key wins and losses, next quarter’s plan, and a Sources & Limits slide. Optional slides (regions, value and expansion health, where deals break down) appear only when their data does. A Word document only if you ask for one.
• Every rate carries its denominator, and every source its as-of date.
• It doesn’t change your CRM. One exception: if you don’t have a weekly action plan yet, reading it creates this week’s plan for you.

What it won’t do

• A customer-facing QBR or value review (use customer-qbr; this deck is internal only).
• A board or investor pack (board-revenue-pack) or the weekly “why did the number move” brief (forecast-delta-brief).
• Rank reps on an executive deck, or invent a benchmark the data didn’t return.

When the data is thin

• Assumptions stated. The reporting period, comparison period, currency, audience and scope it used are listed on the executive summary.
• What it can’t see. Missing data reads unavailable with a reason, never zero or a green status; an empty risk list is indeterminate, not healthy.
• Access. Parts your role can’t see, such as the renewal forecast, are left out and listed as not included (access-gated), without error text.
• Causal words. Account health and execution movement are described as trends, not causes.
• Demo data. On a demo workspace every slide carries Demo data — not for customer or commercial decisions.

See Board pack and quarterly review.

customer-qbr: what value have we actually delivered to this customer?

The job: a customer QBR, executive business review or value review, where every claim has to stand up in front of the customer.

Ask it like this

• Prep a customer QBR for Acme — what value have we delivered this quarter?
• Prep the EBR with Globex.
• Customer-safe value review or mutual success plan for a named customer

What it reads

• Your organisation, to confirm whose data it is.
• The customer’s account and the right contract, and that contract’s agreed outcomes, measures, owners and target dates.
• Dated impact entries for the review period, and the account’s value summary, reconciled against those entries.

What you get

• A customer-safe brief (or a deck outline), in this order: purpose and period, shared objectives, verified progress (baseline, target, observed result and caveat), what the evidence shows, measurement gaps, next-period commitments and any decision needed. A PowerPoint deck or Word document only if you ask for one.
• Internal material is stripped out: health and risk scores, renewal probability, forecasts, rep performance, private notes and anything about other customers.
• It saves nothing. It doesn’t write, send or record outcomes.

What it won’t do

• An internal revenue QBR (use qbr-generator), a renewal-risk case or save plan (renewal-defence-pack), or a board pack (board-revenue-pack).
• Turn the review into a sales pitch, or credit Summit53 with a business result unless the evidence records it.

When the data is thin

• The claim gate. A value claim reaches the customer only if it matches the exact account and contract, links to an agreed outcome, has a dated impact entry in the period, states a value and unit (or a recorded status), and isn’t demo data.
• Not ready. If no claim passes, you get Not ready for customer delivery, the missing evidence by outcome, an internal checklist and, optionally, a claim-free draft agenda.
• No estimates. Missing baselines, targets, dates or values stay missing; summaries without supporting entries are context, not proof.
• Demo data. On a demo workspace every output starts with Demo data — not for customer or commercial decisions.

See Renewal review and Value & Impact, where the contracts and impact entries live.

revenue-action-brief-s53: what should I actually do this week?

The job: Monday morning for a sales leader or CRO, when forty things need attention and the question is which few, in what order.

Ask it like this

• What should I focus on today and this week across the revenue team?
• What needs my attention? or where should I spend my time?
• What do I do today? (a single horizon still checks the others, so nothing that has slipped into today is missed)

What it reads

• Your organisation, to confirm whose data it is.
• The forecast (for the whole organisation unless you ask for your team), coverage, revenue at risk, deal-size patterns and average deal size drift.
• Retention and churn, deal-level risk across your whole pipeline, stalled deals and quiet accounts.
• Rep confidence, team action-plan completion, your own open tasks and, if your plan includes outreach, target-account pipeline generation.

What you get

• A brief in the chat: a one-line posture, then Today, This week, This month and This quarter, each capped at a handful of items, plus Already behind for anything whose recovery window has closed. Each item names the action and its owner, the evidence behind it, and why it belongs in that horizon.
• A Markdown copy if you ask for one. It offers once to help act on an item, and hands a coaching item to rep-coaching-report only if you ask.
• It doesn’t change your CRM. One exception: if you don’t have a weekly action plan yet, reading it creates this week’s plan for you. It never drafts or sends messages.

What it won’t do

• Explain this week’s forecast movement (use forecast-delta-brief).
• Draft emails, write CRM records or create tasks.

When the data is thin

• What it can’t see. A source it can’t read is listed on a closing Not assessed line, and only the checks that depend on it are skipped. Silence from a missing source is never treated as health.
• Nothing pressing. It says No pressing actions detected only when every source answered; otherwise in the areas assessed.
• Whose data. If the forecast figures cover only your team or your own book, a line under the header says so.
• Demo data. On a demo workspace the brief carries Demo data — not for customer or commercial decisions.

See What should I do this week?

deal-review-prep: is this deal real, and what should I ask about it?

The job: a manager-led deal review or pipeline review of one opportunity, where the manager needs the evidence and the right questions before the rep walks in.

Ask it like this

• Review the Acme deal. How healthy is it and should I be worried?
• Give me the full picture on the Acme deal for my pipeline review with the team.
• Is this deal going to close? or what’s the risk on this deal?

What it reads

• Your organisation, then exactly one opportunity. If a name matches more than one deal, it lists them and asks you to pick.
• The deal record, its health score and band, risk factors, how long it has stalled, its notes and activity history.
• Qualification evidence against three frameworks, the deal’s first constraint and win-probability movement where you’re allowed to see them, and the buying committee and its gaps.
• Account health and its trend, other open deals at the account, and, where relevant, the value contract, a battlecard for a named competitor and committee coverage from outreach.

What you get

• A deal review deck (PowerPoint, about 11–12 slides): deal snapshot with an assessed verdict, narrative and recent activity, risk factors, a MEDDPIC scorecard, BANT and SPICED side by side, blockers with a question for each, account and stakeholder context, recommended next steps, and a Sources & Limits slide. A value roadmap appears for closed-won deals.
• Tool scores are labelled as such and kept apart from the reviewer’s assessment, which says which evidence it rests on.
• It saves nothing. For a deal going cold it can add an appendix headed Dark Deal Rescue (Draft — Not Sent); nothing in it is sent or executed.

What it won’t do

• An account executive’s own inspection prep (use ae-deal-inspection-prep) or a disputed chronology (deal-timeline-defence).
• A full competitor response (competitor-response-card); it only adds a short battlecard callout.
• Draft, send or log re-engagement emails or notes.

When the data is thin

• Whose data. It stops if the deal or account belongs to a different organisation from the one you’re signed in to.
• What it can’t see. No risk rows reads Risk factors unavailable, not “no risk”; an empty activity log is a logging fact, not proof the buyer has gone quiet; a missing framework is listed rather than averaged.
• Stale scores. A health score that’s out of date shows its age and carries less weight.
• Causal words. Signal, activity and account-health links are described as consistent with, never as causes.
• Demo data. On a demo workspace every slide carries Demo data — not for customer or commercial decisions.

See Is this Commit deal real?

ae-deal-inspection-prep: what will my manager ask about this deal?

The job: the account executive’s side of a deal inspection: knowing the facts, owning the gaps and having evidence ready for the questions before the 1:1.

Ask it like this

• Prep me for my deal inspection on the Acme deal. What will my manager ask?
• Help me answer questions on the Acme deal before my 1:1 with my manager.
• What am I missing before my 1:1? or give me an inspection cheat sheet

What it reads

• Your organisation, exactly one opportunity, and whether you own it.
• The deal record, its notes, its activity history and your published qualification framework’s evidence.
• Where available and you’re allowed to see them: the deal’s first constraint, its buying committee, committee coverage from outreach and up to three recorded calls.

What you get

• A one-page sheet in the chat: deal snapshot; what the evidence says; gaps to acknowledge; five to eight likely questions, each with the evidence you can cite and a line to say “this is unknown” where proof is missing; three to five next actions you own; and sources and limits. A one-page slide only if you ask.
• Every claim is marked Observed, Inferred, Disputed or Unavailable, with its source and date.
• It saves nothing. It doesn’t create notes, update fields or send messages.

What it won’t do

• A manager-led formal review deck (use deal-review-prep), even if the request mentions an account executive.
• A contested chronology or “when did it go dark” (deal-timeline-defence).
• Judge the account executive’s performance.

When the data is thin

• Whose deal. If you aren’t the owner, it asks whether you’re preparing on someone’s behalf, or routes a manager’s request to deal-review-prep.
• Missing is a gap, not a fact. No recorded call means “recorded call evidence unavailable”, never “no calls happened”. It never infers close-date history from today’s close date.
• Partial frameworks. When a framework component can’t be scored, the analysis is labelled partial, for example Partial — Competition component unavailable.
• Demo data. On a demo workspace every output carries Demo data — not for customer or commercial decisions.

See Is this Commit deal real?

deal-timeline-defence: what actually happened on this deal, and when?

The job: the moment a deal’s history is challenged: when it went dark, who owned it when it slipped, whether a rescue is warranted. The answer has to separate what’s proven from what’s assumed.

Ask it like this

• When did the Acme expansion deal go dark? Show me the evidence trail.
• Defend this timeline or what actually happened on this deal?
• Who owned this deal when it slipped? or should we start a dark-deal rescue?

What it reads

• Your organisation, then exactly one opportunity and its current record.
• The deal’s timestamped activity, account-level inactivity, and actions tracked against the deal or account.
• Where available and you’re allowed to see them: notes for context, buying-committee changes, recorded calls in the disputed window and a read-only rescue diagnosis.

What you get

• A defence brief in the chat: the claim being tested; a one-line verdict (supported, partly supported, unsupported or indeterminate); a timestamped chronology; what is proven, inferred and missing; whether a dark-deal rescue is warranted; the next evidence to get; and sources and limits.
• Every event is marked Observed, Corroborated, Inferred, Disputed or Unavailable.
• It saves nothing. It never writes a CRM note or reproduces a rescue draft.

What it won’t do

• A manager-led review deck (use deal-review-prep) or an account executive’s inspection prep (ae-deal-inspection-prep).
• Draft re-engagement outreach or a CRM note.

When the data is thin

• Whose data. It stops if the organisation or the deal record can’t be confirmed.
• No invented history. Close-date movement is reported only when a source shows both the old and new date; a seller-side handover is Unavailable unless a source shows it.
• Coverage. If the activity history doesn’t cover the disputed period, the chronology says it’s incomplete. A missing recorded call is never “no meeting happened”.
• Causal words. An action’s timing is never presented as the cause of what followed.
• Demo data. On a demo workspace every output carries Demo data — not for customer or commercial decisions.

See Is this Commit deal real? and the opportunity cockpit.

competitor-response-card: a competitor just showed up — what do I say?

The job: a named competitor appears late in an active deal, often as “they’re cheaper”, and the account executive needs a response for the next conversation that keeps what’s known apart from what’s suggested.

Ask it like this

• A competitor just showed up in the Acme deal and they’re saying they’re cheaper. How do I respond?
• How do I respond to <competitor> on this deal?
• Give me a competitive talk track

What it reads

• Your organisation, to confirm whose data it is.
• Your published battlecard for that competitor (an exact match only) and your team’s published plays for the objection, from Knowledge.
• The deal record, only to confirm it’s active and the right one.

What you get

• A one-screen card in the chat: situation; verified evidence (two to five sourced facts, with Unknown where there’s a gap); a suggested response, labelled as suggested; three to five discovery questions; proof to use; what to avoid; and one safe next step.
• No web research unless you ask for it, and then it’s cited separately from your own evidence.
• It saves nothing. It doesn’t send, log or draft emails.

What it won’t do

• Win and loss patterns across many deals, recurring loss reasons or what drives your win rate (use win-loss-analysis).
• Invent pricing, product comparisons, customer references or competitor weaknesses.

When the data is thin

• No exact battlecard. It never falls back to the nearest card; if several could match, it asks you to pick; with none, you get an evidence-gap card instead of competitive claims.
• What you told it. Deal facts you supply are labelled as user-provided, not verified.
• Demo data. On a demo workspace the card starts with Demo data — not for customer or commercial decisions.

See A competitor just showed up late and Blueprint & Knowledge, where battlecards and team plays are written.

win-loss-analysis: what separates our wins from our losses?

The job: a win/loss review, a sales offsite or a process change, where the team needs evidence about why deals close or don’t, rather than anecdotes.

Ask it like this

• Why are we losing deals? What separates our wins from our losses?
• What do our wins against competitors have in common?
• Help me understand our win rate, or how can we improve our close rate?
• Win/loss report for the last 12 months

What it reads

• The closed deals in the window you choose (the last 12 months by default), won and lost, with what counts as each written down before anything is calculated.
• Deal health and qualification scores at close, and framework scores for a sample of the largest won and lost deals.
• Deal notes and recorded loss reasons, and your battlecards for any competitor named in lost-deal notes.
• Logged activity and buying-committee coverage on the sampled deals.
• Your win rate by deal-size band, and, where your organisation has them, win rates by region and ICP fit.

What you get

• A deck of up to 12 slides: an executive summary; win rate overall and by deal-size band; a region and ICP-fit cut when available; framework scores for wins against losses; the patterns associated with wins; themes from deal notes with deal counts; activity patterns; the common loss patterns, split into deals lost during an active evaluation and deals that stalled; recommendations written as tests to run; and a closing slide with the cohort definition and limits.
• It saves nothing. It only reads.

What it won’t do

• Respond to a competitor in a live deal (use competitor-response-card).
• Coach an individual rep (use rep-coaching-report).
• Call a pattern a predictor. Everything it finds is an association in this group of deals until it holds on deals it wasn’t drawn from.

When the data is thin

• Every percentage carries its count, for example 71% (5 of 7). With fewer than 10 deals on either side, findings are labelled early observations; with fewer than 5, it says the pattern can’t be told apart from noise.
• Loss reasons are labelled as stated, inferred from notes or unknown. It never infers a reason from an empty field, and it reports how many are unknown.
• Framework comparisons carry a caveat when lost deals were scored at earlier stages than won ones.
• An incomplete deal list is disclosed on the slides rather than presented as the whole. If your deal-size bands can’t be read, it stops and says which source is missing.
• Missing optional views (regions, ICP fit, rep coaching layers) are dropped and listed in the limits, never filled with zeros.

For a competitor showing up in a deal right now, see A competitor just showed up late. Battlecards themselves are covered in Blueprint & Knowledge.

first-meeting-deck: what do I show in the first meeting?

The job: an account executive’s first or early meeting with a prospect, or with a new stakeholder at an existing customer, where a generic capabilities deck would cost credibility.

Ask it like this

• I’m meeting Acme tomorrow for a first discovery call — make me a deck.
• Prep a discovery deck for the Acme deal, or build slides for my call with Acme
• What should I show in the first meeting?
• Refresh vendor context or refresh vendor branding, to update what it knows about your own company

What it reads

• A short profile of your own company: positioning, real capabilities and limits, proof points, your current offer and your brand. It asks you to set this up once (it drafts it from your website and a few questions) and checks anything date-sensitive before it reaches a slide.
• One opportunity and its account, the deal’s written case for change and value narrative, and whether the account already has contracts (which switches it to an expansion meeting).
• The prospect’s own website language, your Knowledge case studies and positioning, and battlecards for any competitor named on the deal.
• Qualification gaps on the deal, for the account executive’s prep slide only.
• Optionally, target-account research if you use Outreach, and fresh news about the prospect if you say yes when it offers.

What you get

• A customer-facing deck, sized to the meeting (up to 6 slides for 30 minutes, the full 9 for an hour), written as hypotheses the customer is invited to challenge: Is this right? What would you challenge? It names comparison categories, not competitor brands.
• A separate prep slide for the account executive, marked as internal, with what to probe. Internal scores and ratings never appear on the customer-facing slides.
• It saves two things, and asks first. Its account research runs on Summit53’s servers and stores the results on the account. Your company profile is saved only after you approve it. It also runs a few public web searches about the prospect.

What it won’t do

• A value or success review with an existing customer (use customer-qbr).
• A one-screen answer to a named competitor in an active deal (use competitor-response-card).
• An internal, manager-led review of the deal (use deal-review-prep).
• Put a claim about your company on a slide that it couldn’t check against a current source.

When the data is thin

• One deal at a time. If a name matches several opportunities, it lists them and asks.
• No case for change, no deck. If the deal’s written case for change hasn’t been generated, it stops and tells you to create it first. If there’s no value narrative, it leaves out the value slide rather than showing a thin one.
• Unverified claims are dropped and listed in its delivery note as skipped.
• On a demo workspace, every slide says Demo data — not for customer or commercial decisions, and the note says the deck must not be shown to a real prospect.

The deck draws on what you’ve published in Knowledge; see Blueprint & Knowledge.

cs-renewal-forecast: what’s renewing, and how much is at risk?

The job: a customer success leader’s view of the renewal book for the forecast meeting, without dressing a snapshot up as a trend or an NRR forecast.

Ask it like this

• What’s renewing in the next 90 days across my book, and how much ARR is at risk?
• Renewal forecast for my book
• CS renewal outlook for the forecast meeting
• What will NRR be? (it answers with the renewal exposure and says which parts of NRR it can’t see)

What it reads

• The renewal book for a window of 30 to 180 days (90 by default): contract ARR due, contract ARR at risk, and the commit and best-case renewal opportunities.
• For up to five at-risk accounts: their health movement, contracts and value-impact assessment.

What you get

• A written outlook (Markdown unless you ask for another format): the headline exposure, the forecast buckets on their own basis, the top surfaced renewal risks, account-health movement, the NRR and history limits, and evidence questions for the forecast meeting.
• It saves nothing. It only reads.

What it won’t do

• Defend one named account’s renewal (use renewal-defence-pack).
• A customer-facing review (customer-qbr), weekly pipeline-forecast movement (forecast-delta-brief) or an internal quarterly review (qbr-generator).
• Split the renewal book by region: the renewal source has no region dimension today, so it says so rather than joining other views.

When the data is thin

• Who can run it. It needs a manager’s view of the book; without one it stops and says so.
• Two bases, kept apart. Contract ARR and renewal-opportunity buckets are labelled as different measures and never reconciled or subtracted.
• No invented trend. There’s no history of the renewal book to compare against, so it never says the book grew or shrank unless you give it an earlier, matching report.
• Missing history isn’t zero movement. Accounts without enough health history are shown as such.
• Not an NRR forecast. When you ask about NRR, the answer is labelled renewal exposure outlook — not an NRR forecast, with the missing components named.
• The at-risk list is the top five surfaced, not the whole at-risk population.

See Renewal review.

renewal-defence-pack: how do we defend this renewal?

The job: a procurement challenge, commercial review or renewal-risk meeting for one named customer, where every claim about value has to stand up.

Ask it like this

• Procurement is pushing back on Acme’s renewal. Help me defend it.
• Save plan for Acme — their renewal is at risk.
• Build a renewal defence for Acme, or renewal risk meeting prep for Acme

What it reads

• The account and the renewal contract you name (or the active one renewing next): ARR and renewal date.
• The outcomes promised when the deal was won, and the dated impact recorded since.
• The internal value-impact assessment, kept separate from the raw evidence.

What you get

• A written defence (Markdown, or a Word document if you ask): the defensible position, a promise-to-impact evidence table, up to five procurement-ready proof points with safe wording, the gaps that weaken the defence, and questions to resolve before the meeting. Ask for customer-ready wording and internal scores and labels are left out.
• It saves nothing. It never creates or edits contracts, value records, notes or tasks.

What it won’t do

• A portfolio renewal forecast or NRR outlook (use cs-renewal-forecast).
• A customer-facing business review (customer-qbr) or an internal quarterly review (qbr-generator).
• Invent baselines, savings, ROI or percentages, or say the product caused an outcome.

When the data is thin

• No contract, no defence. It stops and says which record is missing.
• No recorded promises, no value case. Without promised outcomes it gives a short readiness finding instead: what’s recorded, and what isn’t yet.
• Partial when impact is missing. With promises but no recorded impact, the pack is marked Partial — delivered-impact evidence missing.
• Only recorded links count. Impact is matched to a promise only when it was recorded against that promise; anything else is shown as context, not contractually matched.
• Estimated dates are labelled planning estimate — confirm before commercial use.

See Renewal review and Value & Impact, where promises and impact are recorded.

rep-coaching-report: what should I coach this rep on?

The job: a routine developmental 1:1, built around one thing worth working on rather than a tour of every metric.

Ask it like this

• Prep my 1:1 with Sam — what should I coach them on?
• Coaching report for Sam, or build a coaching deck
• What’s going on in Sam’s pipeline before our 1:1?
• Log my notes from the session with Sam (see below)

What it reads

• The rep’s coaching record: the stored diagnosis of whether the gap is in results, effort, knowledge or skills, the 1:1 agenda and last session’s actions, meeting analytics, development goals and coaching history.
• Where they are in ramp, their mix of new business and existing customers, and their sales region.
• Their pipeline, deal health, qualification scores, activity, deal drag and action plan.

What you get

• A coaching deck of about 10 slides built around one coaching thesis, kept as a hypothesis to test in the conversation. Last session’s commitments are carried forward, done or not. A rep who’s on target gets a short, honest deck.
• It saves nothing while preparing the deck. After the 1:1 it can turn your notes and agreed actions into a draft. The connector doesn’t save coaching sessions today, so it tells you nothing was saved and to record the session on the Coaching page.

What it won’t do

• Promotion cases, performance plans, calibration or HR documentation (use rep-evidence-dossier). It stops and routes before pulling any evidence.
• Rank reps, show a leaderboard or compare the rep with named colleagues.
• Patterns across many closed deals (win-loss-analysis) or a territory review (territory-review).

When the data is thin

• Whose data. You must be authorised to see this rep’s coaching record; it picks the rep from your coaching roster, not from a name search.
• Benchmarks are references, not averages. The team benchmark is labelled an aspirational reference. Benchmarks built on only a few peers are called out as unstable.
• Missing is not green. Empty or unavailable evidence is left out as indeterminate, never shown as “no drag” or “all engaged”.
• Ramp and region first. Ramp figures carry a caveat about who stays in the cohort, and the rep is compared within their own region or segment before any team-wide comparison.
• On a demo workspace, every slide says Demo data — not for real coaching.

See the rep deep-dive and 1:1 agenda builder guides.

rep-evidence-dossier: what does the record actually show?

The job: a promotion case, talent calibration, performance plan or documented performance concern, where the decision is made by a person and the evidence behind it has to be fair and complete.

Ask it like this

• We’re preparing for talent calibration. Pull together the performance record for Sam.
• We’re considering a PIP for Sam.
• Build the promotion case for Sam, or HR wants documentation on Sam
• What have we done to help this rep?

What it reads

• The rep’s documented coaching record: diagnostic history, development goals, completed coaching sessions and their notes, agendas and meetings analysed.
• The structural factors around the rep, checked first: ramp stage, region and team, segment and the fit between their role and the deals they work.
• Their pipeline, deal health, forecast and activity, each with a named comparator and window.

What you get

• A neutral evidence deck that puts structural factors and the depth of the documented record before any performance evidence, and closes with an evidence register listing every source, date and limit.
• No recommendation. It never says whether to promote, retain or manage out. That decision stays with the person accountable for it.
• It saves nothing. It never creates or backfills a coaching session, goal, note or outcome.

What it won’t do

• Routine coaching or 1:1 prep (use rep-coaching-report).
• Log a coaching session, notes, actions or outcomes.
• Rank the rep, rate them, or show another rep’s evidence beyond what the calibration needs.

When the data is thin

• A depth check comes first. With less than six weeks of diagnostic history, or no development goals, completed sessions or analysed meetings, every evidence slide is marked Provisional — insufficient documentary depth.
• No record, no case. If there’s no documented coaching record at all, it stops before any performance evidence and produces a short record-readiness deck saying the record needed to support a decision hasn’t been created yet. Coaching that happened outside Summit53 won’t appear and needs documenting first.
• Absence is a finding. A missing record is reported, never omitted, and never shown as zero.
• On a demo workspace, every slide says Demo data — not for employment decisions.

See the quarterly development and rep deep-dive guides, where the coaching record is kept.

territory-review: which region is underperforming, really?

The job: a territory or regional review for a CRO, VP or regional leader: where to double down, where to intervene, and which coverage questions to check before moving anyone.

Ask it like this

• How are my territories doing? Which regions are underperforming?
• Where should we prospect?
• Who is selling into which region?
• How penetrated are we in each market?

What it reads

• The territory basis your organisation supports (sales regions, market regions or segments), with ARR, bookings and retention for each.
• Region performance, which sales regions sell into which markets, and the TAM you’ve configured for each market.
• Pipeline and deal health, account health movement, renewals, capacity against target, ramp and the new-business and existing-customer mix per territory.
• Data gaps that hide coverage, such as accounts with no country.

What you get

• A deck of up to 11 slides for a review of about 25 minutes: the territory basis, scorecards, who sells where (when available), pipeline, account and deal health by territory, market penetration and coverage questions, engagement gaps, and three to five recommendations, each labelled evidence-backed or a hypothesis to validate.
• It saves nothing. It only reads.

What it won’t do

• Individual rep coaching (use rep-coaching-report) or a board pack (use board-revenue-pack).
• Recommend moving accounts or reps on raw pipeline alone.
• Call a territory with no records “untapped” without outside market evidence.

When the data is thin

• The basis is stated. Which grouping it used, and at which level, is on the slides.
• Unassigned is always shown. Records with no territory get their own row. If they’re more than about a tenth of accounts or value, comparisons are marked partial.
• Like with like. Territories are compared on normalised measures, and ones still ramping are judged against ramp expectations first.
• Currencies aren’t mixed. Penetration is only worked out when TAM and ARR are in the same currency; a market with any child TAM missing has an unknown TAM, not a smaller one.
• Optional views are optional. Cross-tabs, TAM, capacity by region and ICP fit are dropped and listed in the limits when they aren’t available to you.

See Which region is underperforming, really? and the operating view by region.

Troubleshooting

• The assistant answers from general knowledge. The Summit53 connector isn’t attached or signed in. See the MCP setup guide, then start a new conversation.
• It says your membership isn’t approved. A Summit53 admin needs to approve you; see Teams, roles & invites.
• It asks which organisation you mean. Two Summit53 connectors are attached for different organisations. Name the one you want.
• A slide says Demo data. You’re connected to a demo workspace.

Related

• MCP setup guide
• MCP user guide
• Where is the engine stuck?