# Ideation Deck > Ideation Deck - Skill: `ingridleiria/ideation-deck` (Agent Skill) - Install (CLI): `npx skillmds@latest add ingridleiria/ideation-deck` - Raw SKILL.md: https://api.skillmd.com/api/skills/ingridleiria/ideation-deck/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: ingridleiria (https://skillmd.com/u/ingridleiria) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/ingridleiria/ideation-deck --- # Ideation Deck Ideas do not usually die because they are bad. They die in the gap between enthusiasm and a decision, because the deck that carried them asked for a discussion instead of an answer. The meeting ends with "interesting, let us pick this up next month", the window closes, and the person who raised it learns that raising things produces unrewarded work. Two or three cycles of that and the organisation stops hearing ideas from the people closest to the problem. The opposite failure costs more. A deck that blends what is known with what is hoped, in the same font and the same confident tone, gets approved on evidence that does not exist. The pilot then fails for a reason that was visible before it started, and the credibility spent on it is not available for the next idea. ## When to use this, and when not to Use it when an idea has no decision attached to it yet: a new service line, a product feature, a market or segment entry, an internal initiative, a partnership concept, or a set of workshop notes that needs to become something a leadership team can approve, fund, own, or kill. Do not use it once money is being asked of a client. That is `proposal-writer`. Do not use it when the raw material is discovery notes from one named client and the next step is that client approving an approach; that is `discovery-to-proposal-deck`, which plays their situation back in their words and is judged on whether they recognise themselves in it. This skill covers the earlier and broader case, where the idea has no named buyer yet or the audience is internal. Other adjacent cases. Rebuilding an existing standard deck around one named account is `tailored-client-deck`. Asking investors for money is `pitch-deck`. Where the audience is one person, the options are already known, and what is missing is a decision rather than a picture, `decision-memo` is faster and lands harder than any deck. Gathering the market evidence the deck will rest on is `market-research`, and gathering evidence about a specific company is `external-insights`. ## What you need before starting **The decision the deck asks for.** Approve a pilot, fund a phase, assign an owner, kill it. Missing: this is the first thing to fix, not the last. Offer two or three candidate decisions and ask which one they want in the room, because a deck that asks for nothing gets nothing. **The audience, and their first objection.** Every audience has one objection they raise before anything else: cost, distraction from the core business, "we tried this before", risk to a key account. Missing: ask who has killed a similar idea before and why. Where nobody knows, assume the objection is distraction, which it usually is. **What is known against what is believed.** Facts the person has (a client asked, a competitor launched, the team already has the capability) separated from things they think (the market wants it). Missing: sort their material into two lists in front of them. This conversation usually takes ten minutes and changes the deck more than any other step. **The time budget and the format.** Missing: assume ten minutes, which means eight slides and an appendix. A thirty-minute working session supports about fifteen. **The resource envelope and the likely owner.** What could plausibly be spent, and who would run it if approved. Missing: propose an option costing near nothing so that the answer can be yes even when nobody has budget, and name a provisional owner so the room corrects you rather than deferring. ## The method 1. **Write the ask before anything else.** One sentence naming the decision, the resources, and the date. Everything in the deck is then either supporting that ask or cut. Drafting the ask last is why most idea decks end on a slide that says "next steps: discussion". 2. **Answer the first objection before slide five.** Once you have named it, decide where it goes. Rule: if the objection concerns cost or distraction, it is answered by the options slide, so bring the options forward. If it is "we tried this before", it is answered in why now, which then needs a dated fact about what changed rather than a claim that things are different. 3. **Compress the idea into one sentence: what it is, for whom, and what changes for them.** If it needs a second sentence, the idea is not yet clear enough to judge, and the honest move is to say so and work on the sentence rather than build slides around the ambiguity. 4. **Establish why now with a dated trigger.** A market change, a client request, a competitor move, a new capability, a cost the organisation is already paying. Rule: the trigger carries a date and a source on the slide. An idea with no trigger is an idea that could equally be raised next year, and that is how it gets deferred. 5. **Present the evidence in two visibly different registers.** Facts get a source and a date. Beliefs get the words "we believe" and an appendix line stating what would validate them. Rule: never state a belief in the voice of a fact to make the deck feel stronger, because a leader who catches one instance discounts the whole deck, correctly. 6. **Make the idea visible.** A mock, a workflow, a sample deliverable, a screenshot of a prototype, a one-line example of what a customer would receive. Ideas become judgeable when they become concrete, and this slide is usually the one that converts sceptics. 7. **Build two or three options at different levels of commitment, and include doing nothing.** A two-week test, a quarter-long pilot, a full launch. Each with what it costs, what it needs from whom, and what it would prove. Rule: present doing nothing honestly, with the cost of the status quo, because a straw-man version of it discredits everything else on the slide. 8. **State the recommendation, in the author's own voice.** Which option, and the single strongest reason. A deck that ends on "options for discussion" asks the room to do the author's job, and rooms do not. 9. **Define what would be learned, by when, and what would stop it.** The success metric, the review date, and the kill criterion written before anyone is invested. This slide is what earns approval, because it converts an irreversible-feeling commitment into a reversible one. 10. **Write the titles last, as headlines that carry the argument.** "Three clients asked for this in the second quarter" rather than "Client feedback". Test: read the titles alone, in order. They should tell the whole story and end on the ask. Where they do not, the deck has slides that exist to look complete. ## Evidence discipline and file build Every number on a slide carries its source and a date, in small text at the foot. Market size claims are built twice, top-down and bottom-up, and reconciled, with the assumptions visible; the bottom-up build from the organisation's own numbers is the one a sceptical leader will believe. Client or user quotes are attributed with permission, or anonymised by role and date. Where a claim rests on judgement, the slide says so. Build the file with whatever presentation tooling is available, applying the organisation's brand where one exists. Where none is defined, use one dark neutral, one accent, a single sans-serif family, and generous white space. Charts follow the same discipline as the argument: one message per chart, stated in the chart title, direct labels rather than a legend inside the plot area, series separated by marker shape and dash pattern rather than by colour so the chart survives being printed or projected badly, and the accent colour used only on the series the slide is about. ## Worked example **Situation.** A sixty-person facilities services company. A regional manager wanted to launch a remote monitoring subscription: sensors on client plant rooms, an alert service, and a monthly fee, sold into the existing maintenance base. Three clients had asked about something like it during the second quarter. The leadership team met monthly and the slot was ten minutes. **Task.** Get a decision at the next leadership meeting rather than a fourth deferral. The manager had thirty slides of notes and no ask. **Action.** The ask was written first and took three attempts before it was small enough to be granted: approve a ten-week pilot with eight existing clients, 34,000, decision by the end of the month. The audience's first objection was identified by asking who had killed a similar idea before. The operations director had, twice, on the grounds that new services distract the field team during the peak season from October. That moved the options slide to position four and reshaped every option around field team hours rather than around money. The first version of the deck opened with a market size slide: a 1.2 billion figure for the monitoring market, taken from an industry report. It was cut after a rehearsal, where a colleague pointed out that the chief executive's first question would be what share of that the company could plausibly reach, and that the deck had no answer. It was replaced with a bottom-up build from the company's own data: 340 service contracts, an assumed attach rate of fifteen percent based on the three inbound requests against the eighty-two clients contacted that quarter, at 340 a month, giving roughly 168,000 in year one with the attach rate labelled as the assumption to test. The smaller number was more persuasive than the larger one, because every input in it came from a system the room already trusted. Known and believed were separated explicitly. Known: three inbound requests with dates, a competitor's launch in March, and 240,000 of emergency call-out revenue in the previous year that monitoring would partly displace. Believed: that clients would pay monthly rather than expect it bundled, which became the pilot's primary test. Options ran at three levels. Do nothing, with the cost stated as the call-out revenue continuing to be reactive and the competitor's head start growing. A two-week desk test, effectively free, that would only produce opinions. The ten-week pilot at 34,000 using two field engineers for four hours a week, protected from October by ending in September. The kill criterion was written before approval: fewer than three of the eight pilot clients agreeing to a paid monthly fee at the review meant stop. **Result.** Approved in eight minutes, with the operations director as the named owner, which had not been the plan and turned out to matter more than the budget. Four of the eight pilot clients signed. The one number that did not hold was the price, which landed at 240 a month rather than 340, and the second deck, asking to fund the launch, opened with that correction. Opening with the miss rather than burying it is what made the second approval quick. ### A second scenario, where it goes differently The obvious variant, taking the same concept to a client's leadership team, is not this skill's job. Rebuilding a deck around one named account, in their vocabulary and with their numbers, is `tailored-client-deck`; where that client has already been through discovery and is ready to approve an engagement, the route is `discovery-to-proposal-deck` and then `proposal-writer`. This skill covers the idea nobody has approved yet, which includes one shape people rarely think of as an idea. Eighteen months later the same manager proposed retiring the company's small print management line: eleven clients, about 190,000 a year, which three people had separately concluded was losing money and none had proposed stopping. Three things changed. Why now had no external trigger, so it had to be built from an internal one, and the dated fact became the month the last dedicated engineer left and the work started being absorbed by the maintenance team. The evidence slide inverted: instead of a bottom-up build of a market that might exist, it was a bottom-up build of a cost that already did, at 240 hours a quarter of field time against 47,000 of gross margin, every input from a system the room already trusted. And the visible slide became the letter the eleven clients would receive, drafted in full, because the objection to a stop proposal is never the arithmetic, it is what happens to the people affected. The options still ran at three levels of commitment and doing nothing still appeared honestly, since continuing was a real choice with a stated cost. What replaced the kill criterion was its mirror: the condition under which the line would be restarted, which is what made a decision that sounded irreversible cheap enough to approve. What changed was the direction of the ask, not the spine. A proposal to stop something needs the same eight slides, and the slide it most often omits is the visible one. ## Output An eight-slide spine for a ten-minute slot, with an appendix. Each title is written as a headline. | # | Slide | What it carries | Evidence required | | 1 | The idea in one sentence | What it is, for whom, what changes for them | none | | 2 | Why now | The dated trigger | source and date on the slide | | 3 | The problem, with evidence | Signals, quotes, a number the organisation already pays | each item sourced | | 4 | What it would look like | Mock, workflow, sample deliverable | none | | 5 | Options | 2 or 3 levels of commitment plus doing nothing, each with cost, need, and what it proves | costs labelled as estimates | | 6 | Recommendation | The chosen option and the strongest reason | none | | 7 | What we would learn | Success metric, review date, kill criterion | metric defined and measurable | | 8 | The ask | One decision, the resources, the date, the proposed owner | none | Appendix: the costing detail, the bottom-up sizing with its assumptions, sources, risks and mitigations, and what would validate each stated belief. ## Failure modes **The brainstorm dump.** Recognise it because the deck has more than one idea in it. Pick the one that can be decided this month and put the rest on a single appendix slide headed "not proposed today". **Ending on discussion.** Recognise it because the last slide has no date and no named resource. Rewrite it as one decision, one number, one date. **Belief in the voice of fact.** Recognise it by asking, of each declarative sentence, where the number came from. Anything that cannot be answered gets "we believe" and an appendix line stating what would validate it. **The oversized market number.** Recognise it because the figure is large, from an external report, and has no bridge to what this organisation could reach. Replace it with a bottom-up build from the organisation's own base, even though the result is smaller. **The objection answered in the appendix.** Recognise it in rehearsal: the objection is raised on slide two and the answer is on slide fourteen. Move the answer forward, even if it breaks the narrative. **No kill criterion.** Recognise it because the deck describes what success looks like and never what would stop the work. Approvals are cheaper when they are reversible, so write the stopping rule before anyone is invested in continuing. **Titles as labels.** "Client feedback", "Approach", "Financials". Recognise it by reading the titles alone: if they name topics rather than make claims, the deck cannot be understood by skimming, and senior audiences skim. ## Edge cases **The idea is the leader's own.** The deck's job changes from persuasion to honest shaping. Keep the options and the kill criterion, and put more weight on what would have to be true, since the risk here is a fast approval of an untested assumption rather than a slow no. **No evidence exists at all.** Do not manufacture it. Reframe the ask as permission to gather evidence: a two-week test with a named question, a small budget, and a date. A deck that asks for a small, cheap answer to a real question is easier to approve than one that asserts a market. **The decision belongs to someone who will not be in the room.** Do not present for a decision that cannot be made. Either get that person into the meeting, or change the ask to a recommendation the room can carry to them, and provide the one-page version they will actually read, which is `decision-memo`. **The idea has been raised and rejected before.** Lead with what has changed since, dated. Without that, the room re-runs the previous argument from memory and reaches the previous conclusion. **The time slot collapses to three minutes.** Present slides one, five, and eight, in that order, and send the rest as a pre-read. Compressing all eight slides into three minutes produces a blur that gets deferred. ## Quality bar - Read alone and in order, the titles tell the whole story and end on the ask. - The ask names one decision, the resources, the date, and a proposed owner. - The audience's first objection is answered before slide five, explicitly. - Doing nothing appears as an option with the cost of the status quo stated. - Every fact carries a source and a date on the slide; every belief carries the words that mark it as one. - Any market or revenue figure is built bottom up from numbers the audience already trusts. - The recommended option has a success metric, a review date, and a kill criterion. - Nothing is on a slide because the slide looked empty. ## Adapting this to your context The defaults come from ten-minute slots at leadership meetings in commercial companies of fifty to three hundred people. The spine holds; the slot rarely matches. - **Eight slides for ten minutes.** A ratio, not a rule. A thirty-minute working session supports about fifteen slides. A written-decision culture supports none, and the right artefact there is `decision-memo` with the mock attached to it. - **The bottom-up build.** It assumes the organisation trusts its own numbers. Where the payoff is not revenue, build the same way from whatever you already count: hours released, places filled, incidents avoided, grant income at risk. - **The kill criterion.** Written for a reversible pilot. Where the commitment is a licence, a building or a permanent hire, the equivalent is a stated stage gate with money released in tranches, and it belongs on the slide rather than in the appendix. - **The audience's first objection.** Cost or distraction in most commercial rooms. In regulated, clinical or public settings it is usually risk or precedent, and it is answered by naming the governance route rather than by the options slide. - **What not to change.** Facts and beliefs sit in visibly different registers, and the deck ends on one ask carrying a decision, a resource and a date. ## Related skills `market-research` and `external-insights` produce the sourced evidence this deck rests on, and `customer-interview-synthesis` produces the client signals that make the problem slide credible. `discovery-to-proposal-deck` takes over when the raw material is one named client's discovery notes and the next step is that client approving an approach. `proposal-writer` takes over the moment the client is being asked to buy, and `sow-and-scope` follows once they have. `tailored-client-deck` handles rebuilding an existing deck around a named account, including branding and logo mechanics. `decision-memo` is the one-page alternative when the audience is a single decider and a deck would be theatre, and `pitch-deck` is the investor version. `principal-simulator` rehearses the deck against the specific leader who will react to it, and `program-management` runs the pilot once it is approved.