What Goes in a Pitch Deck Appendix—and What Must Stay in the Main Deck?
What Goes in a Pitch Deck Appendix—and What Must Stay in the Main Deck?
A main slide reports that pilot customers cut fuel spend per delivered case by 11%. The definition of fuel spend per delivered case—which fuel counts, which cases count, what was excluded—sits thirty slides later, in the appendix. Anyone who reads only the main deck gets a number. They don't get a meaning.
That is the failure worth guarding against. Not a crowded slide, not a long appendix. A main deck that has quietly stopped being true on its own.
The remedy isn't to move everything back. It's to run one test on each candidate detail, then organize whatever survives the test around the questions a recipient will actually ask.
Take the detail away and see whether the claim survives
Read a main-deck claim with its proposed supporting detail removed. Ask whether the remaining sentence still means the same thing.
"Pilot distributors cut fuel spend per delivered case by 11%" survives losing the calculation worksheet. It does not survive losing the unit. If delivered case could plausibly mean either outbound cases on customer routes or everything that leaves the depot, a recipient who assumes the second reads the 11% as a different achievement than the one measured. Stripping that boundary off the slide, and letting the reader recover it from slide 34, is not tidying. It's a change of claim in disguise.
So the question is not "is this detail interesting?" but "does removing it let a reasonable reader draw a materially different conclusion?"
The distinction is worth stating plainly, because it's the one that gets fudged: removing detail and removing meaning are different operations that look identical on a slide deck. A chart's underlying table is detail. The exclusion note that explains why the chart starts in March is meaning. The first belongs in the appendix. The second belongs beside the chart, even if that means the chart is smaller and the slide has a line of small print on it.
Consider what the appendix actually is. A reader who opens it has chosen to go deeper than the main argument requires. That's fine—that's the point of the layer. But they can only make that choice well if the main layer was complete on its own terms.
Throughout this piece I'll use one invention to keep the decisions concrete. Halcyon Route is a fictional route-planning product for regional beverage distributors. Its pitch deck claims that four pilot distributors cut fuel spend per delivered case by 11% over six months. Nothing about it is a company record; it's a construction for sorting material. Here is the outline:
- A product claim: the 11% reduction.
- A metric definition: how fuel spend per delivered case is calculated.
- A detailed daily workflow for dispatchers.
- Scenario assumptions for the savings.
- A dated supporting source: the pilot summary.
- A confidential customer record: one distributor's contract terms and route-level data.
Six items, four destinations. Most decks have more of each and need more of the same sorting.
Give each appendix entry a question it answers
The second item is where the split gets delicate, because the metric definition has two parts that belong in two places.
The part a recipient needs to read the claim correctly stays with the claim: the measure counts only company-operated customer routes, and it divides fuel-card charges by outbound cases delivered on those routes. A recipient who doesn't know that could reasonably think subcontracted loads are in the number, or that depot transfers are.
The part that answers a follow-up question goes behind a named question: the full exclusion list, the detailed data-source documentation and extraction fields, and the period-by-period comparisons behind the six-month figure.
The fact that the figure was compiled by Halcyon from the distributors' fuel-card statements and dispatch records, with each distributor reviewing its own numbers, deserves attention for the same reason. It's tempting to push it to the appendix along with everything else methodological. It's not methodology. It's a compiler caveat: it tells the reader how much weight to give the number and who has seen it. It changes the reading. Keep it on the main slide, short, in whatever space the design allows.
Then name the appendix entry for the question it answers. "Q4: How was fuel spend per delivered case calculated?" beats "Additional Metrics." A question-shaped heading tells a presenter what it's for and tells a skimming reader whether to stop.
The workflow item gets the same treatment. The main deck can say the tool fits the existing dispatch routine. The appendix walks the full daily sequence: when routes are proposed, the will-call exception, the driver override, what happens when a truck goes out with a partial load. That's depth for the recipient who has run a depot and will ask. It's also three or four slides that would derail the main argument if left in line.
The scenario assumptions are the most interesting case, because they show the appendix doing something the main deck can't. The pilot's 11% reduction was measured during a period when diesel averaged about $4.10 a gallon across the four distributors' regions. The percentage reduction is in fuel spend per delivered case, and it tracks the proportional reduction in fuel use per case; price cancels out, so the percentage holds across prices. The dollar saving per case is what scales with the price. At $4.10, the saving is roughly 1.1 cents per delivered case. At $3.30, about 0.9 cents. At $4.80, about 1.3 cents.
Why does this belong in the appendix rather than on the claim slide? Not because the per-case rate depends on the recipient's volume—it doesn't. The 11% claim is a percentage; turning it into dollars requires fuel-price and volume assumptions. The appendix can hand over the per-case rate at the pilot's average price and the sensitivity band, and let each recipient multiply by their own case volume. The total dollar value depends on the recipient's volume; the per-case rate and the band depend on fuel price and the measured efficiency gain. That's a real question answered at the right depth, in the right layer.
The dated source goes in with it: the pilot summary, covering 1 September 2025 to 28 February 2026, dated 20 March 2026. Dates and provenance are exactly the kind of thing an appendix exists to carry, because the recipient who cares about them cares a lot.
One caution. An appendix is not where unresolved disagreements get parked. If the main deck argued that Halcyon is a software company and deliberately dropped an earlier slide positioning it as a logistics brokerage, putting the brokerage slide in the appendix doesn't preserve optionality. It tells the reader the deck hasn't decided what it's selling. Depth answers questions; it doesn't reopen verdicts.
Preparing supporting slides for questions beyond the main presentation isn't a new idea. An investor writing for NextView in a September 2019 post, "Pitch Deck Month: Appendix Slides," argued for exactly that in a seed-stage context. It's one investor's practice from one period, not a checklist, and the template the post links isn't part of what I'm drawing on—but the underlying move is sound and durable: the questions come first, the slides follow.
Make the appendix reachable in the room
An appendix that can't be reached mid-conversation is storage.
Two habits help. First, number the questions and the answers so they have short names—Q4 goes to A7. Second, put a pointer at the claim. A presenter who is asked "how did you measure that?" should not be scrolling while the room waits.
The pointer can live in the presenter's notes on the claim slide, which keeps the recipient view clean. Whether those notes travel with an exported file depends on the format and the software that produced it, so if the same deck will also be read alone, check that before relying on it. If the notes don't travel, the pointer has to live somewhere a recipient can see.
Now the choice you'll actually have to make: a topic index at the front, grouping entries under "Metrics," "Workflow," "Scenarios," or a direct question-level reference from each claim. A topic index is easier to maintain across revisions. Under pressure it's slower, because the presenter has to remember whether the fuel definition sits under Metrics or Method. A question-level pointer is faster in the moment, and it forces you to phrase each entry the way a recipient would ask it, which tends to improve the entries themselves.
There's a second decision underneath it: whether appendix entries sit together at the end or ride directly behind the claim they support. Distributed slides are quicker to reach live and can blur the main line for someone reading straight through. Consolidated slides keep the argument continuous and require a jump. Neither is correct in general; pick the one that matches how the deck will mostly be used, and accept the cost of the other.
And then check the thing you've built, in the file you'll actually send. Open the exported deck, click the pointer, see where it lands. Decide on purpose whether coming back means returning to the claim slide or continuing forward into the next argument—if the appendix sits at the end, going forward is usually the wrong move, because forward is the end of the deck. Do not assume a link, a jump, or a return path works because it worked in the file you built it in. Numbering shifts when slides are inserted. Links break on export in ways that are invisible on screen until someone tries them. None of the traversal described here has been run against a real file; it's a plan, and the test is the test.
Make it reachable without you in the room
The same deck will probably be forwarded. Someone will open it with no presenter, no context, and no way to ask what Q4 means—which means the read-alone route has to be built into the artifact, not into the delivery.
The pointer from claim to appendix needs to be visible on the page, not just in the notes: a short line on the claim slide, "Method and exclusions: Q4." The appendix entry needs to restate its question as a heading, so a skimmer can find it without reading everything. Sources and dates go on the entry, not in a final credits slide. And the qualifications that change how a claim reads stay next to the claim, in both modes, because the read-alone reader is the one most likely to miss them.
Then there's the material that shouldn't be in the file at all. The confidential customer record—one distributor's contract terms and route-level data, shared with Halcyon under an NDA—does not belong in a circulated deck, whether it's page 12 or page 60. The appendix rides along with everything else. It is not private by position. If a specific recipient needs that record during diligence, it goes through a separate, permissioned exchange with the distributor's consent, and the deck says nothing about it beyond whatever the main claim already relies on.
This is a content and disclosure decision, not a technicality. Treating the appendix as an informal data room is how decks end up containing things their owners didn't intend to distribute.
Cut what has no question to answer
Sort the remaining candidates into four places, and be able to say why each one landed where it did:
Main deck. The claim, plus every condition or definition that changes what the claim means. Compiler caveats, unit definitions, the measurement window that bounds the claim.
Appendix. Depth that answers a named question a recipient would plausibly ask: method, full exclusion list, period-by-period comparisons, workflow variants, scenario breakdowns, dated sources.
Omit. Anything with no live question behind it, and anything that renews an argument the deck settled. The brokerage-positioning slide. Raw per-distributor tables that duplicate the summary. Nobody is asking.
Separate route. Material that shouldn't travel in the file: third-party records, anything under an NDA, anything requiring an owner's consent before disclosure.
Two things this is not. It isn't a slide-count rule—there's no correct number of appendix slides, and a thick appendix is not evidence of preparation. And it isn't a reason to build a different version of the deck for every recipient you can imagine. The question is where each item belongs in this deck for this audience, decided once and revisited when the audience changes meaningfully.
A cleaner slide is not automatically a better one
The mechanical version of all this is easy: read each claim without its supporting detail, ask whether the meaning changed, and if it did, put the necessary part back. Then give what's left a question to answer, a name, a date, and two routes to reach it—one for the room and one for the forwarded file.
The reason to bother is that the two failure modes look nothing alike from the inside. A bloated main deck is obviously a problem. A main slide whose omitted condition makes it slightly misleading looks like good design. It reads clean. The 11% sits there, unqualified, and the person who needed to know that subcontracted loads were excluded is now building a model on top of a number that means something else.
A main deck should stand on its own. An appendix should answer named questions. When those two sentences pull in different directions, the main deck wins, and the slide gets a line of small print.
Frequently asked questions
How do you decide whether a detail belongs in the main deck or the appendix?
Take the detail away and ask whether the claim still means the same thing. If removing it lets a reasonable reader draw a materially different conclusion, the necessary part belongs with the claim. The article distinguishes removing detail from removing meaning: an underlying table is detail, but an exclusion note that explains why a chart starts in March is meaning and belongs beside the chart.
What makes a good appendix entry heading?
Name the entry for the question it answers. The article gives 'Q4: How was fuel spend per delivered case calculated?' as better than 'Additional Metrics.' A question-shaped heading tells a presenter what the entry is for and tells a skimming reader whether to stop. Entries should also carry sources and dates, and the appendix is not a place to park unresolved disagreements or reopen a verdict the main deck settled.
Why do the fuel-price scenarios belong in the appendix rather than on the claim slide?
The 11% reduction is a percentage and tracks the proportional reduction in fuel use per case, so price cancels out. The dollar saving per case scales with fuel price: roughly 1.1 cents at $4.10, about 0.9 cents at $3.30, and about 1.3 cents at $4.80. The appendix can give the per-case rate at the pilot's average price and the sensitivity band, while each recipient multiplies by their own case volume. Total dollar value depends on recipient volume; the per-case rate and band depend on fuel price and the measured efficiency gain.
How can the appendix be made reachable both in the room and in a forwarded file?
Number questions and answers so they have short names, and put a pointer at the claim. Presenter notes may not travel with an exported file, so check that before relying on them; if they do not travel, the pointer must be visible to a recipient. For read-alone use, put a short line on the claim slide, restate each appendix entry's question as a heading, keep sources and dates on the entry, and keep qualifications next to the claim. The article says to test the exported deck by clicking the pointer and seeing where it lands, and to decide on purpose whether returning means going back to the claim slide or continuing forward.
What material should not be in a circulated deck at all?
Confidential customer records shared under an NDA, such as contract terms and route-level data, should not be in a circulated deck whether they are page 12 or page 60. The appendix is not private by position. If a specific recipient needs that record during diligence, it should go through a separate, permissioned exchange with the owner's consent. More generally, omit anything with no live question behind it and anything that renews an argument the deck already settled.