blog / how-to-improve-a-pitch-deck
How to improve a pitch deck you've already got
Most advice on improving a deck starts with "rewrite it." That's bad advice. A deck you've used has real numbers in it, real photos, and phrasing you've already tested on people. Throwing that out and starting again usually produces something worse that took longer.
The useful move is diagnosis. Here are six checks, in the order that finds the most per minute. You can run all of them in about twenty minutes with the deck you have open.
1. Read only the headlines
Strip everything but the slide titles and read them top to bottom.
If those sentences don't make your argument on their own, that's finding number one and nothing else matters as much. Investors skim headlines first; a deck whose headline-only read works survives being half-read, which is how it will actually be read.
The failure is almost always the same: headlines that name a topic instead of stating a conclusion.
| Topic label | Conclusion |
|---|---|
| Market | The wedge is a $4B niche nobody serves well |
| Traction | We hit $40k MRR in five months, from $12k in March |
| Team | We ran this operation at scale before, at the incumbent |
| Why now | Freight APIs only opened up eighteen months ago |
This is the single highest-leverage edit available on most existing decks, and it costs you nothing but an afternoon. You already know each slide's point — you say it out loud when you present. Put that sentence in the title.
2. Count the points per slide
One slide, one point. If a slide makes two, it's two slides — or one slide with something cut.
The quick test: say the slide's point in one sentence. If you need "and" to join two claims, or a second sentence, it's carrying too much. A live audience can hold one idea per slide. So can a reader, at the speed they're reading.
3. Count the numbers
More than about three numbers in a headline and its supporting line, and the slide has stopped being narratable. Say it out loud and you'll hear it: "SEA aftermarket parts is 8.4 billion growing 11% with 42,000 garages across 3 markets" is a paragraph pretending to be a claim.
Important exception: a chart doesn't count. A chart or a table is inherently many numbers making exactly one point, which is the whole reason those formats exist. The budget applies to the prose — the headline and the line under it — not to the data the visual carries.
4. Ask whether it's a pitch deck or an email deck
This is the question that reframes everything else, and most decks never get asked it.
- A pitch deck is narrated. You're in the room, so the slide holds one idea and you supply the rest. Sparse is correct.
- An email deck is read alone with nobody explaining. It has to carry more per slide, because there's no voiceover.
They are different artifacts. If you're failing checks 2 and 3 on most slides, you probably don't have a broken pitch deck — you have an email deck that someone is presenting live. That's a much more useful diagnosis than "too dense," because the fix is different: either thin it down for the room, or accept it's a reading document and stop presenting from it.
And if you need both — the common case — the answer is a sparse deck plus an appendix, not one deck that compromises and fails at both jobs.
5. Decide what each slide looks like, in plain words
Every slide has a visual form, including the text-only ones. "One decisive sentence, big, with a lot of space around it" is a design decision. Having made no decision is how you end up with a template's defaults.
Go slide by slide and finish this sentence: this slide is…
- "…one big number, and nothing else on it"
- "…a photo of the actual plant, full bleed, no text over it"
- "…the three points on the right, the product shot on the left"
- "…just the one line"
- "…the revenue chart, with the last bar highlighted"
Two rules. Say the medium and what it must show, never the styling — colours and fonts are your brand's job, and deciding them per slide is how a deck stops looking like one thing. And not every slide wants a visual: a deck that's all infographic is as tiring as one that's all text. The restraint is the craft.
6. Find the proof that never made it onto a slide
The most common real defect in an existing deck isn't a bad slide. It's traction the founder has and never put in.
Decks get written early and then patched. The numbers in yours may be from the version you wrote in March. Go and look: what's happened since? Which customer renewed, which pilot converted, which number doubled? That's usually sitting in a Stripe dashboard rather than on slide nine.
While you're there, audit the other direction: every claim, is it evidence or hope? "Three signed LOIs" is evidence. "Strong inbound interest" is hope in evidence's clothing. Investors discount the second automatically, and one uncited market number makes them start discounting everything else too.
What not to change
- Don't touch a slide that already works. A slide that states a conclusion and carries its proof is finished. Say so and move on.
- Don't lose your assets. Real photos and real screenshots of your real product are the hardest things to get and the easiest to drop in a rebuild.
- Don't rebuild all thirty slides. An improve pass that replaces everything is a new deck with extra steps, and you won't recognise it as yours.
- Don't fix it without looking at it. Render or export every slide and actually look. Placeholder text left in the copy, a chart with a stray series, a QR code that won't scan — none of those are visible in the text.
Frequently asked questions
What's the fastest way to tell if a pitch deck is bad?
Read only the headlines. If they don't make the argument by themselves, the deck is doing worse than you think, because that's most of what a first reader takes in. It takes two minutes and finds more than an hour of redesign.
Should I rewrite my deck or fix it?
Fix it, almost always. An existing deck contains real numbers, real imagery and phrasing you've tested on live audiences. Diagnose what's wrong — usually headlines and density — and change that. A full rewrite discards the parts that were working.
How many slides should a pitch deck be?
10–15 for the version you send, since the average first read is two to four minutes. If yours is 30, the answer usually isn't cutting 15 slides but moving them into an appendix you bring to the meeting.
My deck looks fine but nobody replies. What's wrong?
Usually the ask, or the proof. Check the last slide: is there one specific thing you want, small enough to say yes to? "Let's explore working together" reads as no conviction. Then check whether anything in the deck is checkable — a deck with no verifiable numbers gets a polite no regardless of design.
Can I improve a PowerPoint or PDF without rebuilding it?
Yes. Import it, keep the layout, imagery and fonts, and change only what the diagnosis names. The point of importing rather than re-authoring from the text is that re-authoring throws away every visual decision you already made.
Get the diagnosis before you start editing
Point DeckCP's /deck skill at a deck you already have — in DeckCP, or a PowerPoint or PDF you drop in. It reads it, renders every slide, works out the outline it's currently making, proposes a better headline wherever yours is a topic label (keeping your original alongside so you can see exactly what changed), and runs these checks for you. Then it asks the two or three things your deck can't answer.