A founder once told me she'd budgeted her deck for "the standard three minutes" — she'd read it somewhere, built her narrative arc around it, timed her slide transitions to it. She never got three minutes. The investor closed the deck at 47 seconds, on the second slide, and moved on to the next email in his inbox.

She wasn't unlucky. She was working off a number that was already stale by the time she found it.

The Number Keeps Shrinking, Not Holding Steady

Most of the "investors spend X minutes on a deck" advice floating around cites a single DocSend figure — three minutes and forty-four seconds — pulled from a study DocSend ran with Harvard Business School's Tom Eisenmann, first reported by TechCrunch. It's a real number. It's also years old, and treating it as a fixed ceiling is the mistake.

More recent DocSend data, covered in a 2026 release from Spotlight on Startups and broken out year-by-year by PitchBuilder, shows the average seed-stage deck now holds an investor's attention for about one minute and fifty-six seconds — a drop of roughly 24% since 2021. That's not a plateau. It's a trend line, and it's been pointing the same direction for several years running: as investors receive more decks, each one gets less time, regardless of how good it is.

Which raises the actual question founders should be asking. Not "how do I fill three minutes well," but "what happens to my deck in the two minutes I'm actually going to get, and is that different from what I designed for?"

Where the Shrinking Time Actually Goes

Here's the part that complicates "just make it shorter." A 2025 analysis by Papermark, covering roughly 3,000 pitch decks, found that the opening page receives more than twice the attention of any slide that follows it — with subsequent pages averaging only about 15 seconds each once the deck moves past the first page.

So the two minutes isn't spread evenly across a ten-slide deck. It's front-loaded almost entirely onto slide one, and everything after that is competing for scraps. A deck engineered to "build up" to its strongest point on slide seven is built on an assumption about attention that the data doesn't support anymore. If the opening page doesn't do the convincing, there often isn't a second chance for slide seven to do it instead — most of the deck is already being skimmed by then.

Not Every Slide Loses Time at the Same Rate

One thing that surprises founders when they see it: the slide people assume matters most for a first impression is often the one that gets skipped fastest. A 2026 review of investor session data published by Storydoc found the "About Us" slide — despite appearing in the vast majority of top-performing decks — was only opened in about 59% of sessions, and when it was opened, investors spent an average of roughly two seconds on it.

Meanwhile, the team slide is the one section that has actually been gaining attention rather than losing it, per the same DocSend-sourced data cited by Spotlight on Startups. That's a strange pair of facts to sit next to each other: investors are spending less total time on decks, but more of the time they do spend goes to the people behind the company rather than the mission-statement framing of who the company is. A polished "About Us" paragraph isn't earning its slide anymore — a two-line team credibility signal is doing more work in less space.

Shorter Isn't Automatically the Fix

It would be easy to read all this as "cut the deck to five slides and you're done." That's the same trap covered in more depth in a look at whether minimalism actually helps a presentation — fewer slides doesn't buy you more attention, it just redistributes the same shrinking amount of attention across fewer targets. Cut the wrong slides and you've deleted the evidence an investor needed to say yes in the ninety seconds they were willing to give you, not made the deck more efficient.

What actually moves the needle is deciding, before a single slide gets built, which two or three things this specific reader has to walk away believing — and then making sure those specific things land inside the first fifteen to twenty seconds, not the last. The discipline that makes a seven-slide pitch outperform a thirty-slide one isn't really about the slide count at all; it's that someone was forced to decide what mattered most before the reader ran out of patience to find it.

The Deck That "Reads All the Way Through" Is Rarer Than It Sounds

Here's a detail worth sitting with: even among decks that do get opened, DocSend's more recent research (again via PitchBuilder's breakdown) puts the share that get read all the way to the final slide at only 58%. Just over four in ten decks never make it to whatever closing argument the founder built the whole narrative toward.

That changes what "the last slide" is allowed to be. A big-swing summary slide, a call-to-action that assumes the reader has absorbed everything before it, a closing number that only makes sense in context — all of that is a bet on a coin flip. I wouldn't recommend building your strongest single argument for the very end of the deck anymore. If it can survive on its own, near the front, it should live there instead.

What This Actually Changes About Building the Deck

None of this means panic-trimming a deck down to a number that sounds safe. The length debate itself tends to miss the real variable — a longer deck built for diligence and a short deck built for a cold first look are doing two different jobs, and the shrinking-attention data above is specifically about the first look, not every stage of fundraising.

What it does mean, in my experience building decks for founders raising early rounds: assume the reader gives you the opening page and maybe one more slide before deciding whether to keep going. Put the team credibility signal and the core ask somewhere in that window, not buried as a payoff for people who make it to the end — because fewer of them will than you'd expect, and that number has been shrinking every year the data's been tracked.

The deck that wins in that world isn't the shortest one or the prettiest one. It's the one that assumes it only gets read once, from the top, by someone who might stop at any point — and puts nothing load-bearing anywhere else.