What 47 Agency Proposals Taught Us About Why Decks Fail

We tracked 47 client proposals across three agencies from kickoff brief to sent PDF, and the decks that lost did not lose for the reasons the teams expected. When a proposal fell flat, the partners blamed price, or timing, or a competitor who knew someone on the panel. Almost none of them blamed the document. But when you read the losing proposals back the way the buyer read them, alone and fast, the same four failures show up again and again, and every one of them was invisible to the person who wrote it.

TL;DR: Across 47 tracked proposals, decks failed in four repeatable ways: numbers that did not reconcile across pages, arguments written for the author instead of the reader, a brand that drifted slide to slide, and an ask so hedged the buyer could safely do nothing. All four share one root cause, which is that the person who built the proposal is the worst-placed person to catch them. The fix is to review every page the way the buyer will, grounded in your firm's own knowledge, before you send.

The Sample, and What We Could Actually Measure

The 47 proposals came from a strategy boutique, a growth agency, and a creative studio, tracked over roughly six months of live pitching. The hard number that came out clean was time: 14 hours from brief to sent PDF on average, of which about 11 went to assembly rather than thinking. That ratio is covered elsewhere. This piece is about the other thing the sample showed, which is not how long the decks took but why the weak ones did not land.

Failure prevalence is harder to quantify than hours, so we are deliberate here. We are not going to attach a precise percentage to each failure mode, because a sample of 47 read by hand cannot support that kind of decimal. What it supports is a ranking: four patterns recurred often enough, across all three agencies, to be worth naming. The numbers we cite below are the external, published ones that explain why each pattern is fatal.

Failure One: The Numbers Stopped Agreeing With Each Other

The most common defect in the losing decks was internal contradiction. The executive summary said one growth figure, the financials slide said a slightly different one, and the case study three pages later used a metric that was two years stale. None of it was dishonest. It was the residue of copy-paste: a number updated in one place and missed in another.

A buyer who catches one figure that does not reconcile does not file it as a typo. They file it as a reason to distrust every other number, and they are reading without you there to explain. Gartner found that B2B buyers spend only 17% of the purchase journey with any one supplier, and DocSend's analysis puts the average time spent reading a deck at under three minutes. The proposal carries the whole argument alone and fast, and one broken number quietly discounts the rest.

Failure Two: It Was Written for the Person Who Wrote It

The second pattern was proposals that made perfect sense to their author and much less sense to anyone else. The opening assumed context the buyer did not have. The argument skipped the step that was obvious to the strategist and load-bearing for the reader. The proof was there, but buried where only someone who already believed the thesis would look for it.

This is not carelessness. It is the curse of knowledge, first measured by Camerer, Loewenstein and Weber (1989): once you know something, you cannot easily model not knowing it, so you read your own intention into every page. The person who wrote the proposal is structurally its worst reader, which is why the failure survives the author's own final pass.

Failure Three: The Brand Drifted Page to Page

The third pattern was visual and tonal drift. A pasted chart in last quarter's palette. A heading in the wrong weight because it came from a different deck. A section written in a new hire's voice next to a section written in the partner's. The cover looked sharp. By slide nine the document no longer looked like one firm had made it.

The revenue cost of that is measured. Marq's brand consistency research found that consistently presenting a brand can lift revenue by up to 23%, and that 68% of business leaders credit brand consistency with at least 10% of revenue growth. For an agency the stakes are sharper than the average business, because the proposal is the product demo. A buyer who watches you fail to hold your own brand together across twelve slides concludes, reasonably, that you will not hold theirs together either.

Failure Four: The Ask Was Easy to Ignore

The fourth pattern was the quietest and, in the losing decks, the most decisive. The proposal never made saying yes feel safer than saying nothing. The scope was hedged, the next step was vague, the risk of choosing this agency was left for the buyer to work out alone. So they did the safe thing, which was to not decide.

No decision is the outcome agencies underweight most. Between 40% and 60% of qualified B2B pipeline ends in no decision rather than a loss to a named competitor, and buying committees have grown past eleven stakeholders, each able to stall the deal by simply not championing it. A proposal that does not de-risk the yes lets the committee default to inaction, which is where most deals actually die.

The One Thing All Four Failures Share

Read the four back to back and the common thread is not craft. It is blindness. The contradicting number, the reader-hostile argument, the drifting brand, the ignorable ask: none were visible to the person who built the deck, because the builder is too close to see them. The teams in our sample were not bad at their jobs. They were doing the one thing no writer can do well, which is review their own work as a stranger, at 11pm, by eye.

You cannot will yourself out of the curse of knowledge. You need a reader who was not in your head when you wrote it.

How Lurio Handles This

This is the job Lurio's review agents are built for. Before a proposal leaves the building, review agents trained on your firm's own knowledge read every page: Data Integrity checks that the numbers reconcile across slides, Narrative Reviewer checks that the argument lands for someone who was not in the room, Brand Compliance checks that the brand holds from cover to appendix, and Audience Fit checks that the ask is concrete for this specific buyer. Every flag is cited back to the source it came from, so it is a grounded critique, not a guess. You edit anything, and nothing ships without your sign-off.

The strategic thinking still takes the time it takes. What changes is that the four failures that sank the losing decks in our sample get caught by a reader who can actually see them, while you can still fix them.