65% of Your Content Was Already Unused. Then AI Made 42% More of It.

There is a number that has refused to move for over a decade: roughly 65% of the content marketing produces for sales is never used by sales. Forrester named it years ago, and audits keep confirming it — some teams find non-usage north of 80% once they actually count. The standard response has always been to make more. In 2026, more is exactly what we got. Companies running AI now publish around 42% more content each month, and output climbs 77% within six months of rolling the tools out (Averi, State of AI in Marketing 2026). So the graveyard didn't shrink. It got a bigger extension built onto it.

This is the quiet failure mode of the AI content boom. Generation got cheap, so volume exploded, and the thing that was already broken — getting the right page in front of the right buyer at the right moment — got harder, because now there is more to sift through and less of it is on-brand.

The graveyard was already expensive before AI touched it

Strip out the AI for a second and look at what the content library was already costing. Reps burn about 440 hours a year hunting for the right thing to send — that's eleven working weeks per person, spent not selling. Around 28% of sales content isn't even accessible to the reps who need it: it lives in a folder they don't know about, a tool they don't open, or a version that quietly went stale. And when reps do find something, engagement is brutally concentrated — roughly 50% of all prospect engagement comes from just 10% of enablement content. The other 90% is overhead.

So the picture before generative AI: most content unused, a quarter of it unfindable, half the value coming from a tenth of the assets, and reps losing a quarter of their year to search. None of that is a creation problem. Making more case studies was never going to fix a distribution and relevance problem. It just added inventory to a warehouse nobody could navigate.

AI didn't fix the relevance problem — it scaled it

Here is the part the volume metrics hide. When you generate 42% more content but your brand exists as a style guide that's "inconsistently applied" — which is how the majority of teams describe their own setup, with about half still operating at an ad hoc level of AI maturity in 2026 — you don't scale quality. You scale inconsistency. Adobe's framing of agentic marketing put it bluntly: without a shared, machine-readable understanding of the brand, these systems don't scale precision, they scale drift.

That drift compounds with the volume. Every extra asset is another chance for the logo to be slightly off, the positioning to wander, a claim to appear that nobody fact-checked. So the 65%-unused figure isn't holding steady because the content is fine and reps are lazy. It's holding steady because a lot of the new content is generic, off-brand, or unverifiable — and a rep about to send something that represents them can feel that, even if they can't articulate it. They skip it and rebuild from scratch, which is how you get 440 hours of searching and a from-zero rebuild on top.

The buyer changed too, and the library can't keep up

The demand side moved while the content pile grew. The B2B buying committee has gone from an average of 5.4 stakeholders in 2014 to 11 or more in 2026, and on six-figure enterprise deals Gartner-aligned benchmarks now put the group at 10 to 20 people. Each of them arrives with four to five pieces of their own independent research before they ever compare notes with the group.

A static content library was built for a different buyer — one person, one need, one PDF. It cannot serve a fragmented committee where the CFO, the security lead, and the champion each need a different cut of the same argument, read alone, at 9:47pm, with nobody there to point them at the right slide. "Send them the deck" stops working when "them" is eleven people who each read a different 30% of it.

More content is the wrong lever. The right one is per-moment relevance.

The instinct — yours, your CMO's, every platform's roadmap — is to produce more, faster, because AI finally makes that possible. That's optimising the one variable that was never the constraint. The constraint is relevance at the moment of send, on-brand, and verifiable. A library of 300 assets, 65% of which die unused, is worth less than the ability to produce the one right document, on-brand, when a specific buyer needs it.

That reframes the whole job:

Generate for the moment, not the warehouse. Instead of stocking a shelf and hoping a rep finds the asset, produce the specific document for the specific buyer when the deal calls for it — drawn from the brand and the proof you already hold, not the statistical average of the internet. A document built fresh for this committee can't go stale on a shelf, because it was never on the shelf.

Make the brand a system, not a suggestion. The drift problem is solved by treating brand as something checkable on every page — visual grammar, voice, narrative, evidence standard — rather than a style guide someone is supposed to remember. Marq's research found consistent brand presentation can lift revenue by up to 23%, and the only way to hold consistency across 42% more output is to check it automatically, not manually.

Verify before it ships, not after a buyer flinches. When content is generated at volume, the failure mode is a confident claim nobody grounded. Review every page against your own knowledge — the experts, case studies, and numbers your firm actually owns — so a synthetic-sounding stat never reaches a committee that's primed to distrust it.

Track what actually lands. The 10%-of-content-drives-50%-of-engagement reality only became visible because someone measured it. Page-level analytics — what got opened, what got read to the end, what got forwarded to the rest of the committee — tell you which document earned its place and which one was graveyard the day it was made.

The point isn't less content. It's content that doesn't die on arrival.

AI handed every team a content firehose, and the reflex is to celebrate the throughput: 77% more output, 42% lower production cost, a deck in five minutes. But throughput was never the bottleneck — a 65%-unused rate that survived a decade and then survived the AI boom is the proof. The bottleneck is a buyer, often eleven of them, who will read what you send alone and decide in minutes whether it's right.

The teams that win the next two years won't be the ones who generated the most. They'll be the ones who generated the right thing, kept it unmistakably on-brand, checked every page before it left the building, and watched what the committee actually did with it. Everything else is just a bigger graveyard with better tooling.

Stop measuring how much you can produce. Start measuring how little of it dies on arrival.

— The Lurio Team