The Proof Drawer

THE CONTENT ENGINE

The hardest part of content isn't writing. It's having something to say that you can actually back up.

Which is strange, because most businesses generate proof constantly and then throw it away. A customer sends a thank you note and it sits in an inbox. Somebody solves a hard problem on a Tuesday and nobody writes it down. A project delivers a result worth bragging about and the only record is an invoice.

Meanwhile the same business owner sits down on Sunday night, opens a blank document, and tries to think of something insightful about their industry.

You already have the material. It's just filed badly.

I call the fix the proof drawer, and it's the single highest leverage content system I know, because it inverts the problem. Instead of generating ideas and then hunting for evidence, you collect evidence continuously and then let it tell you what to write.

Here's the diagnosis before the cure, because it's worth understanding what's actually broken.

When you sit down to write with nothing in front of you, your brain reaches for the most available thing, which is whatever you've absorbed from other people in your industry. That's why so much business content reads like a slightly reworded version of everything else. It is a slightly reworded version of everything else. It was generated from the same source material: general knowledge floating around the category.

Content built from your own evidence can't do that. If you start from "here's what happened when we tried X with a client last month," you physically cannot end up writing the same post as everyone else, because nobody else has your last month.

This is also, incidentally, why AI written content so often reads as hollow. Not because the writing is bad. The writing is frequently fine. It's hollow because it was generated from general knowledge with no specific evidence underneath it, which is exactly the same failure mode as a human staring at a blank page. Give a model your actual evidence and the output changes character completely.

The bottleneck was never the writing. It's the evidence.

The drawer is one place. It doesn't matter what tool, and people waste enormous energy on this decision. A note app, a doc, a database, a channel. Pick the one you'll actually open.

What matters is what goes in it and how fast.

Questions customers ask. Every time somebody asks you something, particularly if they ask it a second time, that's a piece of content with demand already proven. Write the question down verbatim. The exact wording matters, because their wording is the wording other people are searching with, and yours isn't.

Things that went wrong. Not for public flagellation. For the lesson. Every failure contains a specific, hard won piece of knowledge, and specific hard won knowledge is the rarest thing in business content. Most people only publish wins, which is why win content is worthless. Everyone has wins to talk about. Not everyone will tell you what the recovery actually cost.

Numbers that surprised you. Any time you look at a report and go "huh," capture it. The surprise is the signal. If it surprised you and you're inside the business, it will surprise a reader who isn't.

Decisions and the reasoning. When you choose between two options, write down why in three sentences. This is the most undervalued item in the drawer. Reasoning is what people actually want and almost nobody publishes it. Everyone publishes conclusions.

Things customers said. Verbatim, with context. The thank you email. The offhand comment on a call. The complaint that turned out to be right.

Things you changed your mind about. Gold. Nothing builds credibility faster than "I used to think this, here's what changed."

The rule is: capture in under thirty seconds or it won't happen. No formatting, no cleaning up, no deciding whether it's good enough. Raw and dated. The drawer is a shoebox, not a filing cabinet.

Everybody agrees with the above and almost nobody does it, because the moment of capture is always a moment when you're busy doing the thing worth capturing.

So build the capture into places you already are.

The single best source is calls, because calls are where customers say true things out loud and where your own reasoning gets spoken rather than thought. A tool like Fathom turns every call into a searchable transcript without you doing anything differently, which means the capture cost drops to zero. Once a week you skim, and anything worth keeping goes in the drawer. That one change probably triples most people's raw material.

The second best source is your sent folder. You explain things to customers in writing constantly, usually well, because you're explaining to a specific person with a specific confusion. That's better writing than you'd produce aiming at an audience. Once a week, scan sent mail for any explanation longer than a paragraph. Those are drafts.

The third is a standing prompt to yourself. End of each week, three lines: what surprised me, what did I have to explain twice, what did I decide. Five minutes. It catches the things that didn't happen in a call or an email.

THE AI WORKFLOW BLUEPRINT • $47

The Blueprint includes the capture workflows: the call to drawer pipeline, the sent mail sweep, and the weekly prompts, with the automations already specified. You stop relying on remembering and start relying on a system that collects while you work.

Now the part where it becomes posts, and where most systems break because people try to go from raw note to finished piece in one jump.

There's a middle step, and it's the important one.

Once a week, open the drawer and look for pairs. Not single items. Pairs.

A question a customer asked, plus a decision you made that answers it. A number that surprised you, plus a thing that went wrong that explains the number. A verbatim customer quote, plus a change of mind it triggered.

A single item is a fact. Two items in tension is an argument, and arguments are what people read. The whole reason most content is boring is that it's a sequence of facts with no friction between them.

So the weekly job is pairing, and it takes about twenty minutes. You end with three or four pairs. Each pair is a piece.

Then, and only then, you write. And now you're not staring at a blank page, you're writing up something you already know, which is a completely different and much easier task.

This is also the point where a model becomes genuinely useful. Hand it the pair and the raw notes, tell it your voice, ask for a draft. It's working from your evidence and your reasoning, so what comes back is yours in substance even if the sentences need reworking. That's a fundamentally different exercise from asking it to write a post about your industry, and the output quality difference is not subtle.

Worth saying, because an overstuffed drawer is as useless as an empty one and it fails in a way that's harder to notice.

Industry news goes in the bin. Not because it's uninteresting, but because it isn't yours. The moment a headline goes in the drawer, you've reintroduced the exact problem the drawer exists to solve, which is writing from general knowledge instead of specific evidence. If a piece of news genuinely matters, what belongs in the drawer is not the news, it's what you did differently because of it. That's the part nobody else can write.

Other people's frameworks stay out for the same reason. Save them somewhere else if you like them. They are reading material, not raw material.

Vague positive feelings stay out. "Client seemed happy" is not evidence. "Client said the thing that saved them was getting the draft before the call, not the call itself" is evidence, because it contains a specific mechanism you can argue about.

And anything you'd need permission to publish gets flagged when you capture it, not when you're about to post. Write the note, then add a line about what's shareable and what isn't. Doing this at capture time takes five seconds. Doing it four months later means opening a conversation with a client about something you barely remember, which is how good material dies in the drawer.

The test for everything: does this contain a specific fact, number, quote, or decision that came from inside your business? If yes, in. If no, out. The drawer is not a reading list.

One piece of substance should produce several artifacts, and the mistake is writing them separately.

Write the long version first, always. The full argument, properly made. Then cut down rather than build up, because cutting preserves the reasoning and building up tends to produce something that sounds confident and says nothing.

From one long piece you get the short social version, which is usually the sharpest paragraph plus the conclusion. The email version, which is the long version with a more personal opening. And the reply material, which is the individual points, usable for months as responses in other people's comment sections where being specific and useful builds more trust than any amount of posting on your own feed.

For the scheduling, Buffer does this without ceremony. Queue the variations, let it space them out, stop thinking about it.

And publish the long version somewhere you own. Social platforms are for distribution, not for storage. If your best thinking lives entirely on somebody else's platform you're building on rented land and you'll find out how rented on the day the algorithm changes. This one runs on beehiiv for exactly that reason.

Here's what makes this system worth the effort, and it's not the week to week output.

After six months you have a drawer with a few hundred entries. That's an asset, and it does things a content calendar never will.

You can see patterns. The same question asked eleven times is a product, not a blog post. Three failures with the same root cause is a process problem you'd never have spotted one at a time.

You stop having content emergencies. The blank Sunday night disappears because there's always a drawer with pairs in it.

And your content gets more specific over time rather than less, which is the opposite of what happens to most publications. Most start strong on real experience and drift toward generic commentary as the initial well runs dry. A drawer refills faster than you can publish from it.

Open one document. Call it the drawer.

Put five things in it right now from memory. The last question a customer asked twice. The last thing that went wrong. The last number that surprised you. The last real decision and why. The last nice thing a customer said, in their words.

Set a recurring twenty minutes on Friday to add the week's material and find pairs.

Turn on transcription for your calls so next week's capture is free.

That's the whole system. It's not clever and it doesn't need to be. It just needs to exist before Sunday night does.

THE AI BUSINESS ACCELERATOR • $97

Eight weeks of building the engine with your actual material. We set up the capture, run the weekly pairing with you until it's habit, and build the drafting workflow so your content comes out of your evidence instead of the same ideas everyone else is recycling.

Jordan

The AI Newsroom | Jordan Hale | ainewsroomdaily.com