I Measured My Own Software. Most Of It Was Empty.

Nine months of live use, a real team, my own software. Every field a human had to type was empty. The one question to ask at your next demo.

Every software demo you've ever sat through was staged.

Not a lie exactly. But think about it. Every field on that screen is filled in. Due dates, owners, who's waiting on what, little colored status pills, the whole thing humming along. It looks incredible. It looks like what your shop would feel like if you just got organized.

Here's what nobody shows you. That same screen in month nine, at a real company, with real people who have real jobs to do.

I can show you that one, because I built it.

I sell promotional products. Shirts, mugs, banners, the stuff with your logo on it. I'm also the guy in the back pulling boxes when a job is about to go sideways, which is most of the reason I know anything worth writing down. A while back I got tired of losing jobs across email and Slack and spreadsheets, so I started building software at night to fix it.

Which means I've got admin on a real system, running a real business, for the better part of a year. So a couple weeks ago I stopped building and just went and looked.

If you only steal one sentence from this whole post, steal this one: does the software fill itself, or does your team have to feed it? Everything below is just me proving why that question matters.

What I found

I'm giving you ratios instead of raw counts, because the raw counts would let you work out whose shop this is.

Across every live job in the system, the number with a back-planned deadline chain filled in was zero. Not a handful. Zero.

Every purchase order in there. Supplier ship date entered on zero of them. That's the single most useful date in my entire business and nobody ever typed one.

"Who's waiting on this right now," which is genuinely the field I'd save if the building was on fire, was filled in on about one percent of jobs.

Line items on quotes. There's a whole table for it. Zero rows. Ever.

And there's a mockup generation feature, six database tables deep, that I built and shipped and was proud of. Zero rows across all six. Nobody used it once.

Nine months. About a dozen people. All of them good at their jobs.

The part that saved me

Same system. Same nine months. Same dozen people.

We run a proof approval loop. Client gets sent the artwork, client approves it or asks for changes. Ninety eight percent of those got a decision. Median time to decide, about six minutes. Every single one has the actual file attached. Nobody trained anybody on it. It worked the first week.

Our art request board has over a hundred tickets on it. Every single one arrived through Slack. Not one was ever created inside the tool itself, and it turns out that's the feature, not the problem.

So I've got one system where everything works and nothing works, at the same time, with the same people.

The only difference

The stuff that works reads Slack and asks nobody for anything.

The stuff that's empty waits for a human to type.

That's it. That's the whole difference. It isn't training, and it isn't buy-in. I had all the accountability in the world and I still had zero.

Because here's what actually happens. It's 4:40 on a Thursday, a client is on hold, a box is going out the door, and a dropdown asking you to pick a waiting-on party is competing with a person standing in front of you. The person wins. The person always wins. Every time, for nine months.

One more thing, and this one is embarrassing

While I was in there I found out our assignment feature had never worked.

Not unpopular. Broken. There was a box where you typed somebody's name, and it saved the name but never saved the actual person. So the "my queue" filter, the one that shows a designer their own work, returned nothing. For everybody. For nine months.

Nobody reported it. You know why? Because a broken feature and an ignored feature look exactly the same from the outside. Empty is empty. I only caught it because I went and looked at the database instead of the screen.

Well, now you know.

The question to ask at your next demo

Ask them this. Does this fill itself, or does my team have to feed it?

Then watch the room. Watch whether the answer is an integration or a training plan. If it involves your people remembering to do something new, you already know how that field looks in month nine. It looks like mine did.

The follow up is just as good. Can you show me a real account at month nine, at a customer my size? Not the demo tenant. A real one. Most vendors can't and won't, and why they won't is the entire point of this post.

Now, I run a promo shop, so my examples are purchase orders and art proofs. It doesn't matter what you do. If you run an HVAC crew, an IT shop, a dental office, a landscaping outfit, it's the same thing. Go pull up the software you're paying for right now and check three fields that somebody has to type. I'd bet real money on what you find, and I'd bet it because I already lost that bet myself.

What I'm doing about it

I'm not writing this to tell you my thing was broken and then leave it there.

Everything I build now has to pass one rule. A feature ships only if it can populate itself. If it needs somebody to type something new, it doesn't get built. It gets pulled from somewhere the work already happens, or it doesn't exist.

The dead fields are getting deleted. Not fixed. Deleted. Nine months of zero is an answer.

The system in this post is SIPS, the software I run my own desk on. If you run a promo shop and want to see what self-filling looks like on a real order, that page walks through it. And if you just came for the demo question, take it, it's free, and it works on any software anyone ever tries to sell you.

I'll write up what happens.