The Spaghetti Test

What a vendor does on your worst morning tells you what edtech vendor trust is really worth.

I was in the car with my family, coming home at the tail end of spring break, when my phone started going off. Texts. Emails. Help desk tickets stacking up on a Sunday afternoon. Nobody could log in, and school started in about sixteen hours.

The reason was almost comical. Our licenses had not been renewed in the vendor’s own console. Somebody on their side had simply forgotten, because we were in the middle of a long service credit and there was no invoice attached to us anymore.

When I finally reached the support executive I had on speed dial, she told me she was cooking spaghetti for dinner and would get to it when she was finished.

And that was the moment I started planning to leave. And it wasn’t the outage. I had already survived a worse one.

How we got there

Two years earlier, I had done everything right. I researched. I vetted with other CTOs whose judgment I trusted. I budgeted carefully for a system that touched every teacher, every student, and every morning of the school year. It was expensive, and it promised to automate an enormous amount of manual work for my team that did not have hours to spare.

Implementation was rocky. Everything took roughly twice as long as promised, and most of the extra labor landed on my staff rather than theirs. Features went live without anyone telling us. Changes appeared in our environment that we had not asked for and did not know about until a teacher called.

We got it working, mostly. Only part of the automation ever functioned. We kept doing the rest by hand.

Then came the first day of school, and nobody could log in.

The vendor did not respond that day. Not to me, not to my team, not to the tickets. Teachers were upset. Students were upset. Administrators were upset, and they were right to be.

When I finally got a call the second day, I kept it civil, but I told them something I still believe. A leader with less support than I had could have lost a job over that morning. My superintendent understood that a vendor had failed to show up and that this was outside my control. Not every leader has that superintendent.

What edtech vendor trust is actually worth

They responded the way struggling companies often do. Biweekly calls. More names on the account, including a vice president of something something. Eventually they offered us years of free service, which tells you everything about how badly it had gone.

I took it, and I want to be honest about why.

We had sunk real money and real staff time into the thing. Walking away meant eating all of it. But the bigger cost was political. I had championed this solution inside the district. I had stood in front of teachers and principals and told them it would make their lives easier. Leaving meant walking that back in public, in front of people I had asked to trust me.

So we stayed. We moved to their cloud product because they said that would fix it, and we paid an upgrade fee to do it. Things still broke almost every week. I spent a highly suspect share of my working life babysitting an account that was, on paper, free.

Free was the trap. It took away my financial reason to leave and gave them no financial reason to improve. It also quietly moved us to the part of their system where nobody was watching, which is how a license renewal gets missed on the Sunday of spring break.

Concessions are a way to buy patience. They are not the same as fixing the problem, and we should stop accepting them as though they were.

What it looked like when it worked

The vendor I eventually moved to had been talking with me for two years before I bought anything.

They stayed in touch when there was nothing to sign. No pressure, no quarterly check-in disguised as a favor. When the time finally came, they asked me directly what hurt, and then they built their demos around those specific pains for my tech staff and for our Tech Leadership Cohort of teachers. The people who would live with the system got to interrogate it first.

Their standard rollout was twelve weeks. I told them I needed twice that, because I was not just installing software. I was rebuilding the confidence of staff who had spent two years being told to trust something that did not work. They agreed without argument, and they did not charge me for the additional twelve weeks.

Trust was established before implementation ever started. It came from three decisions that cost them almost nothing: stay engaged without selling, ask before demonstrating, and let the district’s timeline win over the internal playbook.

A partner who lets you go slower is telling you they expect to be around a long time.

What I would tell my earlier self

Slow and messy is a capability problem. Companies can grow out of those, and I have watched good ones do it. Changing a district’s environment without telling anyone, and going silent on the first day of school, is a different category, and that one rarely improves.

Decide your walk-away line while the system is still working. Write down what you would have to see to end it, and tell someone else what you wrote. If you skip that step, the decision still gets made. It just gets made later, in a car full of tired kids, by whoever happens to be standing over a pot of noodles.

I sit on the other side of the table now, which is not lost on me. I spent years as the person whose first day of school depended on somebody else picking up the phone, and I would rather help build the partner I kept hoping for than pretend I never needed one.

Every vendor will eventually have a bad morning. What you are really buying is their behavior on that morning. Pick the ones who answer before the water boils.