Skip to content
NEW: Martech, explained →

Martech, explained · Managing your stack · 1:42

How to change your stack without breaking it

One field renamed in the CRM, three tools broken by Monday: why stacks break in small changes nobody traced, and how a change gets seen before it ships

Embed this film

Paste this where the film should play. It links back here.

The short answer

Most stacks break in small changes nobody traced. Before a change, follow the data downstream: what reads from the tool, and what it does with it. Write the change down, send it to the owner of every tool it touches, agree a way back, and pick the moment: not late on a Friday, not before a big send.

Questions this answers

Why do stacks break on small changes?

Because a small change is rarely traced. One field renamed in the CRM looks sensible on its own; nobody knew three other tools read that field, so the welcome emails went blank, a dashboard read zero and the ad platform stopped getting new customers.

What does tracing a change mean?

Following the data downstream before the change. Ask what reads from this tool and what it does with what it reads. Two boxes joined by a line only say they are connected; draw which way the data flows as well, so you can see which tools the change reaches.

What goes in a change request?

What changes, why, and which tools it touches. Then it goes to the owner of each of those tools, who know what breaks on their side when you might not. A written request gets read before the change ships; a broken field gets found after.

What is a way back?

An agreed undo before the change goes out: how to reverse it, and who decides when to. It is agreed first because the people deciding under pressure on a broken Monday rarely agree on anything quickly.

When should a change ship?

At a moment you pick, not at the end of the week and not the day before a big send. The same rename, with three owners told, the field mapped in each tool and a chosen moment, makes a quiet Monday instead of one spent finding out why.

Transcript

Every chapter as text. Click a time to play the film from there.

0:00 The Monday after

It's Monday. Last week, someone renamed one field in the CRM.

By the time the team logs in, the welcome emails are blank and a dashboard reads zero. The ad platform has stopped getting new customers.

The change was small and sensible. Nobody knew three other tools read that field.

Most stacks break this way. Not in big projects, but in small changes nobody traced.

0:30 Trace what depends on it

Before a change, follow the data downstream. Ask what reads from this tool, and what it does with it.

Two boxes joined by a line tell you they're connected. Draw which way the data flows, too.

0:44 Write it down

Then write the change down: what changes, why, and which tools it touches.

Send it to the owner of each of those tools. They know what breaks on their side, and you might not.

A written request gets read before the change ships. A broken field gets found after.

1:03 A way back

Agree on a way back before it goes out: how to undo it, and who decides when to.

Then pick the moment. Not at the end of the week, and not the day before a big send.

1:16 Before and after

Before: one rename, three broken tools, and a Monday spent finding out why.

After: the same rename, three owners told on Wednesday, the field mapped in each tool, and a quiet Monday.

The change didn't get any smaller. It got seen.

Learn more at Martech Stack Builder, or build your own.