Is your automation running on current data? A 15-minute source check

Most automation checks ask whether the thing ran. This one asks a different question: whether what it ran on was current.

That gap is easy to miss because nothing looks wrong. The job fires on schedule. The log says success. The dashboard is green. And the list it read was exported in August.

Every check you already have asks "did it do what it was told?" — and the answer is yes. None of them asks whether it was told the right thing. That is the whole failure, and it is invisible from the inside.

Budget fifteen minutes. Do it whenever you change a source, and once a quarter regardless.

1. Write down what each flow actually reads

For your three or four most consequential automations, name the source in one line each — not the tool, the source:

• the segment or saved view it queries
• the spreadsheet, export or file it opens
• the list it was pointed at when someone set it up

If you cannot name the source for a flow, that is the finding. Stop here and go find it.

2. Ask when that source last changed — not when the job last ran

These are different dates and only one of them is usually on screen.

| What the dashboard shows | What you need instead | | ------------------------ | -------------------------------------------------- | | Last run: this morning | Source last updated: ? | | Status: success | Rows added since the source was created: ? | | Connected | Is this still the list someone would choose today? |

A job that runs hourly against a file nobody has touched since spring is a healthy-looking job.

3. Count the source against reality

Pick the one number you would notice being wrong, and compare two places:

• how many people are in the segment the automation reads
• how many are in the place you would look if a colleague asked

If those disagree, you have found it. Do not reconcile the numbers in your head — write both down with today's date beside them, because a difference you explain away is the one that persists.

4. Check the defaults nobody chose

When a tool is set up, it fills in a source for you. That default then reads like a decision someone made, which is why it never gets audited.

• Which flows are pointed at "all contacts" because that was prefilled?
• Which report reads last quarter's date range because that was the range when it was built?
• Which export path still has a year in the filename?

A default is an assumption wearing a decision's clothes. It gets the benefit of looking deliberate.

5. Make the source visible where the result is

The durable fix is not remembering to check. It is putting the source's own date next to the output, so the next person sees both without asking:

• the segment name and its last-updated date in the campaign summary
• the export date in the report header, not just the run date
• the list size in the confirmation the flow sends you

If the freshness of the input is not visible beside the result, someone will read the result as current because the machinery is.

Why this is worth fifteen minutes

The expensive version of this failure is not an outage. It is a month of correct, well-monitored, perfectly logged work aimed at the wrong people — the campaign that reached a list from before your best quarter, the follow-up that skipped everyone who arrived since.

Nothing in the system reports it, because nothing in the system is broken.