Who owns the integration?

Between two perfectly good systems there was a small piece of machinery that moved records across every night. Built during the rollout, by a contractor, over a weekend. It had run without complaint for two years, which is precisely why nobody could remember it existed.

When it stopped, the first vendor demonstrated that the data had left their system correctly. The second demonstrated that nothing had arrived. Both were telling the truth, and neither was responsible, because the thing that had failed lived in the gap between two contracts. It took nine days to sort out, and about six of those were spent establishing whose problem it was.

I have come to think of ownership as a decision rather than a discovery. Somebody has to be named. It can be one of the vendors, if you pay them for it and it is in writing. It can be a person on your team, if they have the access, the documentation and an hour a week that actually exists. It can be me. What it cannot be is nobody, and nobody is the default.

I now ask about this in the first week, and I ask for a list: every integration, what it moves, when it runs, who gets called, and how you find out it did not run. Most companies can answer the first four. The fifth column is nearly always empty, and silent failure is the expensive kind.

MMB