The First Month After Go-Live, Honestly

Nothing goes perfectly and that is fine

A first month with zero surprises usually means the automation is not handling much. What matters is whether the surprises are contained, visible and quickly resolved.

We staff the first fortnight accordingly rather than treating it as the end of the project.

1. Edge cases nobody mentioned

The customer with two delivery addresses. The supplier who invoices in a different unit. The order that arrives as a photograph. Month-end, which behaves unlike every other week.

Parallel running catches most of these, which is why we insist on it. The ones that survive parallel running are usually seasonal, and they surface later.

2. The review queue is bigger than expected

Initial confidence thresholds are set cautiously on purpose. That means more items route to a person in week one than in week eight.

We tune during the first month with real data, and the rate falls. A client who was told to expect 10% and sees 25% in week one has not been misled — they have been protected.

3. An integration behaves differently under volume

Rate limits that never appeared in testing. A third-party endpoint that is slower at 9am. A batch that times out at month end because the volume is ten times a normal day.

This is why we build retries, backoff and queueing from the start rather than adding them when it breaks.

4. Throughput dips while people learn

• Two to four weeks of reduced output is normal even when the system is good
• Warn the team in advance so it reads as expected rather than as failure
• Support volume roughly doubles for a fortnight
• Ticket topics in week one are the most useful feedback you will get

What we do about it

Someone available daily for the first week, a defined route for reporting problems, a weekly review of what routed to review and why, and threshold tuning as evidence arrives.

And a stop button, in your hands, that pauses the automation in seconds without calling us.

Frequently asked questions