In this guide 6 sections
Watch for five signs. The person differs from the one you interviewed, work lands outside your repository, questions wait a day, nothing is written down, or engineers sit idle without priorities. Some causes sit with the supplier and some with you. GullySystem’s arrangements are set up to make each of these visible early.
The person on the call is not the one you met
You interviewed one engineer. A different one joined the first stand-up, and nobody explained why. Their experience does not match what was agreed.
This is the clearest early warning. Ask the supplier directly what changed. A sound arrangement tells you before a change, lets you meet the replacement and arranges an overlap.
Work that never shows up in your repository
Open your own source control and look for commits under your branching rules. If work sits on a supplier server or arrives as zip files, you do not hold what you are paying for.
Check the cloud and deployment side as well. Pipelines, credentials and app store listings belong in accounts you control.
- Commits in your history, not a copy sent later
- Deployments running from your cloud account
- Credentials stored where you can rotate them
- Tickets moving in your own tracker
Every question takes a day
A question goes into an overnight queue. The answer comes back half understood, and another day passes. The hourly rate then says little about what gets finished.
Look at the overlap hours you agreed. If they are not happening, raise it. If they were never agreed, set them now and write down which work runs without live answers.
Nothing is written down
Ask to see the runbooks, architecture notes and decision records. If none exist, knowledge is sitting in one person’s head.
That is risky even when the work is good. When the person leaves, only source code remains. Nobody can say why a service is configured as it is.
- A runbook for each deployment step
- Notes on how the main parts connect
- Short records of why key decisions were made
Check your side of the arrangement too
Not every failure is the supplier’s. Engineers without a groomed backlog wait or guess. Unclear priorities produce busy weeks with little output.
Ask who on your side directs the work each day. If nobody does, a lead can be added to plan and review, or the work can move to a fixed-scope project.
Then decide. Fix the gaps, replace a person on agreed terms, or wind down with a proper handover.
Augmentation health check
A short sheet with five rows: identity, repository, response time, documentation and direction. Score each one monthly as clear, unclear or missing. Two missing rows in a month call for a conversation with the supplier.
Open a blank worksheet to printQuestions owners ask
Is slow progress in the first fortnight a warning sign?
Not by itself. Nobody is fully productive on an unfamiliar codebase straight away. What matters is a visible ramp-up with an early small change, not a silent fortnight of reading.
Can we replace one engineer without ending the arrangement?
Usually, if the agreement covers replacement. Raise it early, meet the proposed replacement and ask for an overlap so the context passes across.
Will adding more engineers fix a slow team?
No, not when the cause is direction. More people with no clear priorities produce more waiting. Fix the backlog and the daily decision-making first, then look at size.