The Wrong Diagnosis

When an engineer leaves a project mid-delivery, the standard response from IT consulting firms is to tighten screening. More technical rounds. Tougher culture interviews. Sometimes a probation period for the next hire. The reasoning is intuitive — if the person was wrong for the role, the next one needs to be more right.

Across 30+ enterprise delivery projects, this response almost never solves the problem. The exits keep happening. The screening keeps tightening. The pattern continues until the firm concludes the talent market is the issue.

The talent market is not the issue. The diagnosis is wrong.

What Actually Happens Before a Mid-Project Exit

In most cases observed, the engineer who left was not a bad hire. They were a competent engineer who ran out of information four to six weeks before the exit became visible.

The pattern is structural and consistent:

• Week one to three — engineer onboards, output is normal, both sides feel positive

• Week four to six — engineer encounters friction the SI client does not see, raises it once or twice in muted form, gets ambiguous response, stops raising it

• Week six to ten — engineer's effort decreases gradually, SI client notices but interprets it as a productivity issue rather than a relationship issue

• Week ten or later — exit conversation happens. Both sides are surprised. Neither was operating on the same picture.

By the time the exit is announced, the decision was made weeks earlier in a conversation that did not happen.

The Information Asymmetry That Makes This Inevitable

The deeper problem is that engineers working inside an SI client's project rarely raise concerns to the SI firm managing them. The relationship carries evaluation risk. Raising a concern looks like complaint. Looking like complaint affects the next project assignment.

The SI firm rarely raises performance concerns directly to the engineer either. The relationship carries morale risk. A direct concern can be heard as criticism. Criticism can lead to disengagement before the project ends.

So both sides operate on a sanitized version of reality. The engineer reports that things are fine when they are not. The SI client interprets normal-looking signals as evidence everything is on track. Neither is being dishonest. Both are managing a relationship under structural constraints that prevent direct information flow.

The mid-project exit is the moment that information finally surfaces — and by then the project has absorbed weeks of unaddressed misalignment.

Why Better Screening Cannot Fix This

Screening assesses what an engineer brings into a project. Mid-project exits are caused by what happens to that engineer during a project.

You can hire the most carefully assessed engineer in the market and still see the same exit pattern if the structural information problem is unchanged. The engineer will encounter the same friction. The same evaluation risk will prevent them from raising it. The same morale risk will prevent the SI firm from probing it. The exit will look identical to the previous one.

This is why firms that respond to mid-project exits with tighter screening tend to see the pattern continue. They are solving for the wrong layer.

What Actually Reduces Mid-Project Exits

The intervention that works is not at the hiring stage. It is at the governance layer between hire and delivery.

Three structural changes correlate with significantly reduced exit rates across the projects observed:

1. A third party who carries no evaluation or morale risk

When information has to flow between two parties who both carry relationship risk with each other, it does not flow. Adding a third party — one whose role is specifically to surface friction without affecting either side's ongoing relationship — restores the channel.

This is not a layer of management. It is a layer of observation. The third party's job is to ask both sides what they would not say to each other and to surface patterns before they accumulate.

2. Friction is named early as data, not as failure

When week-four friction is treated as routine project signal rather than a warning of failure, both sides can engage with it without defensiveness. The framing change is small but consequential. "Where is this project encountering friction this week" generates information. "Is anything wrong" does not.

3. Exit signals are tracked deliberately

The signals that precede a mid-project exit are observable: declining response time, narrower scope of comments in standups, lower question frequency, increased delegation back to spec. These are not subtle. They are simply not being looked for.

Projects where these signals are explicitly tracked — not as performance metrics but as relationship health metrics — surface friction earlier and address it before it compounds.

The Implication for Hiring Strategy

None of this argues against careful screening. Tier-appropriate assessment matters. Project fit matters. The point is that screening solves a different problem than mid-project exits.

Firms that experience repeat mid-project exits and respond by hiring more carefully are not addressing the actual cause. They are applying a hiring solution to a governance problem. The exits will continue until the governance layer is built — at which point the original screening process will start delivering significantly better project outcomes, because the engineers who would have left will stay.

The exit is not the problem. The silence was.