Customers

Twenty-seven of forty-one

We are called into stalled ERP implementations regularly, usually on somebody else's product. Twenty-seven of forty-one recovered to a working state. The fourteen that did not are the more instructive number, and they had things in common.

Is yours in trouble?

Tell us where the project is and what has slipped. We will say honestly whether it is recoverable.

1 / 3
41 engagements, 27 recoveredMedian 11 weeks to recoverMostly on competitors’ products

Customers

What recovery actually involves.

Almost every successful recovery followed the same sequence, and the hardest part is the first step.

1. Stop configuring

The instinct is to push through because stopping feels like losing time. Resuming configuration on the foundation that failed is the single most common reason a rescue does not work.

2. Examine the data properly

Not a questionnaire — an actual diagnostic. Data worse than anyone knew was the primary cause in 24% of stalls and was nearly always discovered during load rather than during a nominal assessment phase.

3. Settle requirements against transactions

Derived from a quarter of real data rather than reconstructed from the original interviews. This is where the surprises that stalled the project originally get surfaced deliberately.

4. Name someone who can decide

One person per area, able to decide across departments within a day. No internal owner with authority was the primary cause in 19% of stalls.

5. Re-baseline honestly

A new plan built on what is actually true, including what has to be thrown away. Customers find this harder than the technical work, because it makes sunk cost explicit.

6. Then resume

Median 11 weeks from engagement to a working state. The ones that took longest were those where the organisation wanted to skip to this step.

The fourteen we could not save

These are more useful than the twenty-seven, and they clustered into three patterns.

The organisation could not decide

Six of the fourteen. Not a lack of engagement — sponsors attended every meeting — but no single person who could settle a cross-departmental question about approval thresholds or chart structure. Every decision escalated and stalled again.

A rescue cannot supply organisational authority. Where we identified this and it did not change, we withdrew rather than continuing to bill against a project that was going to stall again.

The sunk cost was unspeakable

Four. The project had consumed enough budget and political capital that re-baselining honestly was impossible — admitting that eighteen months of configuration had to be discarded was a conversation nobody in the room could have.

These usually ended with the project continuing on its original path and failing again some months later.

The product genuinely did not fit

Four, and this is the rarest cause by some distance — consistent with our finding that software incapability is the primary cause of only about 4% of stalls generally. In each case the mismatch had been visible in week one of the original selection and nobody had wanted to say so.

A rescue can fix data, requirements, and sequencing. It cannot supply an organisation with someone empowered to decide, and that is what killed six of the fourteen.

What a rescue costs

We will not quote a rescue without looking first, because nobody honestly can. A paid assessment establishes what state the existing work is in, what is salvageable, and whether the causes are addressable at all.

That assessment produces a written recommendation, and roughly one in five concludes that the project should be stopped rather than rescued. We would rather deliver that conclusion in week two than bill against something that will stall again.

Rescues themselves start at $15,000 and are quoted after the assessment on a fixed scope with written exclusions.

What we do not do

  • Blame the incumbent partner. It is tempting and usually inaccurate — three quarters of stalls trace to requirements, data, or decision authority rather than to delivery.
  • Recommend switching to our product by default. Most rescues end with the customer staying on the system they bought. Changing vendors addresses about a quarter of the actual risk.
  • Take the engagement when the cause is organisational and unchanged. We have withdrawn from rescues for exactly this reason.
Raise it early

Of the forty-one, the ones that recovered fastest were those where somebody raised the alarm in month three rather than month eight. The failure patterns are visible early and the cost of addressing them compounds. If your project feels wrong and you cannot say why, that instinct is usually worth acting on.

Questions

Common follow-ups.

How many rescues succeed?
Twenty-seven of forty-one recovered to a working state. Median eleven weeks from engagement.
Why did the others fail?
Six because the organisation could not supply someone empowered to decide, four because sunk cost made honest re-baselining impossible, and four because the product genuinely did not fit.
Do rescues mean switching to your product?
Usually not. Most end with the customer staying on the system they bought, because changing vendors addresses only about a quarter of the actual risk.
Will you quote before assessing?
No, and nobody honestly can. A paid assessment establishes what is salvageable, and about one in five concludes the project should be stopped rather than rescued.
When should we call?
Earlier than feels comfortable. The projects that recovered fastest were the ones where somebody raised it in month three rather than month eight.

The fourteen are the instructive number.

A rescue can fix data, requirements, and sequencing. It cannot supply an organisation with authority to decide.