After a failed digital initiative, a stronger forecast is rarely enough. The board, delivery teams and affected colleagues need evidence that the firm understands the previous outcome and has changed the conditions that caused it.
This is a trust and governance problem before it is another procurement exercise.
Define the failure without rewriting history
Recover the original objectives, baseline, scope, budget, forecast changes, contracts, decisions, assurance and outcome evidence. Speak to sponsors, users, delivery teams, suppliers and control functions.
Separate several possible failures:
- the problem or expected value was misdiagnosed
- scope or requirements remained unstable
- technology could not meet a material need
- data or integration foundations were missing
- governance delayed or obscured decisions
- delivery quality or supplier capability was weak
- the firm withheld time, access or ownership
- adoption and operation were unfunded
- measurement could not show the outcome
- external conditions changed the case
A project can launch and still fail commercially. It can miss a date while producing valuable capability. Define the gap between intention and result precisely.
Avoid a “lessons learned” workshop that produces generalities such as communicate more. Trace causes and contributing conditions with evidence. Record disagreements and uncertainty.
Allocate responsibility fairly
Blaming the previous supplier entirely is rarely credible. Taking responsibility for everything can conceal contractual or delivery failures that must be addressed.
For each material issue, state who controlled the condition, what was known at the time and which action was or was not taken. Distinguish an unforeseeable event from a risk that was raised and ignored.
This is not an exercise in public fault. It is needed to decide which safeguards actually change the next attempt. A different vendor will not fix absent client ownership; a new steering committee will not fix an unsuitable product.
Consider independent review where the previous sponsor or supplier cannot credibly assess its own decisions. Give the reviewer access to evidence and relevant people, with a clear question.
Close old obligations before proposing new ones
Identify unresolved data, security, licence, intellectual-property, access, support and contractual issues. Confirm who operates what remains and whether clients or colleagues are still affected.
Decide which assets can be retained: research, content, data work, components, integrations, training or process knowledge. Sunk cost cannot justify continued use. Discarding sound work, however, increases the loss.
Update the financial record. The board should see actual total cost and remaining liabilities, not only the original project budget.
If serious incidents or professional duties were involved, complete the appropriate legal, regulatory and risk processes before using the project as a new investment story.
Show what has structurally changed
For each root cause, provide an observable change:
| Previous condition | New evidence or control |
|---|---|
| No outcome owner | Named service owner with time and authority |
| Unbounded scope | Prioritised scope, exclusions and change route |
| Weak data | Profile, remediation and acceptance thresholds |
| Late risk review | Risk owners involved in design and gates |
| Supplier overclaim | Buyer-controlled proof and contractual criteria |
| No adoption capacity | Representative users and funded support |
| Hidden forecast movement | Forecast-to-complete and escalation tolerance |
Do not claim a control is new if it exists only in the proposal. Confirm people, capacity, funding and governance.
Where the causal diagnosis remains uncertain, the next step should test it rather than assume it.
Revalidate the need
The original problem may have changed. Conduct fresh client research, operational analysis, platform assessment or risk review. A failed solution does not prove the need has disappeared, and it does not prove the same investment remains worthwhile.
Compare current options, including improve, replace, contain, defer or stop. Use current costs and market evidence. Peer examples can show possibilities; they cannot establish that another firm's result will transfer.
Build the cost of delay from the firm's evidence. Avoid generic maintenance percentages, breach costs or competitor fear.
Build evidence before a large commitment
A diagnostic, technical proof or bounded service phase can reduce a critical uncertainty. It should not be a demonstration chosen because success is easy.
Define the question, representative conditions, measures, risk controls, cost and decision date. Preserve a genuine stop option. If the first phase depends on later spending to become usable, show that commitment transparently.
Evidence may concern feasibility, data quality, user task success, operating effort or governance. It does not have to produce a commercial uplift within weeks to be valuable.
The board should understand what the phase can and cannot prove.
Address affected colleagues before asking for enthusiasm
People who contributed to an unsuccessful programme may reasonably protect their time. Explain the diagnosis, acknowledge wasted effort and show how participation will differ.
Ask what they need to work responsibly, then allocate it. Do not rely on the same overstretched subject experts under a new project name.
Create a route for dissent and early warning. Colleagues should not have to endorse the business case to report a risk or participate in testing.
Internal trust grows through accurate forecasts, kept decisions and visible response to problems, rather than a renewed launch campaign.
Prepare the board decision openly
The new proposal should begin with:
- what the previous initiative intended and delivered
- the evidenced causes and responsibility
- obligations closed and assets retained
- the current need and options
- changes to scope, governance, capability and assurance
- the present commitment and later exposure
- success, stop and gate criteria
Provide the paper through normal governance and invite challenge before the meeting so missing evidence can be corrected. Do not assemble private allies to manufacture public consensus. Directors should receive the same material information.
Record material dissent and conditions in the decision.
Choose timing by readiness
There is no fixed cooling-off period. Return when the root-cause work is credible, unresolved harms are handled, required people are available and the current need justifies attention.
If the firm is not ready, set a dated preparation plan: close old access, establish ownership, gather evidence or address a foundation. “Wait until people forget” preserves the trust deficit.
Conversely, urgency does not excuse skipping the review. A deadline built on the same unexamined conditions is repetition, not recovery.
Make the second attempt observably different
The difference should live in operating behaviour: decisions made on time, bad news surfaced, scope traded explicitly, users involved, assurance completed and the service owned after launch.
Report against those leading conditions as well as outcome measures. If they deteriorate, intervene before the project reproduces the previous pattern.
The aim is not to erase the previous failure or promise that another cannot occur. It is to show that the organisation learned something specific enough to change how it invests and delivers.
If you want support structuring that conversation - including an honest diagnosis of what went wrong and how to present a genuinely different approach - that's exactly the kind of work we do. Book a recovery planning session and we can work through it together. An outside perspective is useful here, but not because it brings a new strategy. It's useful because someone who's sat in that room before can tell you: this is fixable.
