A COO sits in front of a decorative Gantt chart, a six-figure spend, a delivery partner sending apologies and an internal sponsor who moved role three months earlier.

“Do we just start again?”

The source places this scene at a 250-person consultancy. Verify the engagement and quotation before publication. It frames the recovery question well because sunk cost and frustration make a binary answer attractive.

There are at least three paths: recover with a different model, restructure what has value, or close the programme and begin a smaller or different response. The decision depends on evidence and organisational conditions, rather than the desire for a clean narrative.

Re-test the goal

Ask each critical stakeholder to state the original problem, intended users and outcome. Differences reveal whether the programme drifted or began without shared intent.

Then examine what changed in the business, user need, regulation, technology and economics. A sound goal can become obsolete. A fashionable solution can remain in the plan after its rationale disappears.

Record:

  • Original problem and evidence
  • Current problem and evidence
  • Outcomes still worth pursuing
  • Guardrails and constraints
  • Decision that recovery must enable

Do not treat continuation as the neutral option. It is a new investment decision.

Understand why accountability failed

Identify the individual who was accountable, their authority and the decisions the programme required. A committee can oversee; it cannot substitute for an empowered outcome owner.

Look at client and supplier conditions:

  • Were decisions made when needed?
  • Did the team have access and internal capacity?
  • Were changes governed?
  • Did the supplier surface problems with options?
  • Did commercial terms reward the wrong behaviour?
  • Were assurance and user evidence available?

The source asks the most important uncomfortable question: what did the organisation do that made success harder? Preserve it. Recovery that attributes every cause to the supplier will reproduce unavailable stakeholders, shifting requirements or diffuse authority.

The companion article on why transformation fails examines those patterns.

Value what exists without worshipping sunk cost

Inventory deliverables and evidence:

  • Research and user insight
  • Data and content models
  • Code, integrations and environments
  • Processes and training
  • Supplier contracts and licences
  • Decisions, risks and lessons
  • Content or data already migrated

Assess quality, supportability, rights, security, fit and cost to complete. Something can be technically competent and wrong for the current goal. Something incomplete can still contain valuable research or a reusable integration.

The source describes a financial-services migration at “70% complete” that mainly lacked content authority. Percentage complete is dangerous when tasks carry different value and risk. Verify the case and inspect the remaining path before deciding it needs only governance.

Choose the recovery path

Restart with a different model when the outcome remains valuable and existing delivery or governance cannot support it. Change more than the supplier name: ownership, scope, evidence, commercials and gates may all need reset.

Restructure and retain useful work when core assets are supportable and the main constraints can be corrected. Re-baseline time, cost and outcome instead of preserving the old plan as fiction.

Stop and redefine when the problem is wrong, the remaining value is insufficient or risk and cost exceed the case. Preserve learning, access, data and contractual evidence. Close responsibly.

A fourth path can be deliberate pause when an external dependency or organisational event has a credible resolution date. Assign an owner and review point so “pause” does not become abandonment.

Reset governance around decisions

The recovery owner needs a fresh written mandate. Changing the person can help, though the source's preference for a new owner is not universal. Continuity may be valuable if authority and behaviour genuinely change.

Create a short cadence around:

  • Evidence since the last review
  • Decisions due
  • Highest consequential uncertainty
  • Scope, cost and risk movement
  • Client-side capacity
  • Next gate

Meeting length and frequency should respond to risk. A 30-minute weekly or fortnightly conversation is a useful pattern, rather than a rule.

Use a handful of outcome and guardrail measures. The source's law-firm example includes enquiry completion, publishing time and CMS tickets; verify the project before publishing. Measures need a current baseline and owner.

Earn confidence with a meaningful early result

The source insists on a visible deliverable in 30 days. A bounded result can rebuild trust, but the calendar should reflect the work.

A credible early result:

  • Connects to the core problem
  • Can be completed and assured safely
  • Produces usable value or important evidence
  • Avoids throwaway architecture
  • Demonstrates the new governance

The source's rebuilt enquiry form and automatic acknowledgement allegedly restored confidence in a stalled onboarding programme. Verify the case. Its lesson is strong: a homepage flourish would have been tokenistic because it did not touch the failing handoff.

Some recoveries should first produce an evidence result, such as a tested migration sample or a resolved architecture decision. Do not deploy an unsafe feature merely to create visible momentum.

Manage the human aftermath

Recovery affects three groups differently.

Original backers need a factual account that separates the continuing value of the vision from delivery and governance failures. Avoid language chosen merely to protect status; responsibility still needs naming.

Sceptics may hold evidence the original programme ignored. Give them a defined route to challenge and help test the new controls, without granting an informal veto.

Board or partnership needs causes, structural changes, revised range, decision gates and a clear account of what further funding buys. The related article on board approval for digital budgets can prepare that case.

Delivery teams and affected users also need attention. Repeated resets create fatigue. Explain what is changing, which work is preserved and how new decisions will be made.

Recovery is a new proposition

The source admits Distinction has been close to programmes that stalled. That candour should remain. It avoids positioning recovery as expertise acquired only from other people's mistakes.

The collection's five warning signs article can help identify which signals were missed. The source also links “the next piece” to that same warning-sign article in a way that misdescribes it as a detailed recovery case; keep the useful link and correct the promise.

No framework guarantees recovery. A defensible plan says why the goal remains valuable, what can be retained, which failure conditions have changed and what evidence will govern the next commitment.

The right response to “do we start again?” may be yes, no or partly. The quality of recovery lies in making that answer from current evidence rather than sunk cost, blame or the comfort of the old chart.