Fix or Rebuild a Failing Software Project?

If your software project keeps missing dates, ask to see what works. A fresh promise or a new design is not enough to decide.

You need to know what can be kept and what is left. Then compare the next steps before spending more money.

Warning signs call for a review.

Look for the same tasks staying almost finished for weeks. Other warning signs include new features being added before earlier work is tested.

Vague updates and slow decisions also need attention. These signs do not tell you whether the cause is planning, faulty software or both.

Ask for a working version and a clear list of what is left. Agree who can make decisions and who will answer each open question.

Show what works from start to finish.

Choose a task the software is meant to help someone complete. For a booking system, that could mean choosing a time and getting a booking email.

Try the whole task with test details. Record where it stops, what gives the wrong result and what works.

Ask what has been tested and what still needs checking. A good-looking screen does not prove the steps behind it work.

The GOV.UK guide to showing work and tracking progress describes regular demonstrations and an ordered work list. These give you something clearer to discuss than a percentage complete.

Agree the smallest useful result.

Write down the job the software must do for the business. Separate the features needed for that job from those that can wait.

For example, a booking service might need confirmed appointments before adding loyalty points. This is an example to test against your needs, not a fixed launch rule.

Write what must happen for each part to count as finished. Include faults to fix and checks needed before anyone relies on it.

Choose one person to decide what gets done first. Record decisions so that a new request does not quietly change the agreed result.

Compare the work ahead for each option.

Ask whoever reviews the software to explain the facts behind their advice. If you need a second view, seek someone who can check the work so far.

Compare a repair and a new build against the same agreed result. Include:

• The work still needed, with estimates and clear unknowns.
• The parts that can be kept and how they were checked.
• Any records that must move, and how they will be checked.
• The effect on people already using the software.
• Ongoing bills, support and who will handle faults.

Money already spent explains the pressure you feel. Your next choice still needs to account for the costs and risks ahead.

Fixing may make sense when useful parts work and the faults are clear. To replace part or all, first check that the new plan can meet the same needs better.

Also check whether an existing product could do the job. A new custom build is not the only possible alternative.

Set a clear point to decide whether to continue.

Agree the next piece of work and how you will check it. Review the result before you agree to the next step.

If the result is no longer worth the work, consider pausing or stopping. The GOV.UK guide to deciding whether a service should proceed treats stopping as a valid outcome of research.

Our contract product owner service helps you set priorities and agree what finished work should do. Bring the current plan and the questions you cannot get answered.

The first conversation is free. Work starts after a non-refundable deposit.

Talk to us about your site.

One of us will get back to you the same day, or the next working day at the latest.