Why Transformation Projects Fail, and How to Make Change Stick
Why transformation projects fail: seven failure modes with their early symptoms and fixes, how to make change stick, and a pre-mortem checklist to use first.
Transformation projects fail for predictable reasons, not bad luck: advice and build split across different firms, a first scope that is too big, no single owner, measuring activity instead of outcomes, ignoring the people who do the work, data that is not ready, and no run phase after go-live. Each has early symptoms you can spot and a practical fix you can apply before the money is spent.
If you want to understand why transformation projects fail, do not look at the plan. Look at what happens between the plan and the third month after go-live. That is where almost every failed programme comes apart, and it comes apart for the same few reasons in a 50-person clinic group as in a 400-person manufacturer. The good news is that each failure mode has symptoms you can spot early and a fix you can apply before the money is spent. This guide sets out the seven we see most often, how to recognise each one, how to fix it, and a pre-mortem checklist to run before you start.
Why do transformation projects fail? The short answer
Transformation projects rarely fail because the idea was wrong. They fail because of how the work was set up: who designs it, who builds it, who owns it, what gets measured and what happens after launch. Here are the seven failure modes at a glance.
| Failure mode | Early symptom | Fix |
|---|---|---|
| 1. Advice and build split across firms | The build team asks questions the strategy deck never answered | One team, or one accountable lead, from diagnosis to run |
| 2. Starting too big | The first measurable result is more than a quarter away | One function, one metric, a first result in weeks |
| 3. No owner | A steering committee, but no single name on the outcome | A named process owner with real authority |
| 4. Measuring activity, not outcomes | Status reports count workshops and features shipped | A baseline and an outcome metric agreed before work starts |
| 5. Ignoring the front line | Staff first see the change at training | Involve them in diagnosis, design and testing |
| 6. Data not ready | The new reports are wrong on day one | Data clean-up scoped as its own workstream before build |
| 7. No run phase | The project closes at go-live | A funded, staffed run phase of measuring, tuning and training |
The seven failure modes, with symptoms and fixes
1. Advice and build are split across firms
The classic pattern: a consulting firm diagnoses the business, designs the future state, writes a roadmap and leaves. A separate IT vendor or development agency receives the roadmap and builds. At that hand-off, most of the context disappears: the reasons behind decisions, the exceptions that came up in interviews, the constraints nobody wrote down. The builder builds what the document says, which is not quite what the business needs, and each side blames the other.
- Symptoms: the build team re-runs discovery, change requests start in the first fortnight, and nobody can say who decides when the design and the build disagree.
- Fix: keep the people who designed the change involved in building it, ideally as one team. If you must use two firms, appoint one accountable lead over both and have the designers attend every build demo. This is the core argument for full stack consulting, where the same team owns strategy, software and rollout.
2. Starting too big
Big-bang programmes try to change everything at once: a new ERP, a new CRM, a new structure, every function, one go-live date. They take a long time, the business changes while they run, and people lose faith before any value arrives. By the time the system launches, the requirements are out of date.
- Symptoms: the first measurable result is scheduled more than a quarter away, the scope includes the word "everything", and the budget conversation happens before anyone has measured a baseline.
- Fix: start with one function and one outcome metric, and design the first phase to show a result within weeks. A focused workflow, such as a WhatsApp agent that answers and qualifies new enquiries or automated invoice entry into your accounting system, is typically live in two to four weeks. Prove it works, then fund the next phase from what it saves or earns.
3. No owner
Steering committees do not own outcomes. When a transformation belongs to everyone, it belongs to no one. IT owns the system, the consultants own the plan, the vendor owns the go-live, and nobody owns whether the function actually performs better.
- Symptoms: decisions wait for the next steering meeting, the question "who signs this off?" has three different answers, and the CEO is the sponsor in name only.
- Fix: name one process owner for each function being changed. That person runs the function day to day, has authority over how it works, and is accountable for the outcome metric. The CEO backs them visibly, for example by attending the weekly review for the first month.
4. Measuring activity, not outcomes
A surprising number of programmes report green status every week while the business carries on exactly as before. That happens when the project measures its own effort rather than its effect on the business.
- Symptoms: reports count workshops held, user stories completed, people trained or logins, there is no baseline, and nobody can say which business number has moved.
- Fix: before work starts, agree the baseline and a target for one or two outcome metrics, such as time to first reply on a new enquiry, days from delivery to invoice, cost per order processed or enquiry-to-booking conversion. Add one adoption measure, such as transactions completed in the new system rather than the old one. Review them weekly.
5. Ignoring the front line
The people who answer the phones, raise the invoices and run the shop floor know where the exceptions are. When they are left out, the design misses the exceptions, the system cannot handle real work, and staff build workarounds. What gets labelled as resistance to change is often a perfectly rational response to a system that makes their day harder.
- Symptoms: training is the first time staff see the new process, the old spreadsheet is still open on every screen weeks after launch, and complaints are dismissed as people disliking change.
- Fix: include front-line staff in the diagnosis and in design walk-throughs, use them as testers on real data, name a go-to person in each team, and fix the problems they raise quickly and visibly.
6. Data not ready
Duplicate customers, the same product under three codes, missing fields and stock figures nobody trusts. Load that into a new system and it shows wrong numbers on the first day. People conclude the system is broken and return to the spreadsheet they trust.
- Symptoms: nobody can give a reliable count of active customers, the same item appears under several codes, and monthly reports are reconciled by hand before anyone believes them.
- Fix: assess the data during diagnosis, not during build. Scope the clean-up as its own workstream with an owner and a deadline. Migrate the data you need rather than everything you have, and set simple rules, such as mandatory fields and one place to create a new customer, so it stays clean.
7. No run phase
Go-live is the start of the change, not the end. Without someone measuring, tuning and training in the months that follow, the new process decays. New joiners learn the old habits from colleagues, exceptions get handled outside the system, small bugs never get fixed, and within a year the business is back where it started with an expensive system on top.
- Symptoms: the project team disbands at launch, there is no budget after go-live, and nobody looks at the metrics after the first month.
- Fix: plan and fund the run phase from the start: a weekly review of the metrics, a backlog of tuning changes, training for new joiners, and a clear decision point for when the function is stable enough to move on. This is why Crank Labs prices work as a one-time build fee plus a monthly managed plan, so the run phase is part of the engagement rather than an afterthought.
Why does digital transformation fail when it starts with software?
Most stories about why digital transformation fails are a special case of the seven modes above. A business buys a platform before it has designed the process the platform should run. The software then encodes the old way of working, usually including its delays, and nobody owns the outcome because the project was framed as a system implementation. The fix is to reverse the order: diagnose, redesign, then choose and build the technology. We explain the difference in more depth in digital transformation vs business transformation.
How do you make change stick in a mid-sized business?
In a company of 20 to 500 people, change management is not a communications plan or a slogan. It is a handful of practical habits that make the new way of working the easiest one to follow.
- Explain the change in terms of each person's day. "No more re-typing orders into two systems" lands better than any strategy statement.
- Train on real work in small groups. Use the team's own customers, orders and invoices, not a generic demo.
- Switch off the old way on a fixed date. As long as the old spreadsheet is available, some people will use it.
- Make the new way faster than the old one. If the new process takes more clicks, people will route around it, whatever the memo says.
- Have leaders use it. If the CEO still asks for the old weekly spreadsheet, the old spreadsheet survives.
- Update targets and job descriptions. People follow what they are measured on, so measure the new process.
- Celebrate the first measured result. Showing the team that the change worked builds belief for the next one.
Pre-mortem checklist: run this before you start
A pre-mortem asks the leadership team to imagine that it is a year from now and the transformation has failed, then to list the reasons why. It surfaces risks that nobody wants to raise in a kick-off meeting. Follow it with this checklist. Every unticked item is a risk you are choosing to accept.
- We can state the outcome metric this change should move, and its current value.
- One named person owns the process and that metric, and runs the function day to day.
- The first phase delivers a measurable result within weeks, not quarters.
- The people who design the change will also be involved in building and rolling it out.
- Front-line staff have walked through the design and found its holes.
- We know the state of our data and someone owns the clean-up, with a deadline.
- We know which existing tools we are keeping and how the new pieces connect to them.
- We have written down what we are not changing in this phase.
- The old way of working has a switch-off date.
- There is a funded run phase after go-live, with a weekly review.
- For finance and healthcare workflows, a person approves anything before it is sent or posted.
- The CEO will attend the weekly review for at least the first month.
How do you rescue a transformation that is already stalling?
If you are reading this halfway through a programme that is not delivering, the steps are the same, applied in reverse order of urgency.
- Stop adding scope. Freeze new requirements until something is working.
- Re-baseline. Agree which business number the programme was meant to move, and where it is today.
- Name the owner. If there is no single accountable person for the function, appoint one this week.
- Spend a day with the front line. List every workaround, duplicate spreadsheet and manual step people still rely on.
- Pick the part closest to working and finish it. Fix the main issues, retrain, switch off the old way and measure.
- Then decide the rest. With one function working and measured, you can make an honest call on what to continue, change or stop.
What should you do next?
Run the pre-mortem checklist against your current or planned project and be honest about the gaps. If more than a couple are open, fix them before you commit more budget. Our business transformation framework lays out the seven steps that avoid most of these failure modes by design.
If you want help, Crank Labs runs business transformation as one team from diagnosis through build to the run phase, and our consulting and strategy work starts with a free working session that maps where time and money leak. Book a 15-minute call to set one up.
Frequently asked questions
Is resistance to change the main reason transformations fail?
Resistance is usually a symptom rather than the cause. Staff push back when a new process makes their day harder, misses the exceptions they deal with, or arrives without explanation. Involving the front line in the design, training on real work and fixing their complaints quickly removes most of it. If resistance persists after that, it often points to a design problem worth investigating.
How long should the run phase last after go-live?
Long enough for the outcome metric to improve and hold for several consecutive weeks, and for new joiners to be trained into the new process rather than the old one. For a focused workflow that is often a few months. The run phase should end with a clear handover to the process owner, not simply when the budget runs out.
Should we run a transformation in-house or bring in outside help?
In-house works when you have a leader with time to own it and people who can both redesign processes and build systems. Most mid-sized businesses have one but not the other. Outside help is useful when it covers the gap without creating a new hand-off, so look for a partner that will stay through build and rollout rather than leave after the plan.
What should we do if the software vendor blames our process?
They may be right. Software implementations often stall because the process was never designed, only configured. Ask the vendor to name the specific process decisions that are blocking them, then assign each one to your process owner with a deadline. If nobody on your side can make those decisions, that is the real gap to fix before spending more on the software.
How do we know early whether a transformation is working?
Watch two numbers from the first week after go-live: the outcome metric you set a baseline for, and an adoption measure such as the share of work going through the new system instead of the old one. A short dip followed by steady improvement is normal. Flat numbers after a month mean something in the design or the rollout needs fixing.
Book a 15-min call