Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
  1. Home
  2. Blog
  3. Why Machine Learning Pilots Never Reach Production
Software Strategy

Why Machine Learning Pilots Never Reach Production

The pilot worked and nothing happened. The five reasons projects stop at this point, and what to do differently before the next one.

Updated 2 min readBy SpiderHunts Technologies

Free estimateNo obligation

Get a free estimate

Tell us what you need. A senior engineer reads every enquiry.

Takes under a minute. We never share your details.

  • Free consultation
  • No commitment
  • NDA on request

Prefer to talk? Book a free 30-minute call →

Quick answer — TL;DR

Pilots stall because integration was never scoped, nobody owns the system, the data pipeline was manual, no one budgeted for running it, or the pilot proved accuracy rather than value. All five are avoidable by deciding the production path before the pilot starts.

A common and expensive pattern

A pilot is run, produces good results, is presented, and is received well. Then nothing happens. Six months later it is referenced as evidence that this does not work here.

The pilot was usually fine. What was missing was a route from pilot to production that anyone had thought about.

One: integration was never scoped

The pilot ran on an extract in a notebook and produced a spreadsheet. Production needs it inside the system where decisions are made, which is a different and frequently larger piece of work.

That work is often owned by a different team with its own priorities. Discovering at the end that the necessary change is in next year's roadmap stops the project regardless of how good the model is.

Two: nobody owns it

A pilot is owned by whoever ran it. A production system needs someone accountable for it working - monitoring it, deciding when to retrain, fielding questions when a prediction looks wrong.

If that person does not exist, the system has no home. Projects frequently reach this point and discover that the data team considers it the business's, and the business considers it the data team's.

Three, four and five

  • The pipeline was manual. Someone exported data by hand for the pilot. Production needs that automated, and it is often more work than the model.
  • Nobody budgeted for running it. The build was funded; hosting, monitoring, retraining and support were not, and no budget line exists to create.
  • The pilot proved accuracy, not value. It showed the model predicts well but never established what acting on it would be worth, so there is no business case for the next stage.

That last one is the most common and the most avoidable. A pilot should test whether the improvement is worth having, not only whether prediction is possible.

Decide the production path before the pilot

  1. Write down where predictions will appear and who must agree to that change, before the pilot starts.
  2. Name the person who will own the system in production, and involve them from the beginning.
  3. Establish the baseline first, so value can be measured rather than asserted.
  4. Estimate running costs early and get them into a budget line.
  5. Define what the pilot must show for production to proceed - and what would mean it should not.

None of this makes the pilot much more expensive. It changes it from a demonstration into a decision, which is what it should have been.

A pilot that only proves the model works has not answered the question anyone needed answering.

FAQ

Frequently asked questions

The questions readers ask us after this guide.

Still have a question?

Ask us directly — a senior engineer will get back to you.

Ask about your project

Should we skip the pilot and build properly?

Rarely - the pilot reduces real risk. The fix is scoping the production path alongside it, not abandoning it.

How long should a pilot take?

Weeks rather than months. A long pilot loses sponsorship and tends to drift into building something half-production.

Who should own the production system?

Someone in the business function using it, with technical support. Ownership left with a technical team tends to drift.

What if integration turns out to be impossible?

Better to learn it before the pilot than after. That is why it should be scoped first.

Keep reading

More on Software Strategy

Software Strategy

Machine Learning Myths That Waste Budgets

Eight beliefs about machine learning that quietly inflate project costs, what is actually true instead, and how to spot each one in a proposal.

Start here

Want machine learning project details from us?

Tell us what you are trying to predict and roughly what data you hold. We will come back with an honest view on whether machine learning is the right tool, what the work would involve and a realistic cost range. If a spreadsheet would do the job, we will say so.

  1. You tell us what you needTwo minutes on the form, or a message on WhatsApp.
  2. A senior engineer reviews itAnd comes back with questions, a realistic range and an honest view on fit.
  3. Free 30-minute scoping callWe talk through scope, options and a realistic estimate — with no obligation.
Free estimateNo obligation

Talk to someone who builds this

Send a short brief and we will come back with an honest view and a realistic range.

Takes under a minute. We never share your details.

  • Free consultation
  • No commitment
  • NDA on request

Prefer to talk? Book a free 30-minute call →