Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
  1. Home
  2. Blog
  3. What You Should Get From a Machine Learning Feasibility Study
Software Strategy

What You Should Get From a Machine Learning Feasibility Study

A short first stage should answer whether this is possible, valuable and worth doing - with enough evidence that a no is as useful as a yes.

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

A feasibility stage should deliver a measured baseline, an assessment of whether the data supports the prediction, an early performance indication, and a clear recommendation - including a recommendation not to proceed where that is the honest answer.

The purpose is a decision, not a prototype

A feasibility study exists so you can decide whether to commit to a build. It is not a small version of the project, and judging it on how impressive the demo looks is the wrong test.

The measure of a good one is that afterwards you know something you did not know before, precisely enough to act on.

What it should contain

  1. A data assessment. What exists, its quality, what is missing, and whether outcomes are recorded reliably enough to learn from.
  2. A measured baseline. How well the current method performs, measured properly. Without this nothing later can be judged.
  3. An early performance indication. A rough model on real data, with an honest statement of what it achieved and under what conditions.
  4. The value estimate. What an improvement of that size would be worth, using your numbers.
  5. Risks and unknowns. What could still go wrong, and which questions remain open.
  6. A recommendation. Proceed, proceed differently, or stop - with reasoning.

The baseline is the part most often skipped

Measuring how well the existing process performs is unglamorous and frequently omitted, and it is what makes every later number meaningful.

It also sometimes ends the project usefully. A business convinced its forecasting is poor occasionally discovers the planner's spreadsheet is performing well, and that the money is better spent elsewhere. That is a valuable outcome from a small spend.

A good study can recommend stopping

FindingHonest recommendation
Outcomes not reliably recordedFix recording first, revisit in a year
Model barely beats the baselineDo not proceed as specified
Improvement real but worth littleNot a priority; consider a different problem
Data exists but cannot be joinedSolve the data problem first
Clear improvement, clear valueProceed, with defined scope

A supplier who never recommends stopping is not assessing feasibility. The willingness to deliver bad news is most of what the stage is for, and it is reasonable to ask a prospective supplier when they last told a client not to proceed.

Scope and duration

A feasibility stage should be short and fixed price - weeks, not months. Its purpose is to reduce uncertainty enough to decide, not to build anything durable.

Expect the deliverable to be a written assessment with evidence, not a working system. Anything produced during it is exploratory and should be treated as such - reusing feasibility code as production code is a recognisable way to accumulate problems.

A feasibility study that could only ever conclude 'yes' was not a study.

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

How long should feasibility take?

Usually a few weeks, depending on how accessible the data is. Data access delays are the most common reason it stretches.

Should it be fixed price?

Yes. The scope is defined and the point is a decision at a known cost.

What if we cannot give access to real data?

Then feasibility cannot be properly assessed. Anonymised or sampled data under an agreement is usually the way through.

Do we own what is produced?

Agree it in the contract, as with any project. It is exploratory work, but the findings and any code should be yours.

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 →