Think Build Implement Repeat
London, UK +44 7367 067226
WhatsApp FOLLOW f in X
  1. Home
  2. Blog
  3. We Have Years of Business Data. How Can I Use Machine Learning on It?
Problems We Solve

We Have Years of Business Data. How Can I Use Machine Learning on It?

Lots of data does not tell you what to predict. We start from the decisions your business makes and work back to the machine learning your data can support.

Updated 3 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

The useful question is not what your data can do but which repeated decision would improve if you knew something in advance. Start by listing the decisions made every week on guesswork, check which of them your data could inform, and pick one where a prediction would change an action. That gives machine learning a job, not just a dataset.

The data is there, the value is not

Your business has run on the same systems for years. There is sales history in the ERP, customers in the CRM, jobs in the job management system, and a finance ledger going back a decade. Everyone agrees the data must be valuable. Someone has read that machine learning can find patterns in it. But when you ask what exactly it would predict, the room goes quiet.

Maybe you have already tried. An analyst built a dashboard that nobody opens. A vendor ran a demo of "AI insights" that produced charts of things you already knew. The data sits there, costing money to store, and the question of how to actually use machine learning keeps coming back without an answer.

Why starting from the data goes nowhere

Most attempts start with the data: gather it, clean it, explore it, and hope something interesting turns up. Patterns do turn up, but they rarely change what anyone does on Monday morning. A finding such as "customers in the North buy more in winter" is interesting and does not tell anyone what to do differently.

Machine learning pays off when it informs a specific, repeated decision: how much to order, which customer to call, which job to price higher, which machine to service, how many staff to roster. The decision gives the prediction a purpose and a measure. Without one, even an accurate model has nowhere to go.

What unused data costs

PatternWhat it leads to
Data kept but not usedStorage and upkeep costs with no return
Dashboards nobody opensEffort spent describing the past rather than informing decisions
Decisions still made on gut feelThe same errors repeat despite the history being available
Vague AI projectsBudget spent on exploration with no clear outcome
Scepticism"We tried data science and it did not work" becomes the story

The last one matters most. A failed first project tends to put people off the good projects that come after it.

How we find the machine learning worth building

  1. We interview the people who make recurring decisions, in buying, sales, operations, finance and service, and list the decisions they make on judgement every week or month.
  2. For each one we ask: what would you do differently if you knew X in advance? If the answer is nothing, it drops off the list.
  3. We check the data behind the remaining decisions: whether the history exists, whether the outcome was recorded, and whether it can be reached from your systems through an API or export.
  4. We estimate the value of a better decision in plain terms, such as less excess stock, fewer lost customers or fewer wasted hours, without inventing figures, and weigh it against how hard the data is to use.
  5. We recommend one or two first projects with a clear owner, a clear action, and a way to measure whether the prediction helped, and we say which ideas are not ready yet.
  6. For the first project, we build a simple version quickly and test it against past decisions before anyone relies on it.

Sometimes the answer is that a good report or a simple rule would fix the problem without machine learning. We will say so. Those are often the best first steps anyway.

What you get

A short, specific list of decisions that machine learning could improve in your business, with the data behind each and an honest view of which are ready. Then a first model aimed at one of them, with a named person who will use the output and a way to tell whether it worked.

The data stops being a vague asset and becomes the raw material for particular decisions.

It also sets a pattern for the next project. Once the business has seen one prediction change one decision, and measured it, the second idea is judged the same way: what decision, what data, what action, what measure.

Is this your situation?

  • You have years of data across several systems and little use made of it.
  • Key decisions are still made on experience and instinct.
  • Previous analytics or AI efforts produced reports rather than changes.
  • Nobody can name a specific thing they would want predicted.
  • You want to know how machine learning could help before committing budget.

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

What kind of decisions suit machine learning?

Decisions that are repeated often, depend on patterns in past data, and where the outcome is recorded. Stock ordering, lead follow-up, pricing and maintenance timing are common examples.

Do we need to clean all our data first?

No. Only the data behind the chosen decision needs to be in good shape, and preparing it is part of the first project.

How do we know if it worked?

By agreeing the measure before building, such as stock held or customers retained, and comparing against how decisions were made before.

What drives the cost of this discovery work?

The number of departments and systems involved. It is deliberately a small step compared with building anything.

What do you need from us?

Time with the people who make recurring decisions and an overview of your main systems.

Keep reading

More on Problems We Solve

Start here

Tell us what your data should be telling you

Describe the decision you want to improve and the data you already keep. We will give you a straight answer on whether machine learning fits, and if a report or a simple rule 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 →