When information is available but research takes time
A sales team wants to know which businesses in a particular region might be a good fit for its offer. The information may be scattered across several sources. Before the team has a useful list, someone has to search, compare entries, and fill in missing details. If that work keeps coming around, it is worth asking which parts could be prepared or automated.
One possible approach is automated data transformation: gathering existing information, assessing it for a specific task, and putting it into a usable form. An initial trial would show whether that approach helps in day-to-day work.
An example from sales research
Suppose a business is looking for potential customers by region, company size, and field of work. A system could search suitable sources, bring the details together, and suggest possible matches for review. AI may help interpret descriptions that use different terms. The sales team still decides whether a business is relevant, drawing on its experience and any further information it needs.
Changes over time could matter too. Has something happened that makes a company a better fit than it was before? Monitoring that question would require defined sources, update intervals, and criteria for reassessment. Without those boundaries, a wide search mainly creates more work to review.
Start with the decision the data should support
Before building a pipeline, we would work out which decision the data should help people make. What information does that decision require? Where may it come from? How current must it be? Who checks a record before it is used in day-to-day work?
External sources need to be assessed for access, permitted use, origin, and quality. Several results may refer to the same company; a description may be outdated or misleading. If reviews feed into the process, the team also needs rules for conflicting or potentially unfair comments. An automated assessment should not pass those cases through unchecked.
These are much the same questions we ask before other software projects. In Funded Digitalisation: Create Clarity Before Building Software, we explain how to examine workflows and requirements before making an investment.
B2B
Which data does your team repeatedly research or maintain by hand?
We look at your sources, quality criteria, and current workflow, then assess which steps are worth automating.
Step 1 of 2
Step 2 of 2
Non-binding contact enquiry
By submitting, you are sending a non-binding contact enquiry to develate. We use your details to respond to your enquiry personally.
Review suggestions before automating more
A sensible first step would use a limited set of sources and records. The system proposes entries; someone who knows the subject checks and corrects them. That shows where criteria are missing and where the preparation goes wrong. People who do the research or use the records today should help shape the process from the start.
The trial also needs agreed measures. How many entries are actually useful? How often does someone have to correct them? How long does preparation take, and what are the running costs? Those observations can guide the next decision about automation. The right level depends on the use case and the cost of a wrong record.
The aim is to give people better information for their decisions and reduce repeated research. A small, reviewable trial can show whether that works in a particular business.
