Knowing the data exists somewhere in the business, but being unable to get to it, is frustrating. Sales holds one part of the picture, finance another, and operations something else again. By the time someone has extracted, reconciled and explained it, the opportunity to act may already have passed.
Even when the information is accessible, connecting it can be another problem. Customer records do not match, products have different descriptions, and teams measure the same activity in different ways. A new dashboard might make some of that information easier to see, but the business still needs a dependable way to turn it into action.
These problems have answers. Some need a straightforward fix; others require changes to processes, integration or more substantial engineering. The opportunity is to make useful progress while building the capability to answer better questions, including questions the business has not yet thought to ask.
Make the available data usable
A business can hold years of customer, transaction and delivery information without being able to use it together. Consider a company trying to understand which customers are genuinely profitable. Revenue alone cannot answer that question if discounts, delivery costs, returns and support effort sit elsewhere.
The first task is to establish what is preventing a useful view. It could be an access issue, an unavailable connection between systems, inconsistent identifiers or information captured too late. Each requires a different response, and buying another reporting tool before understanding the obstruction risks adding cost without removing it.
Some improvements are relatively small: agree a shared customer identifier, provide controlled access to an existing source, or replace a repeated spreadsheet export with a reliable connection. Others need the process itself to change. If delivery teams only record additional effort at the end of a project, a live report cannot show costs that nobody has captured.
Ownership matters here. Someone must be accountable for what a measure means, where its information comes from and how errors are corrected. That is a business responsibility supported by technology, particularly when different departments have perfectly reasonable but incompatible definitions of the same thing.
Work from both ends
Starting with the decisions that need improving gives the work a commercial purpose. Where is the business losing margin? Which customers need attention, and which delivery commitments are becoming difficult to meet? Those questions help establish the information, timing and accountability required.
Alongside them, look for useful improvements that are already within reach. A recurring manual report might be a good place to start, even if it is not the organisation’s most ambitious data problem. Removing avoidable preparation work can release time, expose weaknesses in the underlying information and build confidence in the next change.
Technology development is iterative. Teams learn by putting something useful into operation, testing it with the people doing the work and improving it. The first version should have a defined purpose and acceptable quality, but it does not need to solve every future requirement.
The discipline is to connect early wins to the wider direction. Automating one report is valuable if it reduces effort and establishes reusable data or processes. Creating another isolated solution that only one person understands may save time today while making the next improvement harder.
For example, automating part of the monthly board pack could establish consistent definitions and reliable feeds that also support weekly operational decisions. That makes the first improvement a foundation for further value. Leadership can judge the next investment using evidence from work already delivered.
Use AI throughout the work
AI has a role across this process. It can assist with preparing and transforming data, help teams develop automation, and make information easier to explore through questions and summaries. It need not wait until every system is connected and every reporting problem has been resolved.
Used appropriately, AI can help propose transformations, identify inconsistencies and support the work of bringing differently structured information together. Those suggestions still need testing against the source and the business meaning. Two records that look similar are not necessarily the same customer, and a plausible correction can still be wrong.
At the point of use, natural language questions can make analysis more accessible to people who do not build reports. A sales leader might investigate a change in purchasing behaviour; an operations leader might explore the work associated with repeated delays. They need access to the relevant information and a model that understands the measures being discussed.
Business context is essential. Microsoft’s guidance for Copilot in Power BI makes the same point: preparing the underlying data model helps reduce ambiguity and improve the relevance of responses. The interface can be simple, but the definitions, relationships and checks beneath it still require care.
Choose the method that suits the task. A predictable calculation may need ordinary automation, while interpreting varied documents or exploring patterns may benefit from AI. Both can contribute to the same outcome, without making every step more complicated than it needs to be.
Ask what the business has not noticed
Answering known questions is useful, but it limits analysis to what people already know to investigate. There is a bigger opportunity in asking a well-designed analytical system to surface unusual patterns, overlooked relationships and possible opportunities. That gives leaders new questions to pursue, rather than another presentation of familiar measures.
Revenue growth also depends on better choices about which opportunities to pursue and which customers need attention. AI-assisted analysis can help a commercial team prioritise its effort, alongside automating routine sales work. The value comes from improving those decisions and acting on them.
Consider a hypothetical analysis combining purchasing history, product mix, returns and service interactions. It might flag changing customer needs or an overlooked opportunity within an existing account. The team can investigate the evidence and decide whether to prioritise a conversation, resolve a service issue or redirect its effort.
The same principle applies to delivery, where familiar measures can leave problems overlooked. Combining demand, capacity, rework and scheduling information may reveal a neglected handover or a recurring constraint. AI-assisted analysis can challenge assumptions about where the problem sits and suggest changes for the operations team to investigate and test.
For customers, the benefit could be a more useful way to buy: a better product combination, an appropriate replenishment reminder or a simpler ordering process. The test is whether the recommendation helps them achieve something they value. A relevant suggestion can support repeat business; an irrelevant one is simply more noise.
This is where competitive advantage can develop. The business uses its own information and operating knowledge to notice an opportunity, validate it and respond. Access to an AI tool alone offers little distinction if everyone asks the same familiar questions and nobody changes what happens next.
Leaders need the AI skills and confidence to examine that challenge. What is the analysis trying to achieve, which evidence supports it, and why does it suggest a different priority? Check the source records and alternative explanations; an AI-generated explanation is not proof. Understanding the basis of the challenge is part of making the decision.
AI is never accountable. The people making and authorising decisions remain responsible for the judgement, action and consequences. Experience should inform their challenge, while leaving room for evidence to change their minds.
Make the insight change the work
Useful insight needs someone with the authority and support to act. If analysis identifies a customer opportunity or delivery problem, establish who can change the approach, within what boundaries and when they must escalate. Responsibility for spotting a problem is insufficient if every response waits for permission.
Leaders must back those decisions, including when the evidence challenges an established plan or senior opinion. People need a safe route to raise concerns and confidence that a difficult decision made within agreed boundaries will receive support. Build these decision rights into the workflow alongside access to information.
Measure the outcome that justified the effort: less time preparing board reports, earlier intervention on delivery problems, improved customer response or better margin. A dashboard being opened regularly does not, by itself, establish any of those results.
The first step can be practical and contained. Select one decision that needs improving and one recurring activity that consumes too much effort. Establish what information already exists, what prevents its use and which change would produce a useful result soon enough to learn from it.
Relentica’s AI, Automation and Digital Transformation services connect data and analytics enablement, process optimisation and AI adoption to these business outcomes. The work is about improving how the organisation operates, with technology choices serving that purpose.
Before commissioning another dashboard, ask what the business could do differently with the information it already holds. Then ask what it might discover if it could connect that information and look beyond the questions it already knows to ask.
