Three months to clarify priorities, deliver a first use case and hand over the method

IN BRIEF  A 90-day data project should help you make one decision more reliably, establish a foundation you can reuse and give your teams the skills to carry on independently.

When an SME asks what it can realistically change in three months, we start by setting expectations. Ninety days is enough to bring a business problem into focus, get reliable data for a defined use case, deliver a dashboard that helps people make decisions and plan what comes next. That depends on keeping the project focused.

You may recognise this scene. Management wants to “make better use of data”, sales asks for new reports, production needs a reliable view of delays and everyone already has a favourite tool. The calendar fills up before anyone asks the central question. Which decision should be faster or more reliable after 90 days? Without an answer, the programme can keep everyone busy while making little observable difference.

In the projects we support, we treat the three months as a sequence of learning. In the first month, we understand the situation and make choices. In the second, we build the model and test it against real data. The third establishes how the dashboard will be used, who checks it and how your teams take it over. The approach follows Datavizin’s ECSTAA model: explore, understand, structure, analyse and act, then repeat using what you learned on the ground.

Over 90 days, we set out to understand the problem, build a solution, help people adopt it and prepare the next phase.

What can you realistically achieve in 90 days?

The word “transformation” invites big promises. We prefer to be specific. In three months, you can establish what a handful of indicators mean, connect a limited number of data sources, build a first model and test a regular decision-making routine. Replacing the ERP, cleaning up ten years of historical data and rolling out a comprehensive governance programme may take longer. What matters is how dependable that first delivery is.

We therefore make the scope explicit. It defines a process, which records are in scope, a timeframe, a few indicators and one decision. For example, track overdue orders every week so the production manager can prioritise orders at risk. This makes you choose. It also protects the budget. Any extra request can be assessed against the intended outcome before it gets added.

Before day one, bring together the people who know how the work gets done

A 90-day programme is off to a poor start when the only people at the first meeting are a project manager and a provider. We want to hear from a sponsor who can make decisions, someone who knows the data sources and a future user. Their answers may differ. Finance may talk about invoices, sales about orders and management about revenue. Those differences are the work you need to do before building.

We also ask for access to concrete cases: a closed month, a specific order, a product known for its exceptions and the previous reports. A demo on perfect data can reassure people for an hour. A test on a case your teams recognise builds trust. If access, the right people or the priority question is still unavailable, moving the start date can save the entire first phase from being spent waiting.

Days 1–30. Make the problem and success measurable

In the first month, we observe the actual process. Who enters the information? Where is it corrected? Which file is authoritative when totals differ? We map the sources as they are. We want to understand how a figure is produced, what it depends on and who can explain an issue. A short map makes sure that a manual step known by only one person is accounted for in the model.

We then turn the objective into observable criteria. Should the dashboard shorten reporting preparation, reveal delays two days earlier or bring two departments to the same definition? Choose one outcome measure and a few practical checks: a documented source, validated figures, a visible refresh date and named owners. These checks will guide acceptance testing at day 90.

Days 31–60. Build a simple foundation and test it against reality

The project takes shape in the second month. We prepare the data, define the calculation rules and build the model before working on the visuals. Does a row represent an order, an order line, a delivery or an invoice? How do you handle cancellations, currencies and missing dates? The answers determine whether the dashboard is useful, however impressive it looks.

We deliver in short iterations. A first page can be deliberately simple: three to five KPIs, a trend, a filter and a detailed view. Users test it on familiar cases and challenge the definitions. This is the best time to surface disagreements before they cause surprises in production. If the need is well defined from the start, our Data Starter Pack can deliver a first industrial dashboard in ten working days. The rest of the 90 days can then be used to establish routine use, document the work and deliver priority additions.

Days 61–90. Make the dashboard part of daily work

Adoption happens when your teams know when to open a report, how to check a figure and what to do with what they find. In the third month, we test the dashboard in a real meeting. The person in charge should be able to spot a deviation, open the relevant detail and assign a next step independently. We use any difficulties we observe to improve navigation or clarify definitions.

We also make arrangements for ongoing support. Who monitors refreshes? Who corrects source data? Where are the KPI rules kept? Who decides whether a new request comes into scope? Documentation should let someone else understand the flow, identify a problem and know whom to contact. That independence is one of the project's deliverables.

The three-phase roadmap

PeriodGuiding questionExpected evidence
Days 1–30Which decision do we need to make better?Documented use case, sources, rules and success criteria
Days 31–60Does the model hold up in real cases?Prepared data, validated measures and a first dashboard
Days 61–90Can the teams use and take over the dashboard?Tested review routine, checks, training and a roadmap

What we measure at the end of 90 days

We look at whether the priority decision has better support, whether people understand the definitions and whether they can estimate the work for the next step. Time saved is useful. Fewer debates about the figures, more visible exceptions and the ability to take over the process matter just as much.

Sometimes the right conclusion is to wait before extending the scope. If the first use case reveals data entry issues upstream or an unstable source, that knowledge saves you from investing too soon. We favour an honest roadmap over a plan with a series of commitments made in advance. A successful data transformation gives you more control over the decisions that follow.

NEXT STEP  Choose one recurring decision, one reference month and three people: a sponsor, someone who knows the sources and a future user. One meeting should be enough to test whether the scope really fits within 90 days.