What Business Intelligence Consulting Actually Involves

Business intelligence consulting sounds like it should be about dashboards. In practice, if the engagement starts with dashboards it has already gone wrong.

The dashboard is the last five percent. Everything that determines whether it gets used happens before it exists.

What the work actually consists of

Finding out where the numbers come from

Most organizations have several systems producing overlapping data, and one or two people who quietly reconcile them. Mapping this properly is the first job, and it routinely surfaces the fact that two departments have been reporting the same metric differently for years.

Agreeing definitions

This sounds administrative and it is the hardest part. What counts as an active client? Does revenue book on signature or delivery? Until these are settled, every report is contestable, and a contestable report is one people ignore in favour of their own spreadsheet.

This work is not technical. It is a series of conversations between people who did not know they disagreed.

Building the layer underneath

Pipelines that pull from source systems on a schedule, apply the agreed definitions, and produce a consistent set of numbers. Unglamorous, invisible to the end user, and the thing that determines whether the reporting is trusted.

Then the reporting

Built for the person making the decision, not the person who built the database. The most common failure is a dashboard with forty metrics on it, which is what happens when nobody decided what the thing is for.

Signs you need this

• Someone spends multiple days a month assembling a report by hand
• Two teams produce different numbers for the same thing and both are defensible
• Leadership decisions are made on data that is weeks old
• Your reporting exists, and people maintain private spreadsheets anyway

That last one is the clearest signal. Shadow spreadsheets are what people build when they do not trust the official numbers.

Why it matters more now

Every organization is being asked about AI. Almost none of them are ready, and the reason is nearly always the same: the data underneath is inconsistent, undefined and untrusted. Applying AI to that produces confident answers built on disputed inputs.

The BI work is not a detour on the way to AI. It is the prerequisite, and organizations that do it first tend to find their AI ambitions become smaller, cheaper and considerably more likely to ship.