Business Intelligence Consulting

Business Intelligence Consulting Services for Companies That Cannot Yet Agree What the Numbers Mean

Ask three people in different teams to define your primary revenue metric in one sentence, separately. If you get three sentences, no dashboard is going to help you.

That is the engagement this page sells. Business intelligence consulting services usually arrive as a build — a warehouse, a reporting layer, a dashboard refresh. This is the advisory piece of work that has to come first when nobody can yet agree what the numbers mean: not a warehouse, not a reporting build, but a piece of work whose deliverable is your metrics defined in writing, each with a source, a freshness expectation and a named owner, plus the governance model that keeps them true after the people who agreed them move on.

If you already know what to measure, you are on the wrong page. When the definitions are settled and the problem is that the pipelines do not exist or cannot be trusted, that is a build, and it belongs at Analytics & BI. Buying advisory when you need construction is an expensive way to be told something you already knew.

What our own definitions pass produced was a decision list, not a data model

Facet's delivery — engineering, content, research, quality review and reporting — runs through Agent Loops under human command, which meant our own numbers had to be written down precisely before anything could act on them. We expected that to be a modelling exercise. It was not. The output was a list of decisions nobody at the company had ever actually made.

Not ambiguities to be resolved by looking harder. Genuine forks — two reasonable definitions of the same thing, each defensible, each with a team quietly operating on it for years, and no mechanism anywhere in the company for choosing between them. Everybody knew about them. Nobody had ever been in a room where choosing was the agenda.

That is the experience this service is shaped around, and it explains three things about how the engagement runs:

The discovery half is agent-carried, because it is reading rather than judgment. Inventorying sources, profiling fields, finding where two systems disagree and drafting a first-cut definition for each metric is high-volume mechanical work that a senior consultant used to do by hand, by the hour. Putting agents on that pass is why this is a fixed-fee, time-boxed engagement instead of an open-ended reading exercise you fund by the month.
The deciding half is the actual product, and it is entirely human. An agent can show you that two systems disagree about what a customer is. It cannot tell you which one your business meant, and it should not try. What we sell is a structured session with the forks in front of you and somebody responsible for closing each one.
The failure mode is a confident wrong answer. A model asked to interpret an ambiguous field will produce something plausible rather than stop. That is exactly why every drafted definition goes back to a room before anything is built on it, and why agent-drafted work clears the same gate as human-drafted work.

The longer account of how our operation is structured is in We Run Our Own Business on Agents.

Why hire a business intelligence consulting partner?

Because the thing blocking your reporting is usually not technical, and a technical vendor will build around it rather than surface it.

Two teams report the same number differently, and both are defensible. The meeting becomes about the definition instead of the decision. This is the single most common thing we find, and it has nothing to do with tooling.
Nobody owns a metric. A number with no owner has nobody to ask when it looks wrong, and nobody accountable when it is. Software does not volunteer for this.
The reporting is retrospective by design and expected to be predictive. A monthly dashboard is a good instrument for explaining what happened and a poor one for changing it. Companies buy the first and are disappointed it is not the second.
The definition dispute is two years old. Which means it is not an analysis problem. It is a governance problem, and it will survive any tool you buy.

Measure your own version of this. Over your last quarter, count the leadership meetings where time went on what a number meant rather than what to do about it. (Illustrative arithmetic on your own inputs, not a benchmark: four such meetings a quarter, six people in the room, twenty minutes each, is roughly thirty-two hours a year of senior time spent re-litigating something a written definition settles once.) The hours are the visible cost. The decisions deferred while the argument ran are the larger one, and they do not appear on any invoice.

How could AI agents run your reporting once the definitions exist? Book the metric definition workshop — one working session plus one week, fixed fee. You leave with your top operating decisions written down, the metrics behind them defined in one sentence each with source and freshness stated, a named owner per metric, and an honest map of which ones your current data can support and which it cannot.

What the loop actually does

The advisory engagement has a loop inside it, and it runs during the discovery week rather than forever. These passes are agent work because each is high-volume, mechanically checkable, and the kind of reading a person does well for two days and then starts sampling.

Inventory every source that produces a number anybody uses. Systems, tables, exports, the spreadsheet on somebody's desktop that feeds the board pack. The last one is almost always present and almost never in the brief.
Profile the fields behind each candidate metric. Population rates, value distributions, when a field started being used differently, and where the nulls cluster. That is how you find the month somebody changed a process without telling anyone.
Find where two systems disagree, specifically. Not "the CRM and the billing system do not match" but which records, by how much, and what the two systems each believe a customer is.
Draft a first-cut contract for every metric. What it counts, what it excludes, where it comes from, how fresh it should be — written as a proposal for a person to accept, amend or reject, so the session starts from a document rather than a blank page.
Trace each metric back to a decision. A metric nobody makes a decision from is a metric to stop maintaining, and surfacing those is often the most immediately useful finding in the week.

What a person still decides

Everything that matters. The loop shortens the reading so the deciding can be the engagement.

Which definition is the one your business meant. Two defensible answers, one company. This is the fork, and it is the reason you are buying an outside engagement rather than an internal project.
Who owns each metric by name. Not a team, a person. Ownership is the only mechanism that keeps a definition true after the workshop ends, and assigning it is a management decision.
What is worth governing and what is not. Every metric under formal governance is a standing cost. Deciding which numbers earn that and which stay informal is a judgment about where being wrong actually hurts.
What to do when the map says the data is not there. Sometimes the honest finding is that you need instrumentation before you need analysis. Acting on that means accepting a smaller, slower answer than the one you came in wanting.

Business intelligence consulting for professional services firms

Professional services organizations — law firms, consulting practices, recruiting agencies, managed service providers — have a specific and unusually solvable version of this problem, and it is why a disproportionate share of this work lands here.

The revenue data lives in a PSA or billing system that does not talk to the CRM. Utilization lives in spreadsheets. Leadership is making headcount and pipeline decisions from a thirty-day-old snapshot, and the three numbers that actually run the business — utilization, realization and pipeline coverage — are each calculated slightly differently by whoever was asked.

Utilization is a definition argument before it is a reporting problem. Billable against what denominator: available hours, contracted hours, or hours worked? Each answer produces a different number, each is used somewhere in your firm right now, and the gap between them is frequently large enough to change a hiring decision.
Realization exposes what the practice actually collects. Booked against billed against collected, with write-offs attributed to the engagement rather than absorbed centrally. Firms are often surprised by which practice areas look different under that lens.
Pipeline coverage needs one shared idea of a qualified opportunity. When business development and the partners who deliver the work define that differently, forecasting is arithmetic on two incompatible inputs.
Partner compensation makes every definition load-bearing. In a firm where the numbers feed a comp model, a metric definition is not an administrative detail. It is the thing people will litigate, which is exactly why it is worth settling deliberately and in writing rather than in the moment.

We bridge that gap: ingesting project management, billing and CRM data into a shared view, and settling the definitions of the metrics professional services operators actually manage the business on. Facet has been working with law firms, consultancies and professional services organizations since 2014, and this is the sector where the definition work pays back fastest, because the decisions it feeds are weekly rather than quarterly.

Where business intelligence consulting goes next

A ladder, not a bundle. Plenty of firms take the definitions, the owners and the governance model and never build anything, and that is a complete engagement rather than a failed one.

Analytics & BI — the build rung. Once the definitions exist, this is where the pipelines, the warehouse and the reporting get built and then operated.
Information Architecture Analysis — when the naming and structure underneath the data need settling before the metrics can be.
Requirements Engineering & Discovery and Discovery Workshop — when the ambiguity is about the process itself rather than the numbers describing it.
Agentic Engineering and the Agentic Operating System — the rungs above, for when a defined metric should open work rather than get read. A defined metric can have a loop behind it; an undefined one cannot, which is the practical reason the definitions are worth this much attention. The Agentic Operating System explainer covers the shape while those service pages are in production.

Why Facet

Facet Interactive has been a boutique systems integrator since 2014, and CIOReview named us a Top 10 Digital Transformation Services Company in 2021. Our team of 40 spans 9 timezones and 6 continents, backed by a network of 500+ open-source, data and IT specialists.

We are not a pure analytics shop, and for this engagement that matters. Definitions do not live in a data warehouse. They live in your CRM configuration, your billing system, your web analytics and the forms your operations team fills in — and settling what a metric means usually means understanding all of those. Facet's data practice sits inside a firm that also does CRM, web and IT work, which is why the definition conversation does not stop at the reporting layer.

You will not find an industry percentage anywhere on this page. Your definitions, your systems and your governance are specific to you, and the workshop produces your map instead of borrowing somebody else's.

Start with a metric definition workshop

How could AI agents run your reporting once the definitions exist? Book the metric definition workshop — one working session plus one week, fixed fee. We inventory your sources, profile the fields behind your candidate metrics, draft a contract for each one, and run the session where the open forks get closed. You leave with the definitions, a named owner for each, the governance model, and an honest map of which metrics your current data can support. All of it is yours whether or not we build anything afterwards.

Some of these workshops end with the opposite of a proposal. If the map shows your data cannot yet answer the decisions you are trying to make — that what you need is instrumentation rather than analysis — that is the finding, and it points you at a smaller piece of work than the one you arrived asking for. We will also tell you plainly when nobody internally is empowered to settle a definition dispute, because an engagement without a decider stalls in week three and we would rather not start it.

How pricing works. The workshop is a fixed fee, quoted before work starts, with no hourly billing. Facet's project range starts at $6,000 and this engagement sits at the lower end of it by design — it is deliberately smaller than the build it usually precedes.

FAQ

How is this different from hiring a BI analyst?

An analyst produces analysis inside your existing definitions. This engagement exists to settle the definitions, which is not an analyst's job and frequently not within an analyst's authority. A new hire also cannot convene the room — an internal person raising a two-year-old definition dispute is starting an argument, whereas an outside engagement with a date on it is a scheduled decision. That difference is most of what you are buying.

Do we need a data warehouse before this?

No, and buying one first is a common and expensive way to defer the real problem. Centralizing inconsistent data produces centralized inconsistent data on a larger bill. The definitions agreed here stay valid when the architecture changes later, so the work is never wasted — which is the opposite of the usual sequence, where a platform is bought and the definitions are argued about afterwards, on a clock.

What if two departments simply refuse to agree?

Then you have a governance problem rather than a data problem, and we will say so directly. We can facilitate the decision, put both positions in writing with their consequences attached, and make the cost of not deciding visible. We cannot make the decision, and an engagement where nobody internally can is one we will tell you not to start.

How long does it take?

One working session plus one week for the discovery and drafting around it. The engagement is deliberately time-boxed, because the value is in forcing the decisions rather than in the length of the study, and an advisory piece that runs for months has usually turned into an excuse not to decide.

The meeting that settles this has never once been scheduled voluntarily

Every company carrying a definition dispute already knows about it. It is not hidden. Somebody raises it about twice a year, everyone agrees it should be sorted out, and then it is not, because the meeting has no natural owner, no deadline, and no consequence for being skipped. Meanwhile both definitions stay in use and both keep being defensible.

An advisory engagement is, mechanically, a date in the calendar with a document attached and somebody outside the argument holding the pen. That is unglamorous and it is the entire mechanism. Not expertise you lack — you know your business better than we ever will — but a forcing function your organization cannot generate for itself, because the people who would have to convene it are the same people on opposite sides of it.

That is why this is a small engagement rather than a large one, and why we would rather sell it before a build than during one. A fork settled in a workshop is a decision. The same fork settled halfway through a warehouse implementation is a change order.