Data analytics consulting that ends in decisions, not dashboards
Data analytics consulting for growing companies whose numbers live in too many places. We start with the decisions you need to make, agree one definition for each metric, pull clean data from your accounting, CRM, operations and web analytics, and build reports that update by themselves. The result is a handful of numbers your team trusts and uses every week, with the data and the reports in your company's own accounts.
- No minimum contract: book a few hours or a full project
- Scope agreed in writing before work starts
- Independent advice: no reseller commissions
Questions first, tools later
Most analytics projects begin with a tool and end with dashboards nobody opens. We begin the other way round. Before anything is built, we write down the five to ten questions the business actually asks, the decision each one feeds and how often it is needed. Which customers are profitable after service costs? Which marketing channel brings customers who stay? How many days does it take from order to cash? Which jobs overrun their estimate, and by how much?
Each question then gets a metric, a definition, a data source and an owner. Consulting on data analytics is mostly that work: making the questions precise enough that the answer is a number everyone accepts.
Signs you need it
- The monthly report takes days. Someone exports from three systems, pastes into a spreadsheet and fixes the same errors every month.
- Two people quote two different numbers. Sales and finance disagree on revenue or customer counts because each uses its own system's definition.
- Questions wait for one expert. Only the person who built the spreadsheet can answer, and they are busy.
- History disappears. A tool keeps data for a limited time, or a system was replaced and the old data was never moved.
- Decisions run on feel. Pricing, hiring and marketing spend are decided without numbers, because getting them is too slow.
If two or more of these sound familiar, the first step is not a new tool. It is a short audit of where your numbers come from and a written definition of the few that matter most.
What we do
- Data audit. Where your numbers come from today, which spreadsheets sit in between, and where the figures disagree.
- Metric definitions. One written definition per metric: what counts as a customer, as revenue, as an active user, as a finished job.
- Data pipelines. Automatic extraction from the systems you use, such as accounting, CRM, ERP, e-commerce and web analytics, into one place.
- Data model. Tables designed around your business, so new questions can be answered without starting over.
- Reports and dashboards. A few views per role, built for the decision they support, delivered on a schedule.
- Training. Your team learns to read the reports, to change them and to ask the next question.
- Data quality checks. Automatic alerts when an import fails or a figure jumps in a way that suggests an error.
Typical questions and what they need
| Question | Data needed | Usual output |
|---|---|---|
| Which customers are profitable? | Invoices, hours or costs per customer, service tickets | Margin per customer and segment, monthly |
| Which channels bring customers who stay? | Web analytics, CRM, invoices over time | Revenue after 12 months by first channel |
| How long from order to cash? | Order, delivery, invoice and payment dates | Days per step, with the slowest orders listed |
| Which jobs overrun their estimate? | Quotes, hours and purchases per job | Estimate against actual, by job type and team |
| Where are we losing sales? | CRM stages and dates | Conversion and time per stage |
The sources your numbers come from
For most growing companies the data sits in five or six systems: accounting, invoicing, a CRM, an operations or ERP system, an online store or booking tool, and web analytics. Each has built-in reports, and each defines things its own way. The value of analytics comes from combining them, which means pulling the data out on a schedule, matching customers and orders across systems, and keeping the history.
History is easy to lose. One example from web analytics: Google's help page on data retention in Google Analytics says standard properties keep user-level and event-level data for either 2 or 14 months, and that the setting affects explorations and funnel reports, while standard aggregated reports are not affected. (Read on 10 November 2026.) If you want to compare customer behaviour across longer periods, the raw data has to be exported to your own storage before it ages out. The same kind of limit exists in many business tools, so we check retention for each source in the audit.
Tools, chosen for your size
We work with common, well-supported tools rather than one platform we resell. A typical stack for a company of ten to two hundred people is a cloud warehouse or a managed database, scheduled imports from each system, and a reporting tool such as Looker Studio, Power BI or Metabase. Larger setups may add a transformation layer so that every metric is defined in code and versioned. The choice depends on the systems you already have, the licences you already pay for and the people who will maintain it. Everything is set up in accounts in your company's name, with data hosted in an EU region by default.
Data quality and trust
A number that is wrong once is doubted for months. Trust is built with a few simple habits: every metric has a written definition and an owner, every pipeline is monitored, and every report shows when its data was last updated. When a figure changes because a definition changed, the change is recorded. When an import fails, the owner hears about it before the Monday meeting does.
Personal data and the GDPR
Most business reporting works on aggregates: revenue per segment, orders per week, hours per job. Where personal data is needed, for example to follow a customer's history, we keep only the fields the question requires, restrict access by role, set retention on purpose and keep the data in an EU region under a written data processing agreement. Analytics should make the business clearer without making your privacy notice longer.
Where AI fits
Language models are useful in analytics for specific jobs: sorting free text such as support tickets, survey answers or cancellation reasons into categories, summarising long reports, and flagging unusual figures for a person to check. They are not a substitute for clean data and clear definitions; on messy data they produce confident nonsense faster. When AI is the right next step, see AI consulting. For factories, our guide to manufacturing predictive analytics covers the use cases and data. Our guide to NLP applications covers those jobs in more depth. Pipelines, models and reports that outgrow a spreadsheet are usually built in Python; see Python software development.
How a project runs
- Brief. The decisions you want to make better, and the systems you use today.
- Audit and definitions. Sources, gaps and one written definition per metric, billed by the hour against a written estimate.
- First report on live data. One question answered end to end, from source to report, so the pipeline is proven early.
- Extend. The remaining questions, one at a time, each in use before the next.
- Hand over. Documentation, training and a review a few months later.
Analytics often reveals where the processes themselves need work. For that, see digital transformation consulting and business process automation. When the data lives in systems that do not fit, see light ERP and custom CRM. When the questions are clear and the numbers need a home, see dashboard development, or, if your reporting runs on Power BI, Power BI consulting. For property portfolios, see predictive real estate analytics. For the four types of analytics and which one your decisions need, see our guide to types of analytics. For forecasting demand, promotions and staffing in shops, read predictive analytics for retail. See prices. For health and care data, see healthcare analytics solutions; for customer scores in marketing, see predictive analytics for marketers.
Questions we get
We already have dashboards. Why would we need consulting?
Because a dashboard nobody trusts is worse than none. The common problems are not technical: two reports define revenue differently, a key number comes from a spreadsheet only its author can update, or the dashboard answers questions nobody is asking. We fix the definitions and the data first, then the dashboards.
Which tools do you use?
The ones that fit your size and the systems you have. Often that is a reporting tool such as Looker Studio, Power BI or Metabase on top of a small database or a cloud warehouse such as BigQuery. We take no commission on any tool, and everything is set up in accounts in your company's name.
Do we need a data warehouse?
Not always. If your data comes from one or two systems with good built-in reports, a warehouse is overkill. Once numbers have to be combined from several systems, or history must be kept longer than a tool keeps it, a small warehouse is usually the cheaper path in the long run.
Can you add AI to our analytics?
Where it helps, yes: classifying free text such as support tickets or reasons for cancellation, or flagging unusual figures. It only works on clean, well-defined data, so that comes first. See our AI consulting service for the wider picture.
How is personal data handled?
By the GDPR's rules and by keeping as little of it as the question needs. Most business reporting works on aggregates. Where personal data is needed, it stays in an EU data region, access is by role, and retention is set on purpose.
What do we have at the end?
Written metric definitions, data pipelines that run by themselves, reports your team uses, and documentation so any competent analyst can continue. All of it sits in your company's own accounts.
Get a written quote
Tell us in a few lines what you need. We reply within one working day with a written quote: scope, timeline and a fixed price or an hour estimate. Asking is free and commits you to nothing.