Power BI consultancy in Edinburgh: what it costs, what breaks and what to build first

Almost every Power BI problem we are asked to look at in Edinburgh arrives described as a performance problem and turns out to be a definition problem.

Edinburgh is large enough that most organisations here already have several analytics teams and the problem is reconciling them. With around 510,000 people in the city itself and roughly 900,000 across the wider urban area and an economy built on asset and wealth management, retail and commercial banking, and insurance and reinsurance, the reporting questions here are specific rather than generic, and so are the ways they go wrong. This page sets out what a Power BI engagement in Edinburgh actually involves: the metrics we check first, what it costs, how licensing should be chosen, whether you would be better off hiring, and what the compliance position looks like under UK GDPR and the Data Protection Act 2018.

Why reporting breaks in Edinburgh specifically

Generic advice about Power BI is not very useful, because the failure modes follow the industry rather than the tool. Edinburgh's economy is weighted towards asset and wealth management, retail and commercial banking, and insurance and reinsurance, with one of Europe's larger asset management and pensions centres and a strong data science and informatics research base shaping much of the local employment base. That mix produces a predictable set of reporting problems, and we have seen each of them here more than once.

Retail and commercial banking

High-volume transactional data at account and posting grain, with a core banking system that was never designed to be queried analytically and a general ledger that is authoritative but coarse.

The characteristic failure is this: each function builds its own extract from core banking with its own as-at timing, and the resulting small variances are never reconciled. They are simply explained away each month. Where that has already happened, the first deliverable is not a dashboard. It is a written reconciliation that establishes which number is correct and why. In practice we start by building a conformed account and customer dimension with a balance and transaction fact, which immediately resolves the usual disagreement between finance, risk and the front line.

Typical source systems: core banking platform, general ledger, card and payment processors, credit decisioning systems and CRM. Account-level data is both personal data and commercially sensitive, so row-level security and export control are design requirements rather than later hardening.

Tourism, hospitality and events

Booking, channel and rate data with heavy seasonality, plus on-site spend and occupancy that only reconcile once channel commission is accounted for.

Here the recurring problem is different: gross booking value from online travel agents is compared with direct booking revenue, making the most expensive channel look like the best one. The fix is a stay-grain model joining rooms, ancillary spend and channel cost, which is where true channel profitability becomes visible. Guest data is personal data with cross-border transfer implications when group systems sit outside the region.

Software and technology

Product event streams at very high volume alongside subscription, billing and support data that must be joined at account grain to be useful. ARR comes from a finance spreadsheet with manual adjustments, so restating a prior period requires archaeology and diligence becomes painful.

We would usually begin with an effective-dated contract and movement model so ARR is reproducible for any historical date, then product events joined at account grain. Source systems to expect: billing and subscription platform, CRM, product event pipeline, support platform and finance system.

The metrics we check first in a Edinburgh engagement

Sector knowledge in analytics is mostly knowing which definitions destroy a number when they drift. Before we design anything we go through the measures that matter to you and establish, in writing, exactly how each one is calculated and who owns that definition. The table below is the starting list for Edinburgh, derived from the industries that dominate here.

None of this is exotic. Where two definitions are both defensible, an organisation ends up carrying both, and the monthly difference between them becomes a recurring agenda item rather than a resolved question. The output of this stage is a measure dictionary: every KPI, its business definition, its owner, and the DAX that implements it.

What a first Power BI engagement looks like

We work the same way whether the client is a large group with nine existing workspaces: decisions first, sources second, model third, reports last. Reports built before the model are the single most common cause of rework.

1. Decision inventory, week 1We start from the recurring decisions that need numbers, not from the data that happens to be available. Who decides, on what cadence, and what currently blocks them. For a Edinburgh business in retail and commercial banking, that usually surfaces three or four decisions worth building for and a dozen reports nobody needs.
2. Source profiling, weeks 1 to 2Each source system profiled for grain, volume, change tracking, quality and refresh window. core banking platform, general ledger and card and payment processors are the usual candidates here. This is where unpleasant surprises belong, not in week ten.
3. Model build, weeks 3 to 7A star schema with conformed dimensions, measures written against the agreed definitions, and row-level security applied from the start. For retail and commercial banking the argument that has to be settled first is normally net interest margin, where the averaging basis for earning assets must be identical across every report that uses it. Settle it before any report exists, not after two of them disagree.
4. Report build and validation, weeks 6 to 10Reports built with the people who will use them and validated line by line against the source of record until the numbers tie out. Disagreements get resolved and written down, never averaged.
5. Handover, weeks 10 to 14Workshops with your team, documentation in your own systems, and a thirty-day period where we fix what breaks while your people drive. The University of Edinburgh's informatics and data science strength makes the local market unusually deep in modelling and machine learning skills.

What that costs

Build cost follows project size rather than report count, the number of source systems and their state drive almost all of it. These are the bands we quote in United Kingdom, and we publish them because a buyer cannot start an approval chain without a budget figure.

Running it afterwards is a separate line: £3,000 to £8,000 a month for a small estate, £8,000 to £20,000 for a mid-sized one, £20,000 to £50,000+ for a large multi-domain platform. Microsoft licensing is separate again, and is covered next.

Licensing: what Edinburgh organisations should actually pay

This is where money is most often wasted, in both directions. Power BI Pro is listed at around US$14 per user per month and Power BI Premium Per User at around US$24 following the 2025 price revision. Local list prices in sterling, euro and Swiss francs track those figures with regional adjustment, and enterprise agreements usually land below list.

The arithmetic that matters: F64 only becomes cheaper than per-user licensing somewhere in the region of 350 to 600 readers, depending on the discount you actually pay and whether you reserve capacity. Below that, Pro or PPU per user is normally the right answer, and buying capacity for prestige is a straightforward waste of money.

For larger Edinburgh organisations with hundreds of report consumers, a capacity SKU is usually the right answer, and which one depends on measured consumption rather than headcount. We profile workloads on a short-lived trial capacity before anyone signs.

Always price from the current Microsoft rate card for your region and agreement. Any figure in an article older than a quarter should be treated as an order of magnitude, not a quote.

Hiring in Edinburgh versus engaging a consultancy

This is a fair question and the honest answer is sometimes "hire". The University of Edinburgh's informatics and data science strength makes the local market unusually deep in modelling and machine learning skills. Locally, a mid-level analytics engineer costs roughly £45,000 to £75,000 outside London, and £60,000 to £95,000 in London. An experienced external consultant runs at roughly £750 to £1,400 per day, and more through a large systems integrator.

Where a Edinburgh client has told us they intend to build an internal team, we design for it from the start. Teams here expect to own what we build, and we design for that explicitly: scriptable models, conventional naming, documentation in their wiki, and a measure dictionary that survives our departure. Several clients have gone on to need us only for quarterly review, which is a perfectly good outcome.

Data residency and compliance for Edinburgh organisations

Azure UK South and UK West give genuine in-country data residency, and Power BI tenants can be provisioned in the United Kingdom region. Note that some workloads, certain preview features and a handful of AI services, still process outside the region, so residency claims should be checked per service rather than assumed tenant-wide. In practice, Edinburgh deployments normally use UK South (London), with UK West (Cardiff) as the paired region.

Regulatory oversight relevant to analytics comes from the Information Commissioner's Office (ICO) under UK GDPR and the Data Protection Act 2018, plus the sector bodies that matter here: FCA and PRA for financial services conduct and prudential reporting, ICO for data protection, including analytics on personal data and Ofgem and Ofwat for energy and water price-control and performance submissions.

• UK GDPR data-minimisation and storage-limitation duties apply directly to analytics copies of personal data, which is where most estates quietly fail.
• Consumer Duty outcome monitoring has made customer-level analytics a supervisory expectation for regulated firms, not an optional capability.
• SECR energy and carbon reporting requires auditable activity data, which usually means the same lineage discipline as financial reporting.

We cover this properly on the data governance and compliance page, and in more depth in our article on data governance for Edinburgh organisations.

How we deliver into Edinburgh

• Most UK engagements run remote-first with on-site workshops for definition sessions and go-live.
• IR35 and off-payroll rules make a fixed-scope statement of work cleaner than a day-rate contractor arrangement for many clients.
• Public-sector buyers typically route analytics work through established frameworks rather than open tenders.

Delivery is remote-first with on-site workshops at the points where being in the room genuinely helps: definition sessions, validation and go-live. We work in English.

How a Edinburgh organisation actually buys this

Most consultancy pages stop at what the work is and leave the buying process to email. Relatively fast and relationship-led, with procurement entering late unless the value is high. Day rates and fixed prices are both normal, and buyers expect them early rather than after three meetings.

In the private sector: Introduction or inbound enquiry, a technical call, then a written proposal with a fixed price or a day rate. Above roughly £50,000 a procurement team joins and the supplier is checked for insurance, financial standing and security posture. Many larger organisations buy only from a preferred-supplier list, so getting onto that list is a separate exercise from winning the work.

In the public sector: Public bodies buy through Crown Commercial Service frameworks, most often G-Cloud on the Digital Marketplace for cloud and analytics services, or through Digital Outcomes-style routes for delivery teams. Above-threshold requirements are advertised on Find a Tender, and the Procurement Act 2023 regime governs the process. Direct award is possible under a framework and is far quicker than an open competition.

Documents you should expect to request, and that any supplier worth shortlisting will have ready, are certificate of incorporation and latest filed accounts, professional indemnity, public liability and employer's liability insurance certificates, cyber essentials or cyber essentials plus, effectively mandatory for public-sector work and data processing agreement under uk gdpr article 28.

Our own entity details, insurance, processing agreements and contracting shapes are published on the procurement page, so your buying team can start without waiting for a reply.

Signals it is time to call someone

The usual triggers are two reports disagreeing on a headline number, an overnight refresh someone re-runs by hand each morning, a largest model that is one file owned by one person, finance still rebuilding the board pack in a spreadsheet, a legacy BI renewal whose price no longer makes sense, or a question about who can see a sensitive dataset that took a week to answer honestly. If two or more are true, a two to three week paid discovery will tell you what to do about it, and you keep the output whether or not you engage us further. Book a technical review, or read what we actually build:

We work across United Kingdom and deliver into neighbouring markets regularly. Nearby pages worth reading if your group has sites elsewhere: London, Leeds, Bournemouth, Brighton and Hove, Cheltenham and Chester.

Frequently asked questions