Power Apps or a custom business app? Every growing UK business hits this fork eventually. A spreadsheet has quietly become a system, a paper process is falling over, or an off-the-shelf tool almost fits but not quite — and someone has to decide whether to build the replacement on Microsoft’s low-code platform or commission software that is yours outright. The two routes price completely differently, fail differently, and suit different situations, which is why a cybersecurity-grade level of scrutiny on the licence small print pays for itself many times over.
This guide puts real UK figures on both sides of the decision. You will see what Power Apps genuinely costs once licences, Dataverse storage and build days are added up, what a custom build costs at current UK day rates, and where each route runs out of road. There is a worked example for a 40-person business, and a seat-count crossover calculation that tells you when the answer flips.
The short version: for internal tools used by fewer than roughly a hundred people, Power Apps is usually the cheaper three-year decision — but the limitations arrive later, in delegation ceilings, request limits and a licence bill that scales with headcount forever. A custom business app costs more up front, then stops scaling with your team size. The rest of this article shows the arithmetic behind both claims.
Table of contents
- What Power Apps Is — and What Counts as a Custom Business App
- Power Apps Pricing in the UK: The Licence Bill Nobody Itemises
- What a Custom Business App Costs to Build in the UK
- Power Apps vs Custom Business App: The Head-to-Head
- The Power Apps Limitations That Surface After the Demo
- Where a Custom Business App Struggles
- A Worked Example: 40 Users, Three Years
- The Seat-Count Crossover: When Power Apps Stops Being Cheaper
- Best Fit: Which Route Matches Your Situation
- A Hybrid Route: Power Apps First, Custom Where It Earns It
- How to De-Risk Whichever Route You Choose
- Frequently Asked Questions
- References
What Power Apps Is — and What Counts as a Custom Business App
Power Apps is Microsoft’s low-code application platform, part of the wider Power Platform alongside Power Automate, Power BI and Copilot Studio. You assemble screens, forms and logic in a visual designer, connect them to data through several hundred pre-built connectors, and publish the result to your team’s browsers and phones without managing any servers.
The two flavours of Power Apps
Canvas apps give you a blank screen and pixel-level layout control — good for task-shaped tools like inspection forms and approval trackers. Model-driven apps generate their interface from the data model in Dataverse — better for record-keeping systems with views, forms and business rules. Both are built by configuration rather than code, which is where the speed comes from.
What we mean by a custom business app
A custom business app, for this comparison, is software written for you: typically a web application with its own database, hosted in your cloud subscription, built by an in-house developer, a contractor or an agency. You own the source code, you choose the stack, and nothing about it is metered per user. The trade is obvious even before the numbers: you also own every bug, every upgrade and every security patch.
Why this comparison keeps coming up
Microsoft bundles limited Power Apps rights into Microsoft 365, so most UK businesses already have a foot on the platform. The moment a team outgrows those seeded rights, the licence conversation starts — and the honest comparison is not “free vs expensive” but “per-user forever vs one-off plus maintenance”. Our guide to build vs buy vs low-code decisions covers the strategic framing; this article does the money.
Power Apps Pricing in the UK: The Licence Bill Nobody Itemises
The demo is free. The platform is not. Power Apps pricing looks simple on Microsoft’s page — one headline figure per user — but the real bill has four or five lines, and two of them are easy to miss until they arrive.
The current UK price list
These are Microsoft’s published UK prices as of August 2026, excluding VAT, on annual commitment. The per-user licence is the one most businesses end up on.
| Licence or add-on | UK price (ex VAT) | What it actually buys |
|---|---|---|
| Power Apps Premium | £15.40/user/month (£184.80/user/year) | Unlimited apps, premium connectors, Dataverse, 40,000 platform requests per user per day |
| Power Apps Premium (2,000+ seats) | £9.20/user/month | Volume tier — needs a 2,000-licence minimum, so out of reach for SMEs |
| Power Automate Premium | £11.50/user/month | Only needed for flows that run outside the app — in-app flows are covered by the app licence |
| Dataverse Database capacity | £30.80/GB/month (£369.60/GB/year) | Extra structured storage once the included capacity runs out |
| Copilot Studio | £153.80/month per 25,000 credits | Conversational agents on top of your apps — a separate meter entirely |
| Microsoft 365 seeded rights | £0 extra | Standard connectors only, 6,000 requests per user per day — no Dataverse-backed apps |
The three lines people forget
First, storage. A tenant’s default environment includes 3GB of Dataverse database capacity, 3GB of file capacity and 1GB of log capacity, and every additional environment consumes at least 1GB. A record-heavy app crosses those thresholds sooner than expected, and each further gigabyte of database is £369.60 a year.
Second, premium connectors. The seeded Microsoft 365 rights cover standard connectors only. The moment your app touches SQL Server, a third-party API or Dataverse itself, every user of that app needs a Power Apps Premium licence — not just the people who built it.
Third, the build itself. Power Apps removes servers, not effort. A production-quality app still needs requirements, data design, screens, testing and rollout. UK Power Platform contractors bill a median £500 a day, so a 25-day build is £12,500 before a single licence is assigned.
What the platform request limits mean
Each Power Apps Premium user can make 40,000 platform requests per 24 hours; Microsoft 365 seeded users get 6,000. Those sound enormous until you remember that every connector call and every Dataverse read or write counts. A screen that refreshes a gallery against a large table can burn through a seeded user’s allowance in a busy day, and the fix — Premium licences or a 50,000-request capacity add-on — is another line on the bill.
What a Custom Business App Costs to Build in the UK
The custom route prices in days, not seats. The scope we will use throughout this article is a single-workflow business app: six to eight screens, a database, role-based access, a reporting view and an integration or two — the same scope a mid-sized team would put on the low-code platform.
UK day rates in 2026
| Who builds it | Typical day rate | Notes |
|---|---|---|
| In-house developer (£55,000 salary) | £282.60/day loaded cost | Salary plus employer NI and pension, spread over 227 working days |
| Contractor, outside London | £450/day | Power Platform and general development rates converge here |
| Contractor, UK median | £500/day | ITJobsWatch median for Power Platform contract roles, six months to August 2026 |
| Mid-level agency developer | £550–£750/day | Includes project management overhead |
| Senior developer / architect | £700–£950/day | Needed for the data model and integration design, rarely full-time |
| Offshore team | £150–£300/day | Cheaper days, more of them — budget for coordination |
The build budget for our reference scope
At an agency rate of £550 a day, the reference scope needs roughly eight architect days at £850 for the data model and integration design (£6,800) plus 55 build days at £550 (£30,250) — £37,050 in total, or £44,460 once a sensible 20% contingency is applied. That figure sits comfortably inside the £30,000–£60,000 band we see across comparable UK projects in our custom software development cost guide.
The running costs people underestimate
A custom app is never finished. Hosting for this scope is around £3,000 a year in a UK cloud region. Maintenance — dependency updates, small changes, security patches — runs at 15–25% of the build cost annually; we use 20%, which is £8,892 a year on a £44,460 build. Skip that line in the budget and the app becomes the legacy system someone else is asked to replace in five years. Our business process automation cost guide shows the same discipline applied to process projects.
Power Apps vs Custom Business App: The Head-to-Head
Price is one column of a longer table. Here is how the two routes compare on the factors that actually decide the outcome, before any worked example.
| Factor | Power Apps | Custom business app |
|---|---|---|
| Up-front cost | Low — configuration days only | High — full design and build |
| Ongoing cost shape | Per user, forever | Flat hosting and maintenance |
| Time to first release | 2–6 weeks | 2–5 months |
| User experience ceiling | Good forms and lists; bespoke interfaces fight the designer | Whatever you are willing to pay for |
| Large data sets | Delegation rules cap non-delegable queries at 500–2,000 rows | Limited only by your database design |
| External users | Licensing gets complicated fast | Straightforward — it is your app |
| Ownership | Configuration on a rented platform | Source code is a company asset |
| Price-rise exposure | Full — Microsoft sets next year’s rate | Limited to hosting and day rates |
| Skills needed to change it | Widely available; power users can help | Developers only — key-person risk is real |
| Security and patching | Microsoft’s problem below the app layer | Yours, end to end |
The pattern behind the table: Power Apps converts capital cost into operating cost and engineering risk into licence dependency. A custom build does the reverse. Neither conversion is free, and the right choice depends on which risks your business is better placed to carry.
The Power Apps Limitations That Surface After the Demo
None of these are secrets — every one is documented by Microsoft — but they rarely come up in the sales conversation, and each one has ended a project that assumed the platform was infinitely elastic.
The delegation ceiling on large data
When a query cannot be translated (“delegated”) to the data source, the platform pulls the first 500 records — configurable up to a hard maximum of 2,000 — and runs the query locally on that slice. Beyond that, results are silently wrong, not slow: a search over 10,000 job records that uses a non-delegable function will simply miss most of them. Dataverse, SharePoint and SQL Server delegate well for common operations, but complex filters, joins and aggregations hit the wall quickly. Designing around delegation is the single biggest skill gap between a demo app and a production one.
The request limits behind the scenes
The 40,000 requests per Premium user per day and 6,000 per seeded user are genuine ceilings, and every connector call, refresh and Dataverse operation counts toward them. Well-designed apps stay far below the limit; chatty screens that poll data sources do not. When a business automates aggressively — apps plus flows plus integrations — the request budget becomes a shared resource that needs actual management.
Storage that meters in gigabytes, not terabytes
Dataverse is priced like a premium database because it is one: £30.80 per GB per month buys you backup, security and a rich data platform, but ten years of attachments and audit logs adds up fast when the default environment starts at 3GB of database capacity. Custom apps typically pay a tenth of that or less for raw cloud database storage — one of the few line items where the custom route is dramatically cheaper.
The interface ceiling
Power Apps produces clean, functional forms, lists and dashboards. What it resists is bespoke interaction: drag-and-drop scheduling boards, complex visualisations, anything animation-heavy or pixel-perfect against a brand guideline. Teams either accept the platform look or burn budget fighting it — and the fighting days cost the same £500 as the productive ones.
Governance and application lifecycle
A serious deployment needs environments (development, test, production), solution packaging, and rules about who can build what against which data. That is application lifecycle management, and skipping it is how a business ends up with 200 orphaned apps built by people who have since left. The governance work is not optional overhead; it is part of the true cost of the platform, just as version control and CI pipelines are part of a custom build.
The dependency you cannot negotiate
Finally, the structural one: every Power Apps price in this article is set by Microsoft and revised at Microsoft’s discretion. UK price lists have moved with currency and strategy before and will again. A custom app’s running costs float on cloud prices and day rates, which are competitive markets. This is the same platform-dependency argument we examined for Microsoft Copilot versus a custom AI assistant — the buy side is renting, and rent goes up.
Where a Custom Business App Struggles
Symmetry matters: the custom route has failure modes of its own, and they are just as predictable.
Maintenance is a subscription you pay yourself
That 15–25% annual maintenance band is not padding. Frameworks deprecate, browsers change, dependencies publish security fixes weekly. A custom app with no maintenance budget is not cheaper — it is deferring its costs into a future rebuild, with interest.
Key-person risk
The developer who built your app holds a model of it in their head. When they move on, that model leaves with them unless you paid for documentation, tests and handover. Agencies mitigate this with teams; a single in-house builder or one contractor does not. Ask who else could confidently change the system, and treat “nobody” as a project risk with a price.
Security is yours, end to end
On the custom route you own authentication, patching, backups, penetration testing and incident response. None of this is exotic, but all of it is real work that the platform route buys pre-packaged. The NCSC’s cloud security guidance is the right baseline for a UK business hosting its own application, and meeting it is part of the honest cost comparison.
Speed, or the lack of it
A custom build measured in months competes with a Power Apps build measured in weeks. If the process you are fixing bleeds money daily, three months of extra waiting is a cost line too — at £2,000 a month of process waste, a ten-week delay adds £5,000 to the true price of the custom route before any invoice arrives.
A Worked Example: 40 Users, Three Years
Take a 40-person field operations team replacing a spreadsheet-and-email job tracking process. Same scope both ways: six to eight screens, job records, photos, role-based access, a weekly reporting view.
The Power Apps route
Build and rollout: 25 contractor days at £500 is £12,500, which becomes £15,000 with 20% contingency. Licences: every user touches Dataverse, so 40 Power Apps Premium seats at £184.80 a year is £7,392 a year. Iteration: five days a year of changes at £500 is £2,500 a year. Three-year total: £15,000 + 3 × (£7,392 + £2,500) = £15,000 + £29,676 = £44,676.
The custom route
Design and build: £6,800 of architect time plus £30,250 of build days is £37,050, or £44,460 with the same 20% contingency. Hosting: £3,000 a year. Maintenance at 20% of build: £8,892 a year. Three-year total: £44,460 + 3 × (£3,000 + £8,892) = £44,460 + £35,676 = £80,136. No per-user licences at any headcount.
Side by side
| Cost line | Power Apps route | Custom route |
|---|---|---|
| Build (with 20% contingency) | £15,000 | £44,460 |
| Licences per year (40 users) | £7,392 | £0 |
| Hosting per year | Included | £3,000 |
| Changes / maintenance per year | £2,500 | £8,892 |
| Three-year total | £44,676 | £80,136 |
At this headcount the low-code route costs 56% of the custom route over three years — a £35,460 gap.
What would change the answer
Three things swing this result: user count (next section), data volume (heavy storage erodes the licence advantage at £369.60 per GB per year), and interface ambition (bespoke UI work on the platform burns days without ever quite arriving). If any of the three is extreme, rerun the sums before deciding.
The Seat-Count Crossover: When Power Apps Stops Being Cheaper
The structural difference between the routes is that one bill scales with headcount and the other does not. That makes the crossover point calculable.
The arithmetic
From the worked example, the Power Apps route’s fixed three-year cost is £15,000 build plus £7,500 of iteration = £22,500, and each seat adds 3 × £184.80 = £554.40 over three years. The custom route is a flat £80,136 at any headcount. The crossover is therefore (£80,136 − £22,500) ÷ £554.40 = 104 users. Below that, the licence route wins on three-year cost; above it, the custom build does.
Reading the crossover honestly
The 104-user figure belongs to this scope and these rates — it is a method, not a constant. A heavier build moves it up; a leaner custom option or the 2,000-seat volume tier moves it down. What generalises is the shape: per-seat pricing always crosses a flat line eventually, so any Power Apps business case should state the seat count at which its own numbers flip, and how fast the team is growing toward it.
Growth is the hidden variable
A 40-person team that doubles in three years spends licence money like an 80-person team in years two and three. If your headcount plan crosses the crossover during the life of the system, the custom route’s flat line is worth more than the year-one comparison suggests.
Best Fit: Which Route Matches Your Situation
| Your situation | Best fit | Why |
|---|---|---|
| Internal tool, under ~100 users, standard forms and lists | Power Apps | Cheapest three-year route by a wide margin |
| Process bleeding money now, needs fixing in weeks | Power Apps | Time to first release is the deciding cost |
| Hundreds of users, stable process | Custom build | Past the seat-count crossover, the flat line wins |
| Customer-facing product or portal | Custom build | External licensing and the interface ceiling both bite |
| Large or complex data, heavy reporting | Custom build | Delegation limits and Dataverse storage pricing work against you |
| Already deep in Microsoft 365, IT team is small | Power Apps | Security, hosting and identity come pre-solved |
| The app IS the business’s competitive edge | Custom build | Differentiating capability should be an owned asset |
| Unsure the process is even right yet | Power Apps first | Cheap iteration is worth more than ownership while the process is still moving |
A Hybrid Route: Power Apps First, Custom Where It Earns It
The comparison is not actually either/or, and the strongest pattern we see in UK businesses treats the two routes as stages rather than rivals.
Prototype on the platform, harden in code
Build the first version in Power Apps in weeks, run it with real users, and let it prove — or disprove — the process. Most internal tools stay there happily. The minority that outgrow the platform migrate to a custom build with something priceless in hand: a validated specification, written by months of real usage rather than workshops. The £15,000 platform build is not wasted; it is the cheapest requirements document you will ever buy.
Split the estate, not the app
The same logic applies across a portfolio. Put the standard forms-and-lists tools on Power Apps where licence costs are modest, and spend custom budget only on the one or two systems where data volume, user count or interface ambition justify it. The automation equivalent of this split — flows for the simple work, code where the licence maths fails — is covered in our Power Automate vs custom development guide.
Keep the data model portable
Whichever route you start on, the data model is the part worth designing as if you will move. Clean tables, documented fields and sensible keys migrate; a tangle of app-specific shortcuts does not. Business process automation projects live or die by the same rule.
How to De-Risk Whichever Route You Choose
Pilot with a real slice of data
Demo data hides both failure modes. Load a realistic volume before committing: delegation problems and slow screens appear on the platform, and data-model mistakes appear in a custom build — both while they are still cheap to fix.
Write the exit plan first
For Power Apps: how would you export the data and rebuild elsewhere, and what would it cost? For custom: who else could maintain this, and where is the documentation? An exit plan you never use costs a page of writing; needing one you don’t have costs a rebuild.
Budget the whole line, not the first invoice
Both routes have a second-year bill — licences on one side, maintenance on the other. Put both in the business case at year three, not year one, and compare like with like. Every table in this article is a template for doing exactly that.
Get the governance in early
Whoever builds it, someone must own environments, access, backups and change control from day one. On the platform that is ALM and environment strategy; on the custom route it is version control, CI and patching. The route that skips governance is always the expensive one, whatever the licence bill says.
Frequently Asked Questions
Is Power Apps free with Microsoft 365?
Partly. Microsoft 365 licences seed limited rights: standard connectors only and 6,000 platform requests per user per day. The moment an app needs Dataverse, SQL Server or any premium connector, every user needs Power Apps Premium at £15.40 a month. Most production business apps cross that line.
How long does a Power Apps build take compared with custom?
For the reference scope in this article: two to six weeks on the platform against two to five months for a custom build. The gap narrows on complex apps, because delegation workarounds and governance take real time too.
Can Power Apps handle 50,000 records?
Yes, if the app is designed for delegation — Dataverse, SharePoint and SQL Server can all filter server-side. The danger is non-delegable queries, which silently operate on only the first 500–2,000 rows. This is a design skill issue more than a hard platform ceiling, but it is the most common cause of “the app shows the wrong numbers”.
Do I still need developers if we choose Power Apps?
For anything production-grade, yes — the job title changes more than the need. Data modelling, integration design and lifecycle management are engineering work whichever tool renders the screens. What changes is that configuration-level changes stop needing an engineer.
What happens to our app if Microsoft changes Power Apps pricing?
You pay the new price or migrate. That is the honest answer, and it is why the exit plan section above exists: the platform’s three-year cost advantage is real, but it is a tenancy, not a freehold. Price the risk rather than pretending it is zero.
References
Power Apps Pricing — Microsoft UK
Requests Limits and Allocations — Microsoft Learn
Understand Delegation in a Canvas App — Microsoft Learn
Dataverse Capacity-Based Storage Details — Microsoft Learn
Pay-As-You-Go Overview — Microsoft Learn
Application Lifecycle Management Overview — Microsoft Learn
Create Performant Canvas Apps — Microsoft Learn
Cloud Security Collection — National Cyber Security Centre
Annual Survey of Hours and Earnings — Office for National Statistics