In this guide:
- Three UK cost bands for 2026 - matched to project complexity, with April day-rate evidence.
- The 9 drivers that actually move app cost - including integrations, compliance and AI features.
- A decision framework for native vs cross-platform vs PWA, plus UK funding and tax-relief routes.
Every year, thousands of UK SMEs commission a mobile or web app. Most are surprised by the final bill. A 2025 Deloitte study, cited across UK industry reporting, found that 54% of British businesses commissioning a software build ended up spending more than 40% above their original budget. That is not a tech problem. It is an expectations problem - and it is almost always fixable with better information before the work starts.
The public sector offers its own cautionary tales. The NHS COVID-19 app launched four months late and cost around £36m (NAO FOI data). HMRC's Making Tax Digital programme has seen its cost increase by 400%, with the NAO projecting a total £1.27bn.
Whatever the scale, the pattern repeats: weak Discovery, underestimated integrations, missed compliance, and scope creep that nobody catches until the invoice arrives. Those lessons apply just as much to a £40,000 SME app as they do to a national IT programme.
This guide is written for UK decision-makers: founders, managing directors, operations leads, finance directors and first-time CTOs who are considering commissioning an app in 2026. It covers what apps actually cost right now, which levers move the price, where the hidden costs live, how to pick between native and cross-platform, what UK funding is available, and how to tell whether an app is even the right next step for your business.
UK context matters. The UK app development industry was worth £28.3bn in 2024-25, employing around 73,470 people across 14,527 businesses (LuminaryBrands). There are 99.3 million UK mobile connections for a population of under 70 million, and UK users average over five hours per day on mobile devices (DataReportal Digital 2026).
The appetite for well-made apps has rarely been higher. The question is not "should I build an app?" - it is "what is the cheapest sensible way to find out whether mine will work?"
We are a UK-based software development house. We build on .NET and Azure, so the worked examples lean that way, but the numbers and principles apply whatever your stack. When we quote ranges, we cite sources and dates. When we estimate, we show our working. The aim is that you finish this guide knowing what a realistic number looks like for your specific situation - not just another "it depends".
1. What counts as 'an app' in the cost conversation
Before we talk numbers, a word about scope. When a UK SME says "we want to build an app", they usually mean one of five very different things, each with its own economics:
- Native mobile app - built specifically for iOS (Swift/SwiftUI) and/or Android (Kotlin). Full access to the device, best performance, highest cost per platform.
- Cross-platform mobile app - one codebase covering both iOS and Android via a framework like .NET MAUI, Flutter or React Native. Roughly 60-75% of the cost of building both natively.
- Progressive web app (PWA) - a responsive web app that behaves like a native app, can be installed to the home screen, and works offline. Typically 40-60% cheaper than native.
- Responsive web app - a browser-based application accessed through a URL. No app store. Best when you do not need offline mode, push notifications or deep device integration.
- Hybrid / wrapper app - a web app wrapped in a native shell (Capacitor, Cordova). Low cost, limited device access, increasingly being replaced by cross-platform frameworks or PWAs.
The difference between the cheapest sensible option for your use case and the most expensive can be a factor of five or more. If your only reason for "going native" is that the boardroom expects it, that is an expensive habit to be aware of before signing anything. We cover the economics of each option in detail in Section 4. For a broader discussion of what bespoke software is in the first place, our guide to what bespoke software is is a good primer.
2. UK app cost bands in 2026
Most UK app projects fall into one of three cost tiers. The numbers below synthesise current UK agency pricing, freelance market data and our own project experience for April 2026. They assume a single mobile platform or a cross-platform build; expect roughly +50-70% for a second native platform (more on that in Section 4).
UK app cost bands, April 2026
| Tier | UK range (GBP) | Timeline | Typical scope |
|---|---|---|---|
| Simple | £8,000 - £30,000 | 8-12 weeks | Single-workflow MVP, 3-6 screens, basic auth, simple data display, minimal integrations. One platform or lightweight cross-platform. |
| Standard | £30,000 - £80,000 | 12-24 weeks (3-6 months) | Production SME app, 5-15 screens, user accounts, payments or 1-2 third-party integrations, basic backend, push notifications, analytics, polished UX. |
| Complex | £80,000 - £300,000+ | 6-12+ months | Feature-rich platform, multiple integrations, real-time features, advanced security or regulated data (fintech, healthcare), multi-role admin portals, multi-platform, AI features. |
For context, Clutch's directory data pegs the average UK app project budget at around £90,780 as of February 2026, with most projects between £10,000 and £49,999. The average is pulled up by the tail of very large builds; where most SMEs land is the £30k-£80k standard band.
If your project includes multiple native platforms, heavy regulatory work (fintech, health), real-time infrastructure, complex integrations or an AI core, budget for the complex band and treat the lower number as optimistic. For a wider view of custom software costs beyond just apps, see our bespoke software cost guide, which covers enterprise-scale internal systems too.
Typical cost ranges by UK SME vertical
Different verticals cluster at different points on these bands. Synthesised from UK agency pricing data and our own project experience, here is a quick reference for April 2026:
| Vertical | Typical UK range | Comment |
|---|---|---|
| Field service / trades (plumbers, electricians, cleaners) | £40k-£120k | Scheduling, job management, offline mode, integrations with BMS / accounting |
| Hospitality (bookings, ordering) | £40k-£120k | Menu, booking, payments, POS integration; multi-venue adds to upper band |
| Retail ecommerce app (Shopify / WooCommerce / bespoke) | £60k-£150k | Catalogue, checkout, loyalty, push; inventory integration drives cost |
| Property / lettings | £50k-£150k | Listings, viewing bookings, agent CRM; compliance work for AML/KYC adds cost |
| Healthcare (private practice) | £50k-£150k (higher for regulated clinical) | GDPR, clinical compliance and safe data handling materially change scope |
| Fitness / wellness | £25k-£120k | Subscriptions and video content are the main cost drivers |
| Professional services (legal, accounting) | £25k-£80k | Often client portals or document-exchange rather than full native apps |
| Construction / site management | £40k-£150k | Bespoke workflows and asset tracking; offline mode essential on site |
Ranges synthesise UK agency cost bands from multiple 2025-2026 sources (Nordstone, Clutch London, OrionByte, ScaleUpAlly, RiseUpLabs) and our own project experience. Individual quotes will vary.
Want a straight-talking estimate for your app idea? These bands cover most UK SME projects but every app is different. If you'd like a no-nonsense view of what your specific idea might actually cost, timeline, and team shape - and an honest opinion on which platform approach fits - get in touch for a free 30-minute scoping conversation.
Estimate your own project in under 2 minutes
Before booking a call, try the free cost estimator. Answer a few short questions about your project - scope, platforms, integrations, compliance - and get an instant indicative range based on typical UK market rates. It uses the same bands you've just read about.
Bespoke software cost estimator
Get a ballpark cost for your bespoke software project. Answer a few questions about your requirements and get an instant estimate.
3. The 9 things that actually drive app cost
Headline bands give you a rough fit. The real number depends on nine variables. In our experience, the gap between a £40,000 and a £120,000 quote for apparently similar-sounding requirements almost always comes down to how these are scoped.
3.1 Feature scope and complexity
By far the biggest variable. The difference between "a handful of screens and a login" and "a marketplace with messaging, payments, review flows, notifications and a vendor admin portal" is an order of magnitude of build effort. The industry rule of thumb: for every feature added, time-to-build compounds because each new feature has to integrate with everything already built. Keeping feature scope tight at launch is the single most powerful cost-control lever. Our MVP development guide covers how to do this well.
3.2 Platform choice
Covered in detail in Section 4 - but in short: single native ~1x, dual native ~2x, cross-platform ~1.2-1.5x, web-first ~0.4-0.6x. The choice is cost-material in both directions.
3.3 Design and UX depth
UX/UI design typically costs 12-18% of a standard app budget. A 6-screen app with bespoke animations, custom illustrations and a deep design system costs more than one built against a mature design library. Expect £3,000-£15,000 for design on a simple app, £10,000-£30,000 for a standard app, and £25,000-£75,000+ for complex apps requiring motion design, accessibility audits and a reusable design system.
3.4 Third-party integrations
Integrations typically account for 10-15% of project budget and are one of the most frequently underestimated drivers. Each integration - payment gateway, analytics, CRM, ERP, messaging, maps, identity provider - adds developer time, testing effort and ongoing maintenance surface. Stripe is usually a 1-3 day job. A legacy on-premise ERP with SOAP endpoints can be 4-8 weeks. Map them all before you quote.
If your app ties into an existing business system (Dynamics 365, Salesforce, a custom back-office), read our system integration guide for what integration work actually looks like in practice.
3.5 Data and backend architecture
The shape of your backend drives cost far more than users usually realise. A read-heavy app with a simple REST API on Azure App Service is a couple of weeks of backend work. A real-time app with WebSocket messaging, event sourcing, complex data models and multi-tenant isolation is months. Infrastructure costs scale similarly - £500-£2,000/month is typical for a standard SME app, rising to £1,500-£3,000+/month at user-base scale.
3.6 Compliance, security and accessibility
For regulated sectors (fintech, healthcare, education, legal) compliance work adds another 10-15% to build cost. UK GDPR applies to every app handling personal data - initial compliance work runs £3,000-£7,000 and annual maintenance £1,500-£5,000. WCAG 2.2 accessibility is mandatory for public-sector apps and commercially expected. Apple's Privacy Manifest requirements (live since 2024) and Google Play's Data Safety declarations now require explicit disclosure of every SDK you use and what data it handles.
Retrofit accessibility is expensive. Building accessibility in from day one typically adds 5-10% to the design and build cost. Retrofitting it after launch can cost 15-25% of the original build. If public-sector users might ever use your app, build for WCAG 2.2 AA from the first sprint - it is cheaper and it is the right thing to do.
If your app carries user-generated content - reviews, comments, messaging, media sharing - the Online Safety Act 2023 applies. Ofcom's fee regime went live on 11 April 2026, with regulated services paying an annual fee of 0.02-0.03% of Qualifying Worldwide Revenue once thresholds are met. Initial notifications were due by 11 April 2026. Non-compliance penalties are large - fines of up to 10% of global turnover or £18m, whichever is higher. Even if fees do not apply to a small UK SME app, the safety duties (age assurance in some contexts, content moderation, reporting) still do. Factor 5-15% additional engineering and ongoing moderation cost for UGC features.
3.7 Testing and QA depth
Testing budgets run 10-15% for simple apps, 10-20% for complex ones. Automated test coverage takes upfront effort but slashes regression cost post-launch. Manual testing across the UK's fragmented Android device landscape (30+ common device configurations) costs £400-£600 per full regression cycle. Skimping on testing is a false economy; one production bug in a payment flow can cost more in lost revenue and reputation than a month of testing.
3.8 Launch and ongoing maintenance
App store submission, TestFlight/Play internal testing cycles, rejection handling and initial launch support add £1,500-£3,000 on top of the core build. Post-launch maintenance runs 15-25% of initial build per year - a £50,000 app needs £7,500-£12,500 annually just to stay current. We cover this in detail in Section 8.
3.9 AI feature integration
Adding one or two AI features to a traditional app - say a conversational support assistant or a document-classification helper - typically adds £15,000-£45,000 to the build and £500-£2,500/month to running costs. Building an AI-native app, where an LLM or reasoning engine is the core of the product, puts you in the complex band from the start. See Section 9 for the token economics.
4. Mobile vs web vs cross-platform: the cost trade-off
Platform choice is the single biggest lever available before you even start sprinting. Here is how the options compare for a standard SME production app:
Platform cost comparison (relative to single native)
| Approach | Relative cost | Best for | Caveats |
|---|---|---|---|
| Native iOS only | 1x (baseline) | Premium/consumer-first UK audience, deep platform features | You miss Android's ~50% UK share |
| Native Android only | ~1x | Workforce apps, mass-market, lower-ARPU audiences | Hardware and version fragmentation adds testing cost |
| Native both (dual build) | ~2x | Apps demanding deep platform-specific behaviour and top-end performance | Two codebases, two teams' worth of ongoing maintenance |
| Cross-platform (.NET MAUI, Flutter, React Native) | ~1.2-1.5x | Most UK SME apps - iOS+Android coverage at lower cost | Slight perf/platform-feel compromise; talent availability varies by framework |
| Progressive web app (PWA) | ~0.4-0.6x | Content, SaaS admin panels, discovery-heavy apps | Limited iOS PWA feature support; no App Store presence |
| Responsive web app | ~0.5-0.8x | Dashboards, B2B tools, content platforms | Not installable on the home screen without PWA layer |
UK iOS vs Android market share (April 2026)
Before picking a platform strategy, check who your UK users actually are. StatCounter data for the UK shows the split sitting at near parity in early 2026 - Android around 50.1%, iOS around 49.4%. That 50/50 distribution means an iOS-only strategy cuts off roughly half your addressable UK audience. iOS-only still makes sense for narrow premium apps (paid apps, subscription services aimed at higher-ARPU users), but for most UK SME apps aiming at mass-market consumers, workforce, or B2B customers with mixed fleets, both platforms is the right default.
UK OS version distribution (what you actually need to test against)
On iOS, the UK user base is concentrated on the current major release (around 60% of iOS users) plus a long tail on the previous major release (roughly 14% combined) and a small residual on older versions. A minimum support target of current-major plus one-prior usually captures 95%+ of UK iOS users.
Android is more fragmented. The latest Android version typically sits at around 34% of UK Android users, with the previous three major versions each at 8-13% and a long tail of older versions. UK Android support realistically needs to span four or five major releases - a wider test matrix than iOS, and one reason why Android testing costs tend to run 20-40% higher per QA cycle than iOS.
Cross-platform framework comparison (2026 UK market)
If you are going cross-platform (and most UK SMEs should), the three real contenders in 2026 are .NET MAUI, Flutter and React Native. Kotlin Multiplatform and Capacitor/Ionic are in the conversation for specific use cases but have smaller UK hiring pools.
| Framework | UK contractor day rate (ITJobsWatch, April 2026) | Code reuse | Hiring ease in UK | Best fit |
|---|---|---|---|---|
| React Native | ~£515/day | ~85% iOS/Android | Easiest (largest UK JS talent pool) | JS/TS shops; apps with lots of product iteration |
| Flutter | ~£538/day | ~90% iOS/Android/web | Second-easiest (growing UK community) | Consumer apps with bespoke UI; performance-sensitive |
| .NET MAUI | ~£500-£525/day (.NET contractor bands) | ~75% iOS/Android/Windows | Good (.NET community) | UK .NET / Azure shops; Windows + mobile apps |
| Kotlin Multiplatform (KMM/KMP) | ~£500/day (Kotlin contractor) | Shared logic, separate UI | Smaller specialist pool | Teams with strong Kotlin/Android who want to share logic with iOS |
| Capacitor / Ionic | ~£500/day | ~95% (web-wrapped) | Declining UK demand | Content and simple form-entry apps only |
Day rates above are medians from ITJobsWatch for UK contract roles. Seniority spreads 10th-90th percentile typically runs £380-£675/day across these frameworks; senior specialists in London can be materially higher.
A couple of 2026-specific considerations. React Native's New Architecture (Fabric, JSI, TurboModules) became mandatory in 2026 - if you are upgrading an older React Native app, budget for a one-time migration. Flutter's Impeller 2.0 rendering engine improved performance markedly, narrowing the gap with native for most consumer apps. .NET MAUI 8.2 improved ARM performance and Hot Reload, and the UK .NET MAUI community is visibly growing (the first .NET MAUI Day London ran in February 2026).
For most of our .NET-shop clients, we recommend MAUI - the cross-team talent reuse with your backend engineers is worth meaningful time in delivery. For greenfield consumer apps with a bespoke design direction, Flutter is often the strongest pick. For existing JS shops, React Native is the safest step. KMM is worth considering only if your team is Kotlin-heavy already and the shared-logic savings are material; otherwise the thinner specialist pool is a hiring risk.
When cross-platform wins the economics (break-even)
Using the cost multipliers above: if you need both iOS and Android, cross-platform is always the cheaper starting point unless your app specifically needs deep native capability (AR, advanced camera pipelines, certain biometric flows, high-frame-rate games). For most UK SME business apps - field service, customer portals, booking, content, CRM satellites, operational tools - cross-platform is the default right answer.
What the big UK apps actually chose
A quick scan of the decisions behind well-known UK apps is instructive. Monzo chose native mobile development (confirmed in their own community forums) with a Go-based microservices backend that grew from three engineers to around 150 services over time. Revolut runs native mobile apps with a Java/Kotlin microservices backend on Google Cloud, with engineering organised into agile squads of 6-8. Octopus Energy took a hybrid approach - native iOS with SwiftUI, native Android with Jetpack Compose, and Kotlin Multiplatform Mobile (KMM) for shared business logic between the two - sitting on top of their Kraken API-first platform on AWS. Deliveroo runs native iOS in Swift and native Android in Kotlin (confirmed in their own engineering blog and mobile-engineer interview guide) with specialised mobile teams of 15-24 engineers per domain area - consumer, restaurant and rider apps each get their own team - and acquired the Edinburgh studio Cultivate (around 50 engineers, product, UX and data staff) to expand that capability.
The pattern: scale-up fintech and logistics apps with high-performance or security demands tend to go native; mid-market SaaS and workforce apps typically go cross-platform; content and informational apps often start as a PWA or responsive web app. If you are ever benchmarking "could I build an app like Monzo?" - the honest answer is that a public-facing regulated banking app, with the engineering organisation needed to support it, runs into six or seven figures over its first few years, not tens of thousands.
5. UK app developer day rates in April 2026
Every quote you receive is, underneath, a day-rate multiplication. Understanding UK day rates lets you sanity-check any proposal.
UK software developer contract day rates (ITJobsWatch, 16 April 2026)
| Percentile / Region | Day rate (£) |
|---|---|
| 10th percentile | £382 |
| 25th percentile | £438 |
| Median (UK) | £500 |
| Median (London) | £530 |
| Median (UK excluding London) | £510 |
| 75th percentile | £588 |
| 90th percentile | £675 |
Specialist mobile application development contracts show a lower headline median of £360/day (ITJobsWatch), reflecting a different role mix and often less senior contractors than full-stack software developer roles.
Seniority bands (hourly, UK)
- Junior: £25-£40/hr
- Mid-level: £40-£70/hr
- Senior: £70-£150/hr (London specialist)
- Freelance marketplace average (YunoJuno): £63/hr, £505/day
Agency blended rates
UK agencies typically charge a blended day rate of £600-£1,200/day for a mixed team (senior, mid, junior, plus some QA and PM time absorbed). Specialist London agencies charging £100-£150/hr on top-tier engagements imply blended day rates above £1,000. Agency rates include the overhead of running the agency - project management, discovery, QA, delivery, admin - which is why they sit above pure contractor rates. For the right project, it is a price worth paying; for simple, narrow-scope work, contractors or freelancers can be cheaper.
Named UK agency examples give you a calibration point. Apadmi in Manchester quotes in the £90-£135/hr range. Waracle (Dundee and London) sits at £80-£120/hr. Clutch's aggregated listing for London puts typical agency hourly rates at £100-£150/hr with full-feature project budgets of £60k-£150k+. If a UK agency is quoting hourly rates significantly above or below those bands, ask what is driving it.
6. Cost by project stage
A healthy UK SME app budget is split roughly as follows. Exact percentages shift with complexity - regulated projects spend more on discovery, consumer-first apps spend more on design, and testing scales with complexity.
Typical budget split (standard UK SME app)
| Stage | % of budget | Typical weeks | What happens |
|---|---|---|---|
| Discovery | 7-12% | 2-4 weeks | Problem/market validation, scope definition, user research, tech architecture decisions, success metrics |
| Design | 12-18% | 3-6 weeks | Wireframes, prototypes, design system, accessibility review, usability testing |
| Build | 45-55% | 12-24 weeks | Frontend, backend, integrations, sprint-based delivery |
| Test | 10-15% | 2-4 weeks (overlapping) | QA cycles, device testing, performance testing, security review, UAT |
| Launch | 2-5% | 1-2 weeks | App store submission, go-live, early monitoring |
| Post-launch (year 1) | 15-25% of initial build | ongoing | OS updates, bug fixes, small features, analytics, hosting, support |
Skipping or compressing Discovery to save money is a classic false economy - it is the single strongest predictor of budget overrun, per the Deloitte 54% figure. A well-run Discovery usually pays for itself several times over by catching scope issues before they have been built. Our guide to software requirements specifications covers what a proper Discovery output looks like.
What the UK public sector teaches us about overruns
- NHS National Programme for IT (patient record systems) - widely reported to have cost the taxpayer close to £10bn before large portions were abandoned (Guardian, 2013; NAO reports).
- HMRC Making Tax Digital - the NAO's 2023-24 HMRC overview shows programme cost has increased by around 400%, with a projected total of £1.27bn.
- NHS COVID-19 app - around £36m total spend (NAO FOI data) with a four-month delay to launch from the original plan.
The common threads, per NAO analyses: weak early scoping, procurement complexity, and scope creep. UK SMEs can sidestep the worst of it by running a proper fixed-scope Discovery, using small incremental builds, and keeping their contracts clear on change-control.
7. Agency vs offshore vs in-house vs freelance
How you source your development team has a bigger impact on total cost - and total risk - than any other decision beyond platform choice.
Sourcing model comparison
| Model | Typical UK cost | Best for | Main risk |
|---|---|---|---|
| UK agency | £600-£1,200/day blended | End-to-end delivery with minimal internal tech leadership | Highest headline cost; agency overhead baked in |
| UK freelancer / contractor | £300-£800/day | Tightly scoped pieces of work, augmenting an existing team | No integrated QA/design/PM; single-person dependency |
| Offshore team | 40-70% lower headline than UK | Price-sensitive projects with strong internal tech leadership | Timezone, requirements translation, quality variance, IP and GDPR handling |
| Nearshore (EU) team | 20-40% lower than UK | Cost savings with less timezone friction than far-offshore | Post-Brexit contracting complexity; quality varies by country |
| In-house hire | ~£90,000/yr fully-loaded senior; ~£360k/yr for a team of 4 in London | Long-term product with multi-year roadmap | High fixed cost; recruitment lag; churn risk pre-launch |
| UK lead + offshore / nearshore engineers (hybrid) | £70k-£150k/yr lead plus offshore build team | Best cost/quality balance for most UK SMEs | Requires a competent UK technical lead to make it work |
For most UK SMEs commissioning a first app, a UK agency with proven mobile experience is the lowest-risk route. For second or third apps, or for ongoing product evolution post-launch, a hybrid model (UK lead + nearshore or offshore engineers) frequently delivers the best cost/quality balance. Full in-house for a pre-revenue first app rarely makes financial sense - the team costs what a good agency would deliver twice.
Whichever route you take, read our fractional CTO guide if you do not have a full-time technical leader in place - managing any build, in any sourcing model, without one is risky.
Not sure which sourcing model fits your situation? Price isn't the only factor - risk, quality, speed and the shape of your internal team all matter. If you'd like an honest second opinion on the right sourcing model for your specific project, get in touch for a free conversation. We'd rather tell you the right answer than sell you the wrong one.
9. AI features and their cost impact in 2026
AI is the biggest shift in app economics since cross-platform frameworks. The cost pattern is also genuinely different to traditional software: most of the cost is not build, it is operation.
Adding AI to an existing app concept
Adding one or two targeted AI features - a conversational support assistant, a document summariser, a smart search, an AI-assisted onboarding flow - typically adds £15,000-£45,000 to the build and £500-£2,500/month to ongoing running costs for a moderate usage pattern. The three biggest levers on monthly cost are model choice, prompt length and how aggressively you cache.
Real 2026 LLM pricing and worked UK SME examples
Model prices have moved quickly. As of April 2026, published per-million-token pricing for the main providers looks like:
| Model | Input (per 1M tokens, USD) | Output (per 1M tokens, USD) | Good for |
|---|---|---|---|
| OpenAI GPT-5 | $1.25 | $10.00 | General-purpose reasoning, broad SME use |
| Claude Haiku 4.5 (Anthropic) | $1.00 | $5.00 | High-volume, low-complexity tasks (chat, classify) |
| Claude Sonnet 4.6 | $3.00 | $15.00 | Balanced reasoning quality / cost |
| Claude Opus 4.6 | $5.00 | $25.00 | Most demanding reasoning tasks |
| Gemini 2.5 Flash-Lite (Google) | $0.10 | $0.40 | Ultra-cheap background inference |
| Gemini 2.5 Flash | $0.30 | $2.50 | Cost-sensitive production workloads |
| Azure OpenAI (UK South region) | ~$0.605 | ~$1.815 | UK-regulated deployments needing data residency |
Pricing synthesised from provider pricing pages, April 2026. USD/GBP exchange rate 1 USD = 0.7371 GBP (exchangerates.org.uk, 15 April 2026). All providers publish subject-to-change pricing; verify at the point of committing.
What does that mean in real SME money? A worked example: you add a customer support chatbot to your app. 1,000 active users × 10 interactions/month × ~180 tokens per interaction gives about 1.8 million tokens/month. Rough monthly costs before discounts:
- GPT-5 baseline: around £9.40/month
- Claude Haiku 4.5 baseline: around £4.90/month
- Azure OpenAI UK South: around £3-5/month depending on config
Turn on prompt caching (up to 90% off cached input tokens for GPT-5 and Claude) and the batch API (50% off for asynchronous work) and those numbers can drop 40-70%. Double-check the stacking rules with your provider - not every discount composes cleanly.
The UK data-residency picture
Azure OpenAI Service is listed on the UK Digital Marketplace and offers UK data residency in UK South. UK-region pricing is materially lower than US-baseline pricing for the same models, which quietly makes Azure one of the best-value UK SME AI deployments in 2026. OpenAI's direct API, Anthropic's Claude via AWS Bedrock, and Google's Gemini all have UK/EU regional options but check the specific deployment region before committing. If your buyers are UK public sector or regulated (fintech, health, legal), data residency is usually a must-have - and Azure's UK Digital Marketplace listing makes procurement dramatically easier.
Building an AI-native app
Where a reasoning engine is the core of the product (not a feature bolted on), build costs move into the complex band: £80,000-£350,000+ typical UK range. Ongoing operating costs can scale unpredictably with usage. This is the shift most UK SMEs underestimate: unlike traditional software, where running costs are relatively fixed, AI-native apps have genuinely usage-based economics that can surprise you. Our MVP development guide covers concrete 2026 token-cost examples for common AI workloads, and our custom AI solutions guide covers the build-vs-buy decision in more depth.
Cost controls you should insist on before shipping
- Hard token-spend caps per user per day (and per tenant, if multi-tenant)
- Model routing: use a cheap 'mini' or 'flash' model for most requests, reason-capable model only when needed
- Prompt caching (Anthropic and OpenAI both offer up to 90% discount on cached inputs)
- Batch API routes for asynchronous work (50% off across major providers)
- Evaluation harness to catch regressions when you change prompts or models
- For Azure: consider PTUs (Provisioned Throughput Units) once you hit roughly $2-3k/month of steady usage
The top question every UK SME should ask their agency: "What happens to our bill if we suddenly go 10x in usage?" If they cannot answer specifically, keep asking.
The 2026 UK AI regulatory picture
The UK has not enacted a horizontal AI Act like the EU has, but the ICO's guidance on AI and data protection continues to apply and was updated through 2025-2026 with a strong focus on agentic AI and biometrics. The Data (Use and Access) Act 2025 reinforced the 72-hour breach reporting rule and brought PECR breaches into the same window. If your app does automated decision-making with a legal or similarly significant effect on a person, plan a DPIA from the start.
10. Worked example: UK SME field-service app
Let's make the numbers concrete. Here is a realistic worked example for a UK SME commissioning a field-service app in April 2026. Scenario: a 40-person building services company in the Midlands wants a mobile app for engineers to receive jobs, update status, capture photos, collect signatures and sync to their back-office system.
Assumed scope
- iOS + Android coverage via .NET MAUI
- Engineer mobile app (login, job list, job detail, photo capture, signature, status updates, offline mode)
- Back-office admin web app (job dispatch, fleet view, reporting)
- Integration with existing Dynamics 365 Business Central for job data sync
- Azure backend (App Service + SQL, Blob Storage for photos)
- UK GDPR compliance for engineer and customer data
- No AI features in v1
Indicative cost breakdown (UK agency, April 2026)
| Stage | Days | Cost estimate |
|---|---|---|
| Discovery (requirements, architecture, data model) | 10-15 days | £6,000-£10,000 |
| UX/UI design (engineer app + admin portal) | 15-20 days | £9,000-£14,000 |
| .NET MAUI engineer mobile app build | 50-70 days | £30,000-£45,000 |
| Admin web app (Blazor or React) | 20-30 days | £12,000-£20,000 |
| Backend + Dynamics 365 BC integration | 20-30 days | £12,000-£20,000 |
| Testing and QA (device matrix, offline mode) | 15-20 days | £9,000-£14,000 |
| Compliance (GDPR, accessibility, privacy manifests) | 5-8 days | £3,000-£5,500 |
| Launch, App Store submission, go-live support | 3-5 days | £1,800-£3,500 |
| Contingency (10%) | - | £8,000-£13,000 |
| Total initial build | - | £90,000-£145,000 |
Year 1 TCO after launch: add ~£18,000-£30,000 maintenance, £8,000-£15,000 hosting and third-party services, £2,000-£5,000 compliance maintenance. Total Year 1: £120,000-£195,000. This is typical of a mid-complex UK SME app. If the company went offshore, the build cost could drop 30-50%; if it went dual-native instead of MAUI, it would rise 40-60%. The hybrid-sourcing model (UK tech lead + offshore .NET engineers) would land it around £60,000-£90,000 for the build phase.
11. UK funding and tax relief routes for app projects
Many UK SMEs commission apps without realising there is meaningful funding and tax relief available for genuinely innovative work. Five routes worth knowing about in 2026:
Innovate UK Smart Grants
Innovate UK Smart Grants co-fund innovative software and AI work for UK-registered businesses through quarterly assessment rounds. 2026 funding levels run up to £3m for feasibility studies (particularly for frontier AI and ML) and up to £5m for larger collaborative projects enabling adoption of national standards such as the Software Security Code of Practice. SMEs can receive up to 70% of eligible costs for industrial research and 45% for experimental development. Historical success rates hover around 15-20% across rounds. Applications are competitive - your app must be genuinely innovative (not just novel to your business) and you need to make a credible case for commercial impact. UKRI also runs AI-specific opportunities in 2026, including Frontier AI Discovery grants up to £2.5m.
R&D Tax Relief (merged scheme, from April 2024)
The merged R&D Tax Relief scheme gives UK companies a 20% above-the-line Research and Development Expenditure Credit (RDEC) on qualifying R&D spend. After corporation tax this yields a net benefit of around 15% for profit-making companies and around 16.2% for loss-making ones. For loss-making R&D-intensive SMEs - where qualifying R&D is at least 30% of total expenditure - the Enhanced R&D Intensive Support (ERIS) regime offers a more generous 14.5% cash credit, worth up to 27p per £1 of qualifying spend. RDEC and ERIS are mutually exclusive on the same expenditure. Most app development work with a genuine technical uncertainty component qualifies. Speak to an R&D tax specialist at Discovery, not at year-end - the documentation you need must be in place from the start.
SEIS and EIS
The Seed Enterprise Investment Scheme (SEIS) and Enterprise Investment Scheme (EIS) are the standard UK tax reliefs for angel investors backing early-stage companies. SEIS offers 50% income tax relief on up to £200,000 of investment per company; EIS offers 30% on up to £1m per individual. Not direct funding for the build, but they make it dramatically easier to raise from UK angels - which is how many UK SME app projects actually get funded in 2026.
Knowledge Transfer Partnerships (KTP)
KTPs cover roughly 67% of project cost for SMEs when your app has a genuine research question that a university research partner can work on. Typical KTP projects cost around £8,500/month to run, with a total package value of £80,000-£100,000 per year, and typically run 12-36 months. The 2026-2027 round opened in early February and closed in early April, with notifications in June. Best fit for genuinely novel AI, data science or HCI questions - not for a standard CRUD app dressed up.
Start Up Loans
The British Business Bank's Start Up Loans scheme provides loans up to £25,000 per director at a fixed interest rate of 7.5% from 6 April 2026, with eligibility now extended to businesses that have been trading for up to 60 months (previously 36). A modest but useful option for founders who want to fund Discovery or a thin MVP without giving up equity.
Regional UK funding (nations and combined authorities)
Outside the national schemes, regional and devolved funding is often more accessible for smaller projects:
- Scotland: Scottish Enterprise grants up to 45% of eligible costs for small enterprises (35% for medium), plus sector-specific R&D grants
- Wales: Welsh Government £2.1m AI adoption funding (2026), distributed via Business Wales
- Northern Ireland: Invest NI grants and Techstart Ventures for tech-sector funding
- Greater Manchester: multi-hundred-million-pound innovation plan supporting AI and digital skills
- Other combined authorities (West Midlands, Liverpool City Region, West Yorkshire) run their own tech and digital funds - worth a short call to your Local Growth Hub
UK Digital Marketplace / G-Cloud
If you are building for the UK public sector, being on the G-Cloud framework is effectively required to sell without going through full procurement. Not direct funding, but the revenue path it unlocks is material. Azure OpenAI and many other cloud services are already listed, which dramatically shortens the procurement conversation.
12. A decision framework for budgeting a UK app in 2026
Ranges and drivers are useful, but decision-makers still need a framework for working out what their specific number is. Here is the one we use with UK SME clients in Discovery.
Step 1: Start from the business outcome, not the feature list
What has to be true in 12 months for this app to have been worth the money? Revenue uplift? Operational cost reduction? Retention? Compliance? The answer shapes the scope, the target band and the measure of success. If you cannot answer this in one sentence before reading on, go no further - that is your first sprint of Discovery work, not the build team's.
Step 2: Place yourself on the cost band
Use the bands in Section 2. Most first-time SMEs overscope by one tier: they think they need Standard, they actually need Simple. A focused MVP at the Simple tier gives you validated learning in 8-12 weeks at £8k-£30k - use our MVP development guide as a scoping discipline.
Step 3: Pick the cheapest platform approach that meets the requirement
Go through the Section 4 table with the business outcome in mind. Do you actually need a mobile app? Could a PWA do the job? If you need mobile, can a single platform wait 6 months? If you need both, cross-platform is almost always the answer.
Step 4: Choose the sourcing model that matches your risk tolerance and internal capacity
No internal tech leader + tight timeline + limited patience for management overhead = UK agency. Good tech leader + cost-sensitive = hybrid (UK lead + offshore build). Long-term product with a clear multi-year roadmap = eventually, in-house. Tightly scoped augmentation work = freelance. Do not pick the cheapest without considering the risk mix - we have seen more money lost to failed cheap builds than to expensive ones that worked.
Step 5: Add 10-15% contingency and a clear post-launch budget
Never sign a fixed-price quote without a 10% (stable scope) to 15% (exploratory scope) contingency. Never commission a build without a budgeted Year 1 maintenance line at 15-25% of the build. Launching without these is how projects end up at 54%-over-budget - per Deloitte 2025 - and how launched apps die of neglect within 18 months.
Step 6: Match the contract type to the scope certainty
- Stable, well-understood scope → fixed price (with 10-15% contingency, expect a 15-30% risk premium from the agency)
- Exploratory, scope still shaping up → time and materials, with a sprint cadence and weekly burndown
- Mixed → hybrid: fixed Discovery, T&M Build (the most common model in UK SME work)
Not sure an app is even the right next step? Sometimes the honest answer is "a responsive website", "a SaaS tool with some configuration" or "a better internal process" - not "build an app". We'd rather have a 30-minute conversation and talk you out of a build than take on one that shouldn't happen. Get in touch for a no-obligation conversation about whether your idea warrants an app at all.
13. Frequently asked questions
Sources
- ITJobsWatch - UK contract day rates for software developers (retrieved 16 April 2026) - itjobswatch.co.uk/contracts/uk/software%20developer.do
- ITJobsWatch - UK contract rates for mobile application development - itjobswatch.co.uk/contracts/uk/mobile%20application%20development.do
- LuminaryBrands - UK Mobile Application Development Statistics (February 2026) - UK app development industry £28.3bn 2024-25
- DataReportal - Digital 2026 United Kingdom - UK smartphone penetration and mobile connections
- Foundry-5 (2026) citing Deloitte (2025) - 54% of UK software builds run 40%+ over budget
- Innovate UK Smart Grants - grant funding for UK SMEs (ukri.org/councils/innovate-uk)
- HMRC - merged R&D Tax Relief scheme from April 2024, Enhanced R&D Intensive Support (ERIS) for loss-making R&D-intensive SMEs
- Information Commissioner's Office (ICO) - UK GDPR guidance and DPIA templates
- Apple Developer - App Privacy Manifest (required since 2024)
- Google Play - Data Safety section declarations
- YunoJuno - UK freelance developer marketplace rate data (2025)
- SmartWorking, thecodev, Mantar Tech, DITSek, ScaleUpAlly, RiseUpLabs, Code & Pepper, DevStree, BayTechConsulting, EmvigoTech - 2025/2026 UK agency cost band data corroborating the figures in this guide
- Clutch - average UK app project budget figure (February 2026)
- Microsoft - .NET MAUI framework documentation; Azure OpenAI Service - UK Digital Marketplace listing and regional availability
- Flutter (Google) and React Native (Meta) - framework documentation on code-reuse percentages
