The App Services Market in 2026

Start with the distinction that decides everything else in your plan. There is a market for apps, and there is a market for the people who build apps. You are entering the second one. The mobile app development company services market was put at $196.4 billion in 2025, forecast to reach $512.8 billion by 2034 on an 11.3% compound rate (Dataintelo, 2025). Broader definitions that fold in platform licences and in-house build spend run higher: one estimate puts the whole mobile app development market at $145.05 billion in 2025 heading to $776.9 billion by 2035 at 18.25% (SNS Insider, 2025), while another sizes it at $302.1 billion in 2025 growing to $844.5 billion by 2034 (Straits Research, 2025). Those three numbers disagree because they count different things. Say in your plan which definition you are using and why, because a lender who spots an unexplained $150 billion swing stops reading.

Geography matters more here than in almost any other service business, because your competition is priced in another currency. North America held 37.4% of the market in 2025 and Asia Pacific 28.7%, worth $86.70 billion and growing at a 15.9% clip, faster than the global average (SNS Insider, 2025). Read that alongside the rate card and the strategic problem becomes obvious. Clutch's 2026 pricing data puts agency rates at $100 to $150 an hour in the US, Canada and Australia, $50 to $99 in Eastern Europe and $25 to $49 in India (Clutch, 2026). A buyer can get working code at a third of your price. Your plan has to answer, in plain language, why anyone pays the premium.

The answers that hold up are specific. Regulated-sector fluency is one: a studio that has shipped under HIPAA business-associate terms or FCA-adjacent scrutiny carries evidence an offshore bid cannot match on price. Time-zone overlap with the client's product team is another, which is why nearshore pods in Kraków, Lisbon and Monterrey have taken share from pure offshore. Sector depth is a third: an agency that has built fourteen field-service apps knows the offline-sync edge cases before the kickoff call, which compresses a twelve-week build into nine. Generalists compete on price and lose. Specialists compete on risk and win.

Where the demand actually comes from

Three buyer types fund most studio revenue and they behave nothing alike. Venture-backed startups buy speed: an MVP before the next board meeting, fast to sign and fast to churn once they hire in-house. Mid-market companies with no mobile team buy certainty: longer cycles, procurement and security questionnaires, but contracts that renew. Enterprises buy capacity through a preferred-supplier list that takes nine to eighteen months to join and then pays for years. A plan that does not say which of the three it serves ends up with a sales process that fits none of them.

The fourth source, and the one that quietly rescues a lot of studios, is maintenance. Apple and Google ship a major OS release every autumn, and every app in your portfolio needs regression testing against it whether the client budgeted for that or not. Forty live apps under support is a revenue floor you did not have to sell for. Put it in the forecast as a named line.

Questions Founders Ask First

These four come up on nearly every discovery call we run with studio founders, and the answers belong in the plan rather than in your head.

How much do app development agencies charge per hour?

The headline bands are $100 to $150 in the US, Canada and Australia, $50 to $99 across Eastern Europe and $25 to $49 in India (Clutch, 2026). What those bands hide is role mix. The number you quote a client is a blended rate across the people actually on the project, so a team of one senior engineer, one mid-level engineer, a part-time designer and a fractional delivery lead produces a blended rate well below the senior engineer's standalone rate. Price the blend, forecast the blend, and keep a separate internal rate card per role so you can see which roles are being given away.

How do app development companies make money?

Four ways, in descending order of how common they are. Billable delivery is the core: hours sold at a rate above fully-loaded cost. Support and maintenance retainers come second, typically 15% to 20% of the original build value per year, and they carry the highest margin because the work is predictable. Third is resold infrastructure and licences, where you pass through cloud, analytics or subscription tooling with a management uplift. Fourth, and the one most plans overstate, is product revenue: building something of your own, or taking equity instead of cash. Equity deals are real but they are not a funding plan, and no lender will underwrite against them.

Should I build apps for clients or build my own app?

Services generate cash from month one and cap your upside. Product generates nothing for a year and uncaps it. Studios that try both usually starve the product, because the moment a client escalation lands the product engineer gets pulled onto billable work. If you want both, ring-fence it honestly: a fixed percentage of capacity, a named person, and a written rule that the ring-fence is not raidable. Then show the forecast with that capacity already removed from billable hours.

What financial projections should an app development company business plan include?

For a studio, a revenue line is not a projection. The model has to be built bottom-up from headcount: billable heads by month, available hours per head, target utilisation, blended bill rate, and fully-loaded cost per head, so revenue falls out of those assumptions rather than being typed in. Lenders also want a pipeline-coverage ratio against the next two quarters of capacity, a debtor-days assumption (45 to 60 is realistic for mid-market clients), and a cash view of the gap between monthly payroll and 30-day payment terms. That working-capital gap is what kills profitable studios.

What It Costs to Open the Studio

Published ranges for launching an app development company sit between $10,000 and $80,000, with a mid case around $30,000 (HowMuchToStart, 2026). We set the practical floor higher, at $15,000 in the US or £12,000 in the UK, with a realistic ceiling of $80,000 / £60,000 for a studio opening with three or four people on payroll. Most published lists leave out the test-device matrix and treat working capital as optional. Payroll runs monthly; client invoices land 30 to 60 days after the work.

Cost breakdown

• Company formation, contract pack and trademark: $1,200–$4,500 (£600–£3,000). The contract pack matters more than the formation. You need a master services agreement, a statement-of-work template, and an IP assignment that runs from every contractor through you to the client.
• Apple Developer Program: $99 per year, and Google Play developer account: $25 one-off. Trivial amounts that nonetheless gate your ability to ship anything at all.
• Test device matrix: $1,800–$4,500 (£1,500–£3,800). Two iPhones across two iOS generations, two Android handsets at different price points, one tablet. Simulators do not catch camera, Bluetooth or low-memory defects.
• Development workstations: $4,000–$12,000 (£3,400–£10,000). iOS builds require Apple silicon, so there is no cheap route around this line.
• Tooling, CI/CD and cloud: $3,600–$12,000 a year (£3,000–£10,000). Source control, build pipeline, design tool, issue tracker, crash reporting, staging infrastructure.
• Professional indemnity and cyber liability insurance: $1,800–$6,000 a year (£900–£3,500). Mid-market and enterprise clients will not sign without evidence of cover, usually at £2M to £5M.
• Two self-funded reference builds: $4,000–$18,000 (£3,500–£15,000). Nobody buys a studio with an empty portfolio. Budget the cost of building something shippable before your first paying client.
• Business development: $2,500–$15,000 (£2,000–£12,000). A Clutch profile with verified reviews, outbound tooling, and one sector conference.
• Working capital, three months of payroll: $18,000–$90,000 (£15,000–£70,000). The single largest line, and the one most first plans leave out entirely.
• ICO data protection fee (UK only): £40 for micro organisations, £60 for SMEs, £2,900 for large (ICO).

Funding routes that fit a services business

In the US, the SBA 7(a) programme is the workhorse. The SBA approved 70,242 7(a) loans worth $31.1 billion in FY2024, and within NAICS 541511, Custom Computer Programming Services, the average loan size is around $242,000 across roughly 10,000 loans totalling $2.4 billion (PeerSense SBA industry data). The size standard for 541511 is $34 million in average annual receipts, so there is no realistic scenario in which a new studio is too large to qualify. Because you have no collateral, the plan carries the weight: lenders will test your capacity model, your pipeline, and your personal guarantee.

In the UK, Start Up Loans lend up to £25,000 per founder at 6% fixed with twelve months of free mentoring, and up to three co-founders of the same company can each apply, which takes the combined ceiling to £75,000. Beyond that, a services studio's most useful facility is invoice finance against signed client contracts, because it funds the exact gap the business has. Then there is R&D relief. Under the merged scheme that applies to accounting periods beginning on or after 1 April 2024, most SMEs claim a 20% above-the-line credit worth roughly 15% to 16.2% after Corporation Tax, and the Enhanced R&D Intensive Support threshold has dropped from 40% to 30% of total expenditure, giving loss-making research-heavy SMEs relief up to 27% (ForrestBrown). Two cautions worth putting in writing: routine client app builds are usually not qualifying R&D, and relief is a tax outcome, not working capital. Do not model it as cash in month three.

Canada's SR&ED programme, Australia's R&D Tax Incentive and India's zero-rated export treatment shape the same picture elsewhere. The bespoke plan service builds the lender pack for whichever jurisdiction you borrow in.

Hardware, Accounts & Tooling Checklist

A studio's capital expenditure schedule is short, which is why leaving items off it is noticeable. This is the list we use when building the asset and opex schedules for a software services client.

Hardware, per engineer

• Apple silicon laptop, 32GB where possible: $2,000–$3,500 (£1,700–£3,000). Non-negotiable for iOS builds. 16GB works until you run a simulator, a local backend and a design tool at once.
• External display and peripherals: $350–$700 (£300–£600).
• Mobile device plan for on-device testing: $25–$50 a month per shared SIM.

Shared test device matrix, per studio

• Current-generation iPhone plus one two-generation-old handset: $900–$1,800. Clients still have users on older hardware, and performance regressions only show there.
• One flagship Android plus one mid-tier Android: $500–$1,200. Fragmentation is a real cost line, not a talking point.
• One iPad and one Android tablet: $500–$1,000, if you take tablet work at all.
• Accessibility test kit: screen reader, contrast tooling and switch control. Costs little and is increasingly a procurement requirement.

Accounts and platform fees

• Apple Developer Program: $99 a year. Organisation enrolment requires a D-U-N-S number, so start it before you need it.
• Google Play developer account: $25 one-off, with identity verification under the Android developer verification programme announced in August 2025.
• Cloud and backend: $150–$900 a month across staging and production environments before any client-specific infrastructure.

One procurement decision shapes the whole schedule: whether client infrastructure sits in your cloud account or theirs. Billing it through yours is simpler to set up and becomes a liability the moment a client leaves owing money. Our recommendation is that production infrastructure lives in the client's account from day one, with you holding delegated access.

Rate Card, Utilisation & Margin Maths

Custom software firms should run at roughly a 30% to 50% gross margin, and the healthy agency band is 30% to 60%. Anything below 30% almost always means untracked time or underpriced fixed bids, and hidden time sinks such as app-store review cycles, third-party API defects and repeated QA passes can cut realised margin by 20% to 40% against what the proposal assumed (8base, agency margin analysis). Those two figures contain the whole operational challenge of running a studio.

Project sizes you will actually quote

Buyers ask for an app, not for hours. Current benchmarks for a first release put a simple MVP at $8,000 to $25,000 over four to six weeks, a mid-complexity build at $25,000 to $55,000 over six to ten weeks, and a complex app with real-time features, machine learning or regulated data at $55,000 to $150,000 and above across ten to eighteen weeks (Helpware, 2026). Translate each tier into hours at your blended rate and you have a rate card a client can read and a lender can test.

The worked example that belongs in your plan

Here is the calculation, run twice, that decides whether a studio survives. Take six billable engineers. Each is contracted for roughly 1,800 available hours a year after holiday and public holidays. At a 75% utilisation rate and a $95 blended bill rate, that is 1,350 billable hours each, so 6 × 1,350 × $95 = $769,500 of service revenue. Now cost it. Using the US median software developer wage of $133,080 plus about 22% employer burden, each engineer costs roughly $162,400 fully loaded, so six of them cost $974,400. Delivery payroll alone exceeds revenue. The studio is insolvent before a single salesperson or office cost is added.

Raise the rate and the utilisation and it is still not enough. At 82% utilisation and a $125 blended rate, the same six engineers bill 1,476 hours each, so 6 × 1,476 × $125 = $1,107,000 against that $974,400 of payroll. That is a 12% gross margin, less than half the floor for a viable agency. Push the rate to $150 and revenue reaches $1,328,400 for a 27% margin, which is still under the 30% line. This is the calculation almost no studio plan runs, and it is why so many of them fail in year two: at the US median wage, an all-senior team cannot be billed profitably at mid-market rates.

The fix is the seniority pyramid, and it is the single most valuable thing to model properly. Staff the same six seats as two senior engineers at $162,400 fully loaded, two mid-level at roughly $120,000 and two juniors near the lower decile at about $97,000, and delivery payroll falls to $758,800 with a blended cost of $126,467 per head. Hold 82% utilisation and a $150 blended rate and you bill $1,328,400 against $758,800, which is a 43% gross margin, squarely inside the 30% to 50% band. Nothing about the revenue changed. The mix did. Build your model so seniority mix, utilisation and blended rate are three visible inputs on one tab, then show a lender the sensitivity grid. That one table does more for credibility than ten pages of narrative.

Two caveats keep the maths honest. The pyramid only works if juniors are genuinely productive, so mentoring and code review are real non-billable costs that belong in the utilisation assumption. That is also the arithmetic behind nearshore pods: at Eastern European cost levels the same structure produces a wider margin, which is why your pricing has to be defensible against the $50 to $99 band. And 100% utilisation does not exist. A senior engineer who bills 85% of available hours is performing very well; if your spreadsheet says 95%, a lender who has funded a services business will find it in the first ten minutes.

Revenue lines beyond the build

• Support and maintenance retainers: 15% to 20% of original build value per year. Highest-margin line in the business and the one that smooths cash flow across the gaps between projects.
• Annual OS-release readiness work: every live app needs regression testing against each new iOS and Android major version. Sell it as a scheduled package rather than firefighting it for free.
• Discovery and technical due diligence: two to four week paid engagements at premium rates. They also qualify the client and de-risk the fixed bid that follows.
• Dedicated-team retainers: a monthly fee for named capacity. Lower rate per hour, far higher revenue certainty, and the line that makes a studio fundable.

Fixed Bid vs. T&M vs. Dedicated Team

Studios in the same market behave like three different businesses depending on how they contract. Name the model you lead with, because it sets your cash cycle, your margin volatility and the kind of client you can serve.

Most studios start on fixed bid because it is what first-time clients sign, then move towards retainers because that is what makes the business financeable. Put that migration in the plan as a target mix: 70% fixed bid and 30% retainer in year one, inverting by year three. Retainer-led agencies keep clients roughly three times as long as project-based ones (8base).

SBA Lending & Developer Wage Benchmarks

SBA 7(a) lending into software services

Within NAICS 541511, Custom Computer Programming Services, the average SBA 7(a) loan is approximately $242,000, drawn from around 10,000 loans worth a combined $2.4 billion. Across the whole programme, 70,242 loans worth $31.1 billion were approved in FY2024 (PeerSense SBA industry data). The adjacent code, 541512 Computer Systems Design Services, is worth knowing about too: many studios classify there instead, and the choice affects which lender comparables your application is benchmarked against (PeerSense, NAICS 541512). Pick the code deliberately and state it on the first page of the plan.

Three implications for the funding request. $242,000 is a realistic ask, not $2 million, and an amount far outside the sector norm invites scrutiny you do not want. The size standard for 541511 is $34 million in average annual receipts over the prior three fiscal years, so eligibility is never the obstacle. And because a studio has no equipment or property to secure the loan against, expect a personal guarantee and expect the lender to underwrite the pipeline: a signed contract schedule, a debtor-ageing assumption and the capacity model are what move the decision.

Wage data: your single largest cost line

US software developers under SOC code 15-1252 had a median annual wage of $133,080 in May 2024, with the bottom 10% below $79,850 and the top 10% above $211,450. Around 1.69 million people held those jobs, and employment is projected to grow 10% between 2025 and 2035, adding roughly 174,700 positions (U.S. Bureau of Labor Statistics).

Three things follow. Add 20% to 25% on top of base salary for employer taxes, benefits and software licences to get fully-loaded cost per head, which is the figure the margin calculation needs. Use the distribution rather than the median: a two-year engineer near the 25th percentile affects your margin very differently from a staff engineer near the top decile. And treat the projected 10% employment growth as a cost warning as much as a demand signal, so build an annual salary uplift assumption into the model and say what it is.

Legal, Tax & App Store Compliance

There is no professional licence for writing software, which leads founders to assume compliance is a non-issue. The opposite is true: the obligations arrive through the app stores, through data protection law, and through whatever sector your clients operate in.

United States

• No state professional licence is required to develop apps in any US state. What you do need is a general city or county business licence, an EIN, and in many states a sales-tax permit if you resell packaged software or subscriptions.
• Entity formation and IP assignment. Filing fees run $50 to $800 by state. The real risk is not formation: without a written work-for-hire and IP assignment clause in every contractor agreement, your client does not cleanly own the delivered code, which surfaces in their next funding round and lands back on you.
• Apple Developer Program, $99 a year. Organisation enrolment requires a D-U-N-S number and legal entity verification, which can take up to a week. Apple's commission on client app revenue is 15% below $1M a year and 30% above, which matters when you are advising on a client's pricing model.
• Google Play developer account, $25 one-off. Under the Android developer verification programme announced in August 2025, developers must supply legal name, address, email, phone, government photo identification and a Google payments profile before apps can be installed on certified Android devices.
• Sector overlays you inherit from clients. Health apps pull you into HIPAA business-associate agreements, apps aimed at under-13s bring COPPA obligations, and payment features trigger state money-transmitter analysis for the client. Write a sector policy before a salesperson signs something you cannot deliver.

United Kingdom

• ICO data protection fee, annual and mandatory for controllers: £40 for micro organisations with turnover up to £632,000 and no more than 10 staff, £60 for SMEs up to £26 million turnover and 250 staff, £2,900 for large organisations. Failure to pay where required can attract a fine of up to £4,350 (ICO data protection fee). A studio is a processor for client data but a controller for its own staff and prospect records, so the fee applies regardless.
• UK GDPR Article 28 processor agreements. Any client handing you production data needs a written processor agreement naming your sub-processors (your cloud and analytics vendors), covering international transfers and setting a breach-notification window. Have your own version ready; negotiating theirs from scratch costs weeks.
• IR35 and the off-payroll working rules. If you staff projects with limited-company contractors, you carry status-determination risk and potential PAYE and National Insurance exposure. The practical control is a documented status determination statement per contractor plus genuine project-outcome contracting rather than rolling day-rate substitution.
• Companies House and Corporation Tax. £50 to incorporate, 24 hours online. Relevant because the merged R&D claim is filed through the CT600 and first-time claimants must submit a claim notification form within the statutory window, or the claim is lost.

European Union

The requirement that surprises studios most is Digital Services Act trader status. Developers must declare and verify trader status in App Store Connect and Google Play Console before any new app or update can be distributed in the EU, and the declared business name, address, phone number and email must be correct and verified on both platforms (Apple Developer, App Store Connect help). The trap for an agency is publishing a client's app under your own developer account as a convenience: you become the trader of record, your details appear on the listing, and the client is stranded if the relationship ends. Publish from the client's own account with your team added as users, and write that into the statement of work.

India

For studios delivering from India, tax treatment is the whole story. GST registration applies at ₹20 lakh of aggregate turnover (₹10 lakh in Manipur, Mizoram, Nagaland and Tripura) but is effectively mandatory for exporters, since export counts as an inter-state supply. App development falls under SAC 998314 and IT consulting under 998313, with domestic IT services at 18% GST. Export of services is zero-rated under section 2(6) of the IGST Act when all five conditions are met, and a Letter of Undertaking (Form GST RFD-11) must be filed annually to export without paying IGST upfront. SOFTEX filing is due within 30 days of the invoice date; STPI registration is optional (RegisterKaro, GST for software and IT services). A studio that forgets the annual LUT renewal spends the following quarter funding IGST it will eventually reclaim, which is a working-capital event worth avoiding.

Six Ways Studio Plans Fall Over

We review a lot of software-services plans. The same six faults appear, and each is visible to a lender inside five minutes.

1. Writing a plan for an app instead of for an agency

The most common fault by a distance, and it comes from reading guides written for app founders. The plan arrives full of daily active users, retention curves and in-app purchase conversion. None of that is your business. You sell engineering capacity, so the metrics are billable headcount, utilisation, blended bill rate, gross margin per project, pipeline coverage and debtor days. If a reader cannot find those six numbers, they cannot underwrite the plan.

2. Quoting fixed bids off a feature list

A client sends a list of screens, you send a number, and the number was wrong before you pressed send. Every serious studio sells a paid discovery phase first, two to four weeks, producing the technical approach and risk register before any fixed price. Without it you absorb every integration surprise at zero margin, and since hidden time costs erode realised margin by 20% to 40%, quoting blind is the fastest route below the 30% floor.

3. Modelling utilisation nobody can hit

Spreadsheets with 2,080 billable hours per engineer are a tell. Pre-sales estimating, internal tooling, recruitment, mentoring, holiday and bench time between projects are all real and none of it is billable. The honest band for a senior engineer is 75% to 85% of available hours. State the assumption and what the non-billable time buys, and the rest of the model becomes believable.

4. Letting one client become the business

Concentration risk ends studios that were otherwise doing fine. Once an account passes about 40% of revenue you have no negotiating room at renewal and no ability to walk away from scope abuse. Put a concentration cap in the plan, say what you will do when you approach it, and show the revenue split by client across the forecast period.

5. Breaking the IP chain

Your client's investors will eventually ask whether the client owns its own code. If you used a contractor with no written assignment the honest answer is no, and the problem becomes yours. The chain has to run from every contributor to your company and from your company to the client, in writing, with open-source obligations disclosed. Keep a dependency manifest per project.

6. Publishing client apps from your own store account

It is quicker at the start and creates a hostage situation later. Your details become the declared trader under the EU Digital Services Act, your developer account carries the enforcement risk, and the client cannot leave without a store transfer process.

The Stack a Lender Expects to See

An operations plan for a studio is mostly a description of its toolchain, because the toolchain is the production line. Naming real tools rather than writing "modern CI/CD" is what separates a plan that was thought about from one that was generated.

• Build and delivery: Xcode Cloud or Bitrise for pipelines, with Fastlane handling signing, screenshots and store uploads. Automating releases removes the single most common cause of a slipped client deadline.
• Cross-platform decision: React Native, Flutter or Expo versus fully native Swift and Kotlin. Pick a default and say why. A studio that takes every project in whatever stack the client names cannot build reusable internal libraries, and reuse is where your margin advantage over a cheaper competitor comes from.
• Design and handoff: Figma, with a component library you own. Reusing your own design system across clients is a legitimate, disclosable efficiency and it shortens every project after the first.
• Project and capacity tracking: Linear or Jira for the work, Harvest for time, Forecast or an equivalent for capacity planning. You cannot report utilisation you do not record, and utilisation is the number the whole financial model turns on.
• Quality and observability: Sentry or Firebase Crashlytics for crash reporting, plus automated UI tests on the device matrix. Clients who see a crash-free-sessions figure in your monthly report renew.
• Commercial tooling: RevenueCat if you implement subscriptions for clients, since it removes weeks of receipt-validation work. A verified Clutch profile for inbound credibility, since that is where mid-market buyers shortlist.

Treat the internal component library as an asset in the plan. Each project that feeds it reduces the delivery cost of the next, and that compounding is the honest answer to how you compete with a $35-an-hour rival without working for free.

Sample Business Plan Preview

Here is an extract from an app development studio plan written by our team, so you can see the level of specificity we work at:

What's in the Template

Every Avvale business plan template includes these sections, pre-structured for your industry:

• Executive Summary: Your studio at a glance: positioning, sector focus, capacity, and the funding ask
• Company Overview: Legal structure, founder equity, delivery locations, and the origin story behind your niche
• Industry Analysis: Market size by definition used, regional rate pressure, and the demand drivers you are betting on
• Customer Analysis: Which of the three buyer types you serve, their procurement process, and typical contract value
• Competitor Analysis: Local and offshore competitive mapping plus the risk-based reason buyers pay your rate
• Marketing Plan: Referral engine, sector content, directory presence, and outbound to named account lists
• Operations Plan: Delivery methodology, toolchain, QA and release process, and the capacity model by month
• Management Team: Founder engineering credentials, delivery leadership, and the first three planned hires

The optional Financial Forecast add-on (included in our $300/£250 and $1,000/£800 packages) provides a 5-year Excel model with income statement, cash flow, balance sheet and break-even analysis. For a studio we build it headcount-first, so utilisation, seniority mix and blended rate are inputs you can flex rather than hard-coded revenue.

Building something adjacent? We publish matching templates for a software development company, a SaaS business, an IT consulting firm and a web development company. You can also browse the full free template library or read how we work in our client case studies.

Frequently Asked Questions