Choosing a mobile app agency: criteria, questions and pitfalls
10 min read
Key takeaways
- No-code is unbeatable for validating an idea in a few weeks, but not for carrying a product that has to last five years.
- A senior freelancer costs less than an agency per day, but the real risk is depending on a single person.
- An agency is justified as soon as the project combines design, mobile, back-end and store publication with a date to hit.
- Compare the total cost over two to three years, including maintenance and any takeover, not the initial quote.
- Whichever option you choose, insist that the code, the store accounts and the hosting access are in your name.
You have an app idea, a budget, and three options on the table: a freelance developer recommended by a friend, a no-code tool everyone is talking about, or an agency. All three can deliver something that works. They don't expose you to the same risks at all.
We are an agency, so we won't pretend to be neutral. But we regularly turn down projects for which a freelancer or a no-code tool would objectively be a better choice, and every year we take over apps that should have started differently. This guide gives you the criteria to decide, backed by figures. If you'd like to see how we work first, the mobile app development page details our method.
The three options in one sentence
- No-code (Bubble, FlutterFlow, Adalo, Glide, etc.): you assemble the app yourself or with a specialist, in a visual tool that hosts the result.
- Freelancer: one person, sometimes two, writes real code, and you manage the project.
- Agency: a multidisciplinary team (project manager, design, mobile and back-end development, testing) takes on the project end to end, with a contractual commitment.
The real difference isn't code quality. An excellent freelancer often writes code as clean as an agency's. The difference is who carries the risk: of dependence, coordination, delay and continuity.
The comparison
The comparison below applies to a mobile app of medium complexity: user accounts, five to ten main screens, a simple back office, notifications, store publication.
| Criterion | No-code | Freelancer | Agency |
|---|---|---|---|
| Cost of a first version | Lowest at the start | Varies with the person | On quote, at low prices (single-platform MVP or full iOS + Android) |
| Lead time for a first version | 2 to 8 weeks | 2 to 5 months | 4 to 8 weeks (MVP), 3 to 4 months (full) |
| Recurring cost | Platform subscription, which rises with usage | Time spent, depending on availability | Monthly maintenance contract, on quote |
| Skills covered | Those of the tool | Those of the person | Design, mobile, back-end, QA, publication |
| Code ownership | No, or partial | Yes, if the contract provides for it | Yes, full assignment |
| Main risk | Technical ceiling and dependence on the vendor | Unavailability or departure of the person | Higher initial cost |
Two remarks on this table. First, freelance figures vary enormously: a junior developer at a low rate and a senior who has published thirty apps are not in the same category. Second, the low no-code cost assumes you build it yourself; entrusting the build to a no-code specialist agency quickly brings the bill close to conventional development.
2 to 8 wksFor a usable first no-code version
1 personThe single point of failure of a freelance project
2 to 3 yearsThe horizon over which to compare the real cost of the three options
No-code: perfect for learning fast, fragile for the long run
No-code has matured. For some projects, it's the best choice by far.
When it's the right option
- You're testing a hypothesis. You don't yet know whether anyone will pay for this service. Investing in custom code before you know is a pointless gamble.
- It's an internal tool. An app for twenty employees, handling forms, lists and a few approvals, doesn't need an architecture designed for a hundred thousand users.
- The product is essentially content and forms. Directory, catalogue, simple appointment booking, request tracking.
Where it gets stuck
The limits aren't visible at launch. They arrive when the product starts to work, which is the worst moment.
Performance. Beyond a few thousand active users or large lists, load times degrade, and your room to optimise is whatever the vendor leaves you.
Native features. Bluetooth, robust offline mode, background geolocation, in-app payment, widgets, sensors: some are possible through extensions, others not at all, and quality varies a lot.
Business logic. Dynamic pricing, complex allocation rules, regulated calculations: in a visual tool, these rules become tangles of blocks that nobody dares touch after a year.
Dependence. The vendor hosts your product, sets its prices and can change its terms. If the subscription doubles or a feature disappears, you have no quick plan B.
Above all, remember this: you don't migrate from no-code, you rewrite. The day you move to code, you recover the data, the validated journeys and what you've learned, not the app itself. That's not a problem if you planned for it from the start. It is one if you discover this reality with a growing product and users waiting.
The freelancer: excellent value for money, with conditions
A good senior freelancer is often the best deal on the market. No overheads, a single point of contact who codes themselves, great responsiveness.
When it's the right option
- The scope fits in one head. A single-platform app, a simple back end or an existing back end to plug into.
- You know how to run a project. You (or someone on your team) will write the specifications, approve the screens, test each delivery and arbitrate priorities.
- You already have a technical team. The freelancer reinforces an existing team that can review their code and take over.
- The design is already done, or very simple.
The risks to face squarely
The bus factor. If your freelancer falls ill, takes a full-time contract or goes on parental leave for three months, your project stops. This isn't a criticism of the person: it's a mathematical property of a one-person team.
Coordination. A full mobile app needs a product designer, a mobile developer, a back-end developer and someone to test. A profile that does everything very well is rare. In practice, you end up with two or three freelancers, and you become the project manager, with the time that implies and responsibility for the gaps between each person's scope.
Continuity. An app is never finished. Annual iOS and Android updates, new store requirements and security flaws in dependencies all demand regular maintenance. Two years on, is your freelancer still available? Our article on the cost of maintaining an app details what that involves.
The non-negotiable rule, whichever option you choose: the Git repository, the Apple Developer and Google Play accounts, the domain name and the hosting must be in your name from the first week. We regularly take over projects where the only store access belonged to a provider who had become unreachable. Recovering an app under those conditions costs weeks, sometimes a complete republication.
How to choose a freelancer without getting it wrong
- Download their apps. Not screenshots: published apps that you can install and use.
- Call two former clients. Ask them what happened when there was a problem, not whether everything went well.
- Insist on a Git repository in your name and regular deliveries, at least every two weeks, that you can install on your phone.
- Plan for a replacement. Ask who can take over the code in case of prolonged absence, and require minimal documentation.
- Split payment into milestones that are delivered and tested, never a large deposit before the first line of code.
The agency: more expensive at the start, less risk over time
An agency charges more per day than a freelancer. It pays for a team, tools, offices, insurance, and a margin. What you buy in return is mainly risk reduction.
When it's the right option
- The project combines several areas of expertise: product design, iOS and Android, back end, back office, integrations (payment, CRM, ERP), possibly AI.
- A date matters. A commercial launch, a trade show, a season, a fundraise. A team absorbs an absence or an unforeseen event; a single person can't.
- The app is at the heart of your business. If it goes down, you lose revenue or customers. You need on-call cover, contractual maintenance, and someone who answers.
- You don't have a technical team and you don't want to become a part-time project manager.
- The project has regulatory constraints: health data, GDPR, accessibility, AI systems subject to the AI Act. On these subjects, a design error costs far more than the difference in rate.
What to demand from an agency
An agency isn't a guarantee in itself. Check the same points as for a freelancer, plus a few more:
- A scoping workshop before the firm quote. A fixed-price quote given without having understood the project will either be inflated out of caution or renegotiated along the way.
- A named team. Who actually works on your project, and for what percentage of their time.
- Full assignment of the rights to the code, delivered and documented in your repository.
- Regular demos on real installable builds, not presentations.
With us, these points are in every contract. Our plans are detailed on the pricing page, and our work (Decathlon Camp, ARENA 5, Yellow Kitchens, VIKFIT, memreez, among others) can be viewed in our portfolio.
Compare the real cost, not the quote
The most common trap is comparing initial quotes. On a product that lives on, developing the first version often represents only part of the total cost over three years.
Do the exercise over a horizon of two to three years, with four lines:
| Item | What to estimate |
|---|---|
| First version | The quote, plus a 15 to 25% contingency margin if the scope isn't defined |
| Enhancements | The features of versions 2 and 3, which you will inevitably have |
| Maintenance | OS updates, security, stores, hosting, subscriptions |
| Takeover scenario | What a rewrite or a change of provider would cost if the chosen option reaches its limit |
The fourth line is the one nobody fills in. Yet it's the one that changes the decision. A cheap no-code build that has to be entirely rewritten after eighteen months isn't cheaper than a custom app that lasts five years, unless those eighteen months taught you something that was worth the difference. Which is sometimes the case.
If you're also weighing up hiring in-house, our article on in-house team or agency does the same calculation for a salaried team.
The decision grid
Answer these questions honestly.
| Question | If yes, lean towards |
|---|---|
| You don't yet know whether the need really exists? | No-code |
| It's an internal tool for fewer than 50 people? | No-code |
| You have a technical team or someone able to manage the project? | Freelancer |
| The scope fits on one platform with a simple back end? | Freelancer |
| The project requires design, iOS, Android and back end at the same time? | Agency |
| A launch date is set and can't be moved? | Agency |
| The app carries your revenue or sensitive data? | Agency |
| You want a single point of contact responsible end to end? | Agency |
Hybrid paths that work
The best projects don't always pick a single option for their whole life.
- No-code, then agency. You validate the need in a few weeks with a prototype, then have the real app built with actual usage data. Scoping is faster and the scope more accurate. Our guide on MVP scope helps you make that selection.
- Agency, then in-house team. The agency builds and stabilises version 1, then trains your first technical hire and hands over the documented code.
- Agency plus freelancer. The agency carries the architecture and maintenance, and a freelancer you know reinforces on a well-defined scope from time to time.
The path that works badly is the one you're forced into: a no-code build that has grown too big, or a freelancer who has become indispensable and unreachable, and an emergency takeover.
In short
No-code buys speed of learning. A freelancer buys value for money, provided you know how to manage and accept dependence on one person. An agency buys risk reduction and continuity, at a higher entry price. No option is good in the absolute: the right one is the one whose main risk you can accept.
For the full budgets of an app by complexity, also read how much does a mobile app cost in 2026, and for our detailed approach, the mobile app development page.
Torn between these options for your project? Describe it to us in a few lines via our contact page or on +33 1 59 13 25 55. We reply within 24 hours, including to tell you honestly if a freelancer or a no-code tool would do the job better.
Sources
- Article L131-3 - Code de la propriété intellectuelle (in French) — Légifrance
- Règlement (UE) 2024/1689 sur l’intelligence artificielle (AI Act) (in French) — EUR-Lex (European Union)
