Web applications9 min
Custom web application development, explained for the owner who pays for it
When a business needs custom web application development rather than a website: three cases that pay back, how a build runs, cost and timeline.
Last March the owner of a sailing school a few kilometers east of Cassis asked me what it would cost to put online booking on her website, and I ended up telling her she needed custom web application development instead, which wasn't what she'd called about. The weekend before, the wind had canceled all her lessons and she'd spent two evenings texting about forty clients one at a time to move them, working off a paper sheet in the clubhouse.
Her website itself was fine. The way I see it, if the people using the thing have to log in or be recognized, and what they do once they're in changes something on your side like a schedule or a deposit or an invoice, then you're buying an application, and if it's no to either one you only need a website and it'll cost you a fraction of the price.
Website vs web app
For her it was yes to both. Her clients have lessons booked and pay a 40 € deposit, and every time a Saturday gets canceled there are about forty credits to sort out, which she'd been doing on a calculator and writing on the back of the sheet in pencil. With six boats and four instructors and around 350 lessons a month in July and August, that's a lot of evenings.
I haven't always caught this early myself. In 2023 I built a quote calculator into a carpenter's website in Aubagne, and four months later his office manager was retyping every quote from it into their invoicing software, about thirty a month, so what he needed by then was an application and what he had was a form.
The reason it's worth sorting out early is the price. A website with us starts at 1,900 € before VAT and takes two to four weeks (I've gone through what a website costs in another article), whereas an application starts at 9,000 € and takes four to eight, and most of the extra money goes into user accounts and who's allowed to see what, plus a database that's backed up every night, since people book at eleven on a Sunday night.
Three cases that paid back
Every app I've built that paid for itself was replacing a cost the owner already knew about, in euros or in hours, and up to now it's nearly always been one of three kinds, the ones in the web applications column on the services page.
A client portal
An estate agency on the Vieux-Port in Marseille looks after seasonal rentals for about sixty owners, and most of the owners don't live down here. The manager had counted her emails for two weeks before she called me: about forty a week, mostly owners asking whether the July tenant had paid or wanting their statement again. One of her staff spent about four hours every Friday finding PDFs to answer them.
Now each owner logs in and sees their statement for the month and the documents for their apartment, sorted by year, plus a calendar of when it's rented. She'd asked me for a dashboard with charts too, but none of her owners had ever asked for that, so we didn't build it. It came to 9,800 € and was live in five weeks, and she's down to about ten emails a week from forty, mostly owners asking what they should charge next summer.
I go by something Nielsen Norman Group wrote back in 2014, that a login wall is only worth putting up if the people you ask to log in get something real out of it. Her owners log in once or twice a month to check the rent and their documents, so they do. A prospect who comes to your site once doesn't, and if most of your clients are like that I wouldn't build a portal. There's more on what goes in one in the article on client portals for small businesses.
Booking
At the sailing school people book on their phones, often from the quay where the 4G drops out behind the cliffs, so I kept the pages light. Google said in 2017 that the average mobile landing page took 22 seconds to load fully and 53% of mobile visitors left if it took more than three. Put a calendar, the boat list and a payment form on one booking page and load them all at once, and you'll get to 22 seconds without trying.
She didn't believe that asking people to create an account first would cost her bookings, so I pulled up Baymard's list on my laptop. Take the US online shoppers who'd walked away from a checkout for some reason other than just browsing: 18% of them quit because the site wanted them to create an account, and 17% because the checkout was too long or complicated. That was enough for her, so at the school you book and pay the deposit first and make an account afterward if you want to.
There are four screens, one each for the lesson, the slot, the deposit and the confirmation, and when the mistral's too strong she has a button on her side that cancels the lessons, puts the deposits back as credits and texts everyone.
The back office
A refit subcontractor at the La Ciotat shipyards was the third. He has five crews doing paint and woodwork on yachts, and until last year they wrote their hours on paper sheets that the office manager typed into the invoicing software at the end of each month, two days of her time every month, and some of the sheets never reached her at all.
The app doesn't do any invoicing, since their accounting software already does that. It puts the job sheet on the foreman's phone by boat and by task with the hours and a photo, and it sends the totals to the invoicing software at the end of the month. It cost 14,500 € and took seven weeks, and in the first three months they billed about 6% more hours than in the same months the year before on about the same amount of work, which, together with the office manager's two days, paid for it within the year.
Stripe and the Harris Poll surveyed more than a thousand developers and a thousand executives in 2018 for The Developer Coefficient, and the developers put the time lost to maintenance (debugging, refactoring, bad code) at 17.3 hours of a 41.1-hour week. The French ones put it highest of the five countries, at 20.9. I take that as a reason to keep a small business's app small, since every screen you add has to be maintained for as long as you use it, and the refit app has five.
What custom web application development costs, and how long it takes
At the studio an application starts at 9,000 € before VAT, which is a little over $10,000 at the moment, and takes four to eight weeks. For that you get one kind of user, one process such as a booking or a job sheet, and one connection to software you already use, usually the invoicing tool, and every extra role or process or integration adds a few days. The three projects above came to 9,800 €, 11,400 € and 14,500 €, and the article on custom software development cost goes through what moves the number.
For the sailing school I spent the first Saturday at the clubhouse seeing how bookings were really being made, and I gave her a fixed price on the Monday. In weeks two and three she clicked through a mockup with all the screens and no real data and changed her mind about two things, which costs next to nothing at that point. Then I built it, and the instructors did the testing in the last week, since they use it all day and she doesn't.
It was a week late anyway. In week six one of the instructors happened to mention that club members don't pay deposits. Nobody had told me, and the booking and credit screens had to be redone.
After it goes live there's a monthly cost for hosting, backups, security updates and small fixes, between 100 and 300 € a month for an app this size, and the sailing school pays 140 €. If nobody looks after it an application stops working sooner or later, usually because the payment provider or some other supplier changed something.
Choosing who builds it
Most people comparing custom web app development companies look at the portfolio first, and I'd ask two things before that. Who owns the code and the database at the end? You want the answer to be you, written into the contract. And will they come and see how you work before they give you a fixed price? If they can give you one after a single phone call they're guessing, and you'll pay for that later in change requests.
When I say no
I turn down about a third of the application projects I'm asked about. Last spring a dentist in La Ciotat wanted appointment booking built from scratch, and I sent him to a booking subscription at 30 € a month, since a custom version would have cost him at least 9,000 € plus hosting to do what the subscription already did. When something exists as a product like that, whether it's a CRM or a point-of-sale app or online accounting, I'd buy the product.
Building makes sense when the process is specific to your business, a way of booking, delivering or billing that the products on the market don't handle, and the piece on custom software for small businesses is about telling those cases apart.
So far every client I've built an application for has kept their website, because the site is what brings in new people and the application is for the ones who already know you. The studio is in La Ciotat, between Marseille and Cassis in the south of France, and most first conversations happen on the phone.
If you're not sure which one you need you can book a fifteen-minute call, and I'll tell you if a website is enough.
FAQ
What's the difference between a website and a web app?
I'd put it like this: anyone can look at a website and they'll all see the same thing, but a web app knows who you are and lets you do something, like book a lesson or download last month's statement, and what you do gets saved on the business's side. If your customers or staff log in and change your data, you need an app.
Does a web application work on phones?
It does. It runs in the phone's browser and people can put an icon for it on their home screen, and you don't have to go through Apple or Google to publish it. At the sailing school I built one for, nearly every booking comes in from a phone. A native app only makes sense to me for heavy offline use or Bluetooth hardware, or when the brand wants to be in the app stores.
How much does custom web application development cost?
From 9,000 € before VAT at Studio Calanque, which buys one kind of user, one process and one connection to software you already have, in four to eight weeks. Three I built recently came to 9,800 €, 11,400 € and 14,500 €, and each of those clients pays between 100 and 300 € a month for hosting and maintenance.
Who owns the app once it's delivered?
You own it, including the code, the database and the hosting account, which we hand over in your name at delivery so you can take it to another developer later. With a subscription you're paying rent, and your data stays with the company you rent from.