How to read a software quote (and when not to build custom)

Why three quotes differ so much

It is the standard experience of anyone buying software for the first time: you ask for three quotes, they come back with figures that look nothing alike, and you have no way to judge which is reasonable. The temptation is to assume one of them is lying. Usually something else is going on: all three understood a different job, and none of them wrote it down.

• Different scope behind the same words. “Online store” can mean a shop window, or a shop window wired to the ERP, the warehouse and the carriers. There is twice the work between them, and the heading on the document is identical.
• Different definitions of “done”. Migrating the data, training your team, going to production and standing behind the first two weeks of real use is all work. If it is not in the document, you will still pay for it, just later.
• Different things included forever. A price with the first year of maintenance inside it and a price without cannot be compared from the front page.
• Risk allocated differently. A fixed price carries the buffer of whoever signs it. An hourly price moves that risk to you and nearly always looks cheaper up front.

What a quote has to state

• Who owns the code, the repository and the app store accounts. If they are not in your name, you have not bought software: you have rented a dependency.
• What year two costs. Hosting, third-party services and maintenance, with a figure beside them. Ours is published: from €150 a month, and from €450 once there is live data and native apps.
• What it takes for granted. “We assume an API exists”, “we assume you supply the copy”. Assumptions are where the overrun lives.
• What happens if you change your mind. How a change of scope gets priced, agreed before there is one.
• Who does the work. Not the company: the actual people, and how much of their time.
• What is delivered besides the software. Documentation, credentials, and the session where your team learns to run the admin panel without calling anyone.

A quote that does not answer those six questions is not cheaper: it is less complete. The difference shows up on the second invoice.

Compare like with like

The only way to know whether a figure is reasonable is to know what that figure buys. These are our published starting prices, and they work as a reference for reading anybody else’s:

When not to build custom

The useful question is not “what does it cost to build?” but “does this already exist?” If somebody sells what you need, buying usually wins, and not because of the initial price: it wins because the maintenance, the security work and the new features are paid for by thousands of customers between them instead of by you alone.

• What you need looks a lot like what everybody needs. Invoicing, taking payments, sending newsletters, selling standard products.
• Your advantage is not there. If your business does not make money by doing that part differently, doing it differently only costs money.
• The figure you were given sits below the real floor for that kind of system. What you are buying then is a first version somebody will have to rebuild.
• Nobody is going to operate it afterwards. A custom system with no maintenance agreement is a loan repaid in year two.

We apply this to ourselves. Eolas Prints manufactures its own filament — there the advantage is real — and resells the 3D printers, laser engravers and scanners: we did not build a printer in order to sell printers. In software we do the same. The store is Shopify, because it is nearly always the right answer, and we say so even though it is not the most expensive option for us. Simientes Infinitas sells into 27 EU markets on that same machinery.

When you should

• When the off-the-shelf option would force you to change how you work, and how you work is your advantage.
• When the joining is the product. Nobody sells the connection between your ERP, your warehouse, your carriers and your store: that always gets built.
• When the data you need together lives in places that do not talk to each other. That is literally what SomosTorre does with more than twenty-five municipal sources.
• When the per-seat licence, multiplied by your headcount and by three years, already looks like the cost of building it.

And one rule that saves a lot of argument: buying and building are not opposites. Almost every good system is a well-chosen bought tool plus the custom integration that wires it into the business. The expensive mistake is not buying or building — it is building the part that was already for sale.

The cheap shortcut: paying to find out

On a large project the expensive decision is not who you hire, it is what you ask them for. That is why we sell discovery as a product at a fixed price: €4,500, two weeks, ending in an architecture, a phased plan, the risks and a firm quote you could hand to any other studio. If we build it, the fee is credited back against the project in full.

That is not generosity. A discovery billed by the hour rewards dragging it out, and a figure that comes out of two weeks of work is a figure that does not move afterwards. Three quotes written against the same document can be compared. Three quotes written against a half-hour call cannot.

The dangerous quote is not the highest one: it is the one that does not say what it takes responsibility for.