Don’t Build Toasters

I saw a TED talk the other day about a guy who built a toaster from scratch, I mean really from scratch, he mined his own iron ore, made plastic from oil and pretty much did everything using pre-industrial tools and technology. He had a great time doing it and it’s a fascinating talk, link below.

Mr Thwaites’ project was a labour of love, his goal was to see if it could be done, he succeeded and all credit to him. However, had his goal been to produce toast then this was a truly terrible idea.

  • It made bad toast
  • It cost £1187.54
  • It took nine months

Amazon will deliver a Russell Hobbs toaster to your door, today, for £16.99. Failing that you could hold bread over your cooker or light a fire or use a blowtorch. In fact pretty much any method of toasting bread you care to name would be faster, cheaper and better in every way than making your own toaster.

And yet this is precisely what many businesses do when it comes to software. Rather than use products that are available now, reasonably priced and someone else’s responsibility to maintain they build or commission something bespoke that costs a fortune, takes forever, isn’t very good and diverts attention and resources from actually running the business.

Usually this happens when a company has internal processes that don’t align with the way that off the shelf software does the job. For example a lead management process that doesn’t fit easily in to off the shelf lead management tools like PipeDrive or a project management workflow that doesn’t fit with tools like Jira.

Think about this for a moment, these tools are built to do a single job well, the processes they are built around are researched and optimised by experts and have been honed by thousands of existing users providing feedback. If your process doesn’t fit, it’s entirely possible that it’s better than the combined wisdom of industry experts and thousands of other businesses like yours but before you go and build “The Homer” be honest about it. Could you conceivably change your processes to work with existing tools? Might it be better?

Another reason bespoke software happens is when someone wants features from various tools in one place.

This spawns monstrosities that do lots of things less well than the single purpose solutions the features came from. It ends up kind of like a Swiss army knife, yes, there are scissors and tweezers but compared to a proper pair of scissors and a proper pair of tweezers they’re crap.

Could you use two or more separate tools? Most modern SaaS solutions integrate with each other anyway, either directly or via services like Zapier. Even if you can’t find tools that will integrate, the chances are they’ll have APIs and that you can have an integration built for a tiny fraction of the cost of bespoke software.

Bespoke isn’t always bad, sometimes there’s a genuine need for a completely new piece of software and a bespoke solution is the only or best way, I’m not saying never go bespoke but do exhaust all the other possibilities first. Don’t build a toaster.

Link to the Ted Talk:



Leave a Reply

Your email address will not be published. Required fields are marked *