Innovation Tokens
Every organisation can afford only a few novel technologies. Spend them where they create real advantage.
“Let's say every company gets about three innovation tokens. You can spend these however you want, but the supply is fixed for a long while.”
To an executive
A company can only absorb a few new technologies at once. Spend that limited appetite for risk on what sets you apart, and buy proven tools for the rest.
To an engineer
Default to well-understood technology. Before adopting something new, ask whether it differentiates the product and whether anyone on call will know how it fails.
The idea
In his 2015 essay "Choose Boring Technology", Dan McKinley suggested imagining that each company gets about three innovation tokens. Every new, unfamiliar technology spends one, because its failure modes are unknown and the organisation has to learn them the hard way. The supply is fixed for a long while, though a mature, stable company may earn a few more.
Boring technology is not bad technology. It is technology whose failure modes are well understood, documented and operable by people who are not its original enthusiasts.
Why it works
Unknown failure modes are the real cost. A proven database has its sharp edges written up; a new one finds yours in production.
Every technology adds to the operational surface. Each one needs monitoring, patching, skills and on-call knowledge. The cost is paid across the whole stack, not just by the team that chose it.
Scarcity forces a conversation about value. If you can only afford a few novel choices, you have to say which one matters most to the business.
Where it applies
- Technology radar and standards decisions, where the enterprise decides what is adopted, trialled or held.
- Vendor and platform selection, especially when a new product promises a lot but few people in the organisation know it.
- Greenfield programmes, where it is tempting to modernise everything at once.
- Portfolio reviews, counting how many distinct technologies the estate already has to support.
How to apply
- Name the differentiator: the capability where doing something new would really help the business.
- Count the tokens already spent by the team and across the estate.
- Default to the proven option everywhere else, ideally one the organisation already runs.
- Make adding a technology a deliberate decision, with a plan for who operates it and how you would back out.
- Retire technologies as well as adopting them, to free tokens up.
Example scenario
A digital team adopted a new graph database, a novel stream processor and a bleeding-edge front-end framework at once. Most of its incidents in the first year came from those three, and few people elsewhere in the organisation could help.
The enterprise architect asked which of the three actually differentiated the product. Only the stream processor did. The team moved the graph workload onto the PostgreSQL platform the organisation already ran, and used the standard front-end stack. Incidents dropped and the team got back to product work.
Questions to ask
- Is this new technology part of what differentiates us, or could a boring option do the job?
- How many innovation tokens has this team, or this estate, already spent?
- Do we know how this fails, and who will fix it at 3am?
When not to use it
- Research and prototyping, where learning about new technology is the goal and nothing will be operated in production.
- When the boring option cannot meet a real requirement, which is exactly what a token is for.
How it misleads you
Never adopting anything new. Today's boring technology was once novel, and an estate that never spends tokens slowly falls behind. Spend them deliberately, not never.
Further reading
Last reviewed · Suggest an improvement