Every company gets about three innovation tokens. The supply is fixed. Spend one on an unproven database, and it is gone. Spend one on an unproven framework, and it is gone. You will be tempted to believe you have more than three, and you do not.
The essay was written for engineers picking databases. The idea applies at least as strongly to a business owner picking an invoicing tool, and almost nobody applies it there.
Boring does not mean bad
Worth clearing up immediately, because the word invites the wrong reading.
There is technology that is boring and bad. Avoid that. There is a great deal that is boring and good, or good enough, which is the category we are talking about.
Boring means the thing has been in wide use long enough that people have found the problems. The capabilities are known. More importantly, and this is the actual point of the essay, the failure modes are known.
With a proven tool, you know how it breaks. With a new one, you do not know what you do not know, and that unknown is where the expensive surprises live.
The asymmetry nobody prices in
When you compare two tools, you compare features and price. Both are visible. Both are in the marketing material.
What does not appear in that comparison is everything that happens after you commit.
With something established: the answer to your question already exists on the internet. A search returns a result rather than a forum thread with no replies. Other tools already connect to it. Your accountant has seen it before. If you hire somebody, they may already know it. And it will probably still exist in three years.
With something new: you are the documentation. Every integration is a request rather than a setting. Support is one overworked founder. And there is a real chance the company is acquired, pivots, triples its price, or disappears, and you are migrating on their schedule rather than yours.
None of that appears in the comparison table, and all of it costs more than the subscription difference.
The new argument, which did not exist in 2015
Here is what has changed recently, and it is concrete rather than philosophical.
Language models learned from what has been written on the internet. Established tools have millions of pages written about them: tutorials, forum answers, documentation, worked examples, people solving the exact problem you are about to have.
A tool that launched fourteen months ago has almost none of that.
Which means when you ask an assistant how to do something in a widely used accounting package, you get a specific and usually correct answer. Ask the same question about a newer competitor and you get a confident guess, because the model is filling a gap rather than recalling a fact.
The boring choice used to be the safe one. It is now also the faster one, because the help you need is available at a level the new tool cannot match regardless of how good its interface is.
How many tokens does a business of one get
Fewer than three. Honestly, about one.
McKinley was writing about companies with engineering teams who could absorb a bad choice. A business with one or two people cannot. There is no colleague who knows the workaround and no capacity to run a migration in a bad quarter.
So the rule tightens. Everything in your operations should be the dull, obvious, widely used option, and your one token gets spent on whatever actually distinguishes your business.
If you are a design studio, spend it on design tools. If you are a restaurant, spend it on something that materially changes the guest experience. Do not spend it on the thing that sends invoices.
A photographer chooses an established accounting package, an established email platform, an established website builder, and an established booking system. Four boring choices, all made in an afternoon, all with answers available for any question. Then she spends her single token on a newer editing tool that genuinely produces better results than the standard option. That is the correct allocation. The reverse, which is common, is exciting infrastructure supporting ordinary work.
What counts as boring in small business software
Not a product list, since those change. The signals.
- It has existed more than five years under the same name and ownership
- You can find recent independent tutorials that are not published by the vendor
- Other software you use already lists an integration with it
- Your accountant, or somebody in your industry locally, has used it
- The pricing page has tiers that have been stable rather than a single figure that recently changed
- A search for the product name plus a common problem returns actual answers
Six signals. Four or more and it qualifies.
Boring versus abandoned
This is the honest difficulty, because the failure mode of this advice is choosing something genuinely dead.
Boring software is maintained, unexciting, and improving slowly. Abandoned software has stopped changing entirely. They look similar from the outside and are very different to depend on.
Three checks separate them. When was the last update, and is it within the last several months. Is support answering, which you can test before subscribing with a real question. And is the company still selling, meaning are they publishing, hiring, or appearing anywhere, rather than quietly maintaining a customer base while planning an exit.
Slow is fine. Stopped is not.
When to spend the token
Not never. The exceptions are real and worth naming.
Spend it when the new option does something the established one genuinely cannot, and that capability affects what customers experience. Spend it when the established option is failing you in a specific, describable way rather than a vague one. Spend it when the cost of being wrong is contained, meaning you could leave in a week without losing data.
Do not spend it because something is new, because a competitor mentioned it, or because the current tool is merely unexciting. Unexciting is the goal.
Why this matters more than it sounds
The appeal of new tools is that they feel like progress. Adopting one produces the sensation of having improved the business.
Most of the time what actually improved is your familiarity with a settings screen.
Every hour spent evaluating, migrating, and learning is an hour not spent on the work customers pay for. The businesses that do well are rarely running the most interesting software. They are running dull, dependable software they stopped thinking about years ago, which is precisely what freed them to think about something that mattered.