digi.ai.in
← digi.ai.in / blog
2026-07-318 min readHiring a studio

Why We Default to Boring Technology on Purpose

Because unproven tech spends a budget you can't see until it's gone: your team's attention.

When a client asks why we're not building on whatever framework is trending this month, the answer is simple: predictable technology ships faster, breaks less often, and costs less to maintain. That's not a lack of ambition. It's a deliberate trade-off, and there's a name for the thinking behind it.

The idea: you only have a few "innovation tokens"

The clearest articulation of this comes from a widely-cited 2015 essay by former Etsy and Mailchimp engineer Dan McKinley. His argument: every organization has a small, fixed budget of what he calls "innovation tokens" to spend on unproven technology. As he puts it, every company gets about three innovation tokens, and you can spend these however you want, but the supply is fixed for a long while.

Spend a token on a trendy database, and you've committed some of your organization's limited attention and troubleshooting capacity to that one bet. McKinley's own example: if you choose to write your website in NodeJS, you just spent one of your innovation tokens, and if you choose to use MongoDB, you just spent one of your innovation tokens. Spend all three tokens on the stack, and you have none left for the actual business problem you're supposed to be solving.

This is the part people misread. "Boring" doesn't mean "outdated" or "low quality." It means the tool's behavior under stress is already known. In McKinley's words: the nice thing about boringness is that the capabilities of these things are well understood, and more importantly, their failure modes are well understood. A more recent summary of the essay puts it the same way: McKinley introduces the concept of innovation tokens as a simple way to track and limit the introduction of unfamiliar technologies or architecture.

Why a client should care, not just a developer

Most of the businesses we build for aren't technology companies. A retail chain, a logistics operator, a D2C brand — none of them are in the business of running infrastructure. Their business is serving their customers. A LinkedIn analysis of the same idea makes this point directly: for most businesses, using a boring technology stack lets you get the benefits of recruiting, speed of development, and scalability, because your company is in the business of solving your customers' pain points, and they don't care what technology you use.

Put plainly: your customers don't know or care whether your site runs on a two-year-old framework or a two-month-old one. They care whether checkout works, whether the app loads, and whether support tickets get resolved. Every hour your developer spends debugging an obscure edge case in a new framework is an hour not spent on those things.

A newer reason to stay boring: AI coding tools work better on well-known stacks

McKinley wrote his essay in 2015, before AI coding assistants existed. A decade later, commentators are adding a second, distinct argument for boring tech that has nothing to do with his original point about organizational risk. As one 2024–2025 piece argues: now with AI, boring technology has gained another concrete, measurable advantage — most LLMs are trained on the internet, meaning established, well-documented, stable technologies are extremely well represented.

In practice: when our developers use AI tools to speed up routine work, those tools are far more reliable on a mainstream stack with years of public documentation, Stack Overflow threads, and GitHub issues than on a framework that's six months old with a thin community around it. A retrospective on the original essay from mid-2025 summarizes the durability of the underlying idea: McKinley's core argument was that companies have limited "innovation tokens" and should spend them strategically on established, well-understood technologies rather than exciting but unproven ones. The AI angle is a new argument stacked on top of an old one — same conclusion, different reasoning.

The part you can actually measure: speed and revenue

Separate from the "boring stack" argument, there's a directly measurable reason to avoid unnecessary technical complexity: heavy, unproven front-end frameworks and bloated third-party scripts slow sites down, and slow sites lose money. Google's own performance team is blunt about this: slow sites have a negative impact on revenue, and fast sites are shown to increase conversion rates and improve business outcomes.

This isn't theoretical for Indian businesses — two direct examples

Google's own Core Web Vitals case studies include two Indian businesses, which makes the numbers harder to wave away as "not relevant to our market."

redBus, India's largest online bus-ticketing platform, worked on performance rather than feature additions and saw real gains. Per Google's case study: Core Web Vitals fixes contributed to 80-100% mobile conversion rates and significant domain ranking uplift across redBus's global market properties. The technical fix wasn't exotic — it was fixing slots for page components and removing unoptimized tag insertion scripts to improve CLS, and reducing TTI from around 8 seconds to around 4 seconds. No new framework. No rewrite. Just removing what didn't need to be there.

The Economic Times, India's leading business daily, is cited by Google for a similar result: how The Economic Times passed Core Web Vitals thresholds and achieved an overall 43% better bounce rate.

Both examples point the same direction: the return on "boring and fast" was bigger than the return would likely have been from adopting something new and untested.

A rough decision checklist we use

When a client or our own team is tempted by a new framework, library, or platform, we run through a short set of questions before spending an innovation token:

Question Why it matters
Is this solving our actual bottleneck, or is it interesting to us as engineers? McKinley's point is that tokens are wasted on problems the business doesn't have source
Do we understand its failure modes, or just its happy path? Boring tools are boring because their failures are documented source
Will this slow down page load or add third-party script weight? Even 0.1 seconds moves conversion numbers source
Is it well-represented enough for AI coding tools to help us maintain it later? Well-documented stacks get better AI assistance source
Can we hire and train someone else on this in six months? Recruiting and scaling are easier on familiar stacks source

If a new technology doesn't clear most of these, we default to the boring option and put our attention into the actual product — the checkout flow, the search, the thing customers touch.

Where this argument gets misapplied

It's worth being precise about what "boring" actually protects against, because the term gets stretched in two different directions in practice.

McKinley's original framing is organizational: a team only has so much attention, and unfamiliar tools consume it in ways that are hard to predict in advance https://mcfunley.com/choose-boring-technology. A lot of the web-performance literature reframes "boring vs. exciting" as purely a front-end speed and SEO question https://web.dev/learn/performance/why-speed-matters. These two arguments overlap — a slow, unfamiliar framework hurts you on both fronts — but they are not the same claim, and it's easy to conflate them when reading advice online. Slow doesn't always mean unfamiliar, and unfamiliar doesn't always mean slow. We check both separately.

What this looks like in a project

For most of the web and app work we take on — business sites, booking flows, internal tools, e-commerce — there is no problem novel enough to justify spending an innovation token on unproven technology. The interesting work is in the business logic, the data model, the integration with whatever payment gateway or logistics API the client already depends on. That's where we spend the attention budget. The stack underneath stays boring on purpose, so it can be maintained by whoever picks it up next, debugged at 11pm without a search through GitHub issues from three weeks ago, and kept fast enough that the numbers above stay on the right side of the ledger.

Questions people also ask

Does "boring technology" mean using old or outdated tools?

No. McKinley is explicit that boring means well-understood, not old — the point is that the tool's capabilities and failure modes are already mapped out, which is different from simply being outdated source.

Isn't this just an excuse to avoid learning new things?

McKinley's argument allows for some new technology — that's what the innovation tokens are for. The point is spending them deliberately on problems that actually need it, rather than on every available trend at once source.

Does site speed really affect revenue, or is that assumed?

It's measured, not assumed. Google cites the BBC's finding of a 10% user loss per extra second of load time, and a Deloitte/Google study found a 0.1-second improvement lifted retail conversions by 8.4% and travel conversions by 10.1% source, source.

Written by the studio behind digi.ai.in. Questions? hi@digi.ai.in or WhatsApp +91 98605 96905
WhatsApp us