The annual “trooping the colors'' for our industry is that time when Gartner publishes their hype cycles and quadrants. It is crazy to see the incredible number of cycles that Platform Engineering is on this year. Not to mention Humanitec itself is on a whopping 9 cycles! That’s just crazy. Time to reflect on where we are headed as a movement!
While right now everything is cozy and colored in pink, the uncomfortable awakening will come. Gartner calls it the trough of disillusionment and it will come as certain as the amen in church (as we say in Germany). First question: when? And I think we can be surprisingly precise here.
First for the trough to hit, we need a critical mass of failed platform engineering initiatives. From our data, the progression of large mandated platform engineering initiatives started later than you think and was less sizable than one assumes. In 2022, although PlatformCon did already exist there were very few initiatives, at best a few dozen in the US and the EMEA region. In 2023 Gartner assumes there were around 1,000 cases, our data indicates that we’re more in the region of 600. Steep growth but still ridiculously small numbers. 2024 we’re estimating a range between 1,600 and 2,500 initiatives of all sizes. Humanitec alone is starting the execution of an initiative with a multi-national team with +500 developers every three days at this point. Those numbers are slowly becoming relevant.
But if we can assume that you would need at least 24 months delay until the frustrated and failed initiatives report their failure publicly (and that you will likely not hear about 80% of those cases) we can derive that in 2025 we will slowly start to see a few dozen public failure stories.
By 2026, the trough feeling will become more apparent, and by the end of that year we’ll all be troughing like there’s no tomorrow.
What will be the factors that will lead to initiatives failing? I wrote an article about the biggest fallacies back in 2022. I’m not in the business of crystal ball interpretation but here are my best guesses:
- Missing focus on return on investment
Something I am observing consistently is that people are asked to “do platform engineering” and are starting to do things that just don’t change the top or bottom line. Fancy engineering is useless if it doesn’t have an effect. If you cannot show me in plain numbers how productivity changes, time to market drops, innovation goes up and revenue is boosted you are on the road to failure. You should be absolutely obsessed with ROI, impact, measurement etc. - Missing adoption strategy
SUPER important. My favorite example remains Netflix (watch their PC22 and PC23 talk on how they admit the fancy UI they built wasn’t used by developers). You have to think long and hard about how to choose the right first teams, the scope of your platform, how much adoption speed you promise and anchor etc. - Never finish anything and never reach production readiness
Countless platform engineers I talk to tell me about “plans” and just listening to them overwhelms me. If you think you can build performant platforms from the ground up across any compute type for instance you have to ask yourself: how come Humanitec with all its engineers and full focus on the topic is still primarily focusing on K8s as compute? The answer is pretty simple: it’s super tough to support all those different provider APIs (to avoid confusion we of course support any managed service like DBs, DNS, file storage, etc). If you want to succeed, you need to focus and understand that Platform engineering is an 80/20 game.
Ajay Chankramath, CTO & Managing Director, Platform & Product Engineering at Brillio, and I will be chatting with James Brookbank, Cloud Solutions Architect Manager at Google Cloud at a GCP webinar tomorrow about these fallacies, may be worth tuning in…
Cheers
Kaspar
