In 2019 I walked away from a 20-year investment banking career and went back to school. I spent a transformative year at London Business School discovering a huge amount (mostly about myself) including learning that the whole “First Mover Advantage” thing is not actually a thing at all.
Everyone assumes Facebook was the first social network, but actually MySpace and Friendster got there earlier. If you think Google invented search, you are too young to remember that AltaVista and Yahoo were not just there earlier but were also the dominant search engines.
Looking at today’s landscape between OpenAI, Gemini, Claude and others: at the beginning there was clear blue water (IMO) between the efficacy of the LLMs and ChatGPT was beating the others hands down. Now, I don’t believe there is a huge amount between them. Differences are emerging in specific use cases, interfaces and how connected they are to your underlying data. Who will be a winner? Who knows!
Over the past few months, we’ve won three competitive enterprise RFPs in autonomous negotiation. We were not first to market, nor the biggest nor the most funded. It has made me reflect on how often first-mover advantage works and whether the dynamics are changing because of AI.
What the Evidence Actually Shows
Research across industries, including analyses published in Harvard Business Review, finds that pioneers do not de facto become long-term leaders. In many markets, “fast followers” outperform. The burden of being first is heavy, all the mistakes are on you!
- They educate the market at their own expense.
- They make early technical assumptions under uncertainty.
- They lock in processes and teams around those early decisions.
- They give competitors a live experiment to learn from.
The Architecture Trap
In technology, underlying architecture decisions compound.
Here at Nibble, we started experimenting with LLMs from the moment ChatGPT was released. It was a bit crap at first, so we stuck to the fixed copy and NLU (natural language understanding, i.e. how old chatbots were made) approach – it categorically outperformed in A/B tests. We threw away more than one iteration of the technology, but without those experiments we would never have been ready for “agentic” in 2025.
Funnily enough, we also lost some good stuff in the old bot when we did switch to agentic. Some of the personality and some of the fixed copy in the first generation was awesome and made me grin. But you can’t cling onto tiny features in the face of overwhelming benefits from a new technical approach.
In a more well-known comparison, Netflix was not first in home entertainment. Blockbuster already dominated physical rentals, but their infrastructure, incentives and retail footprint were optimised for stores. Netflix designed for the distribution economics that were emerging, first DVD-by-mail, then streaming.
You can say the same about Kodak. When faced with the opportunity to lead in digital photography, it felt like throwing away too much so they famously fell by the wayside.
AI Accelerates the Cycle
AI amplifies this dynamic because you can pivot and shift your business focus and / or your tech stack capability so fast. We are living proof of our pivot to procurement at the same time as our tech stack went agentic. In an old world this could have been suicidal to completely rebuild the company from the ground up, but AI makes it remarkably achievable on so many levels.
It can also remove the moats you think you have built. We thought our database of 2 million historical autonomous negotiations was a unique asset. To a degree it still is, as this was from real people (not AI simulations) chatting to our bot. But 2 million is not a huge dataset in model-training terms, it’s much less than the volume you need for machine learning / reinforcement learning techniques.
What it does give us is a place to A/B test live on 50–100,000 autonomous negotiations a month so we can ensure the all refinements we are building into Nibble’s procurement bot improve performance.
The sands shift. It’s not a moat anymore, but it is an opportunity.
We couldn’t be more acutely aware of these risks and challenges from AI in today’s environment.
- Early technical advantages erode quickly.
- Systems built on earlier model assumptions require retrofitting.
- New entrants can build natively on improved infrastructure.
When the frontier moves this fast, any first-mover advantage is fundamentally weakened.
And Sometimes the First Mover Wins
First-mover advantage is not a myth, it does work sometimes, it works often, just not every time.
Amazon entered ecommerce early and compounded logistics scale over decades. Switching costs and infrastructure depth became formidable barriers.
Microsoft established Windows as a platform early in personal computing. Network effects and developer ecosystems reinforced dominance.
First movers win when technology evolves slowly enough relative to their ability (often scale) to pivot. As technology moves faster, it becomes more likely that the big players will win — I’m thinking Microsoft, Google, etc. Have you seen their capex plans lately?
There is considerably more detail here — recommended reading if you are considering stock market investment in AI soon:
https://giftarticle.ft.com/giftarticle/actions/redeem/1e26635e-ca50-4181-868e-c824f5bc437d
Why This Matters Now in the age of autonomous negotiation
Technology is changing fast and the market is still maturing. I am not doing myself a favour in saying so, but this doesn’t feel like the right time to sign a 5 year SaaS contract with anyone, and certainly not because they are the first mover in the market 😉
There you go – Free negotiation advice for SaaS procurement leaders.
Just One More Thing
Parenting is full of ambition for your kids but according to The Economist, there is no need to be a “tiger mom”, it doesn’t work.
Stay calm, let them nurture a wide variety of interests and tread their own path because many of the most successful people in life achieve were relatively normal in childhood.
“Nobel-prizewinning scientists were less likely to have won academic scholarships than those nominated for a Nobel who did not win. They also took longer to reach senior academic positions, had less impressive early publication records, and maintained interest in fields beyond that for which they won their prize.”
