September 22, 2026 · 7 min read

Does being first still matter, or is being best enough?

Pioneers Fail:
47%

Being first has almost never mattered as much as the people selling urgency would like you to believe. Peter Golder and Gerard Tellis analysed roughly 500 brands across 50 product categories using historical records rather than surveys of survivors, and found that market pioneers failed 47% of the time and held an average of 10% market share. The early leaders who entered afterwards failed around 8% of the time and held 28%. In only 4 of the 50 categories did the pioneer hold long-term leadership. The average early leader arrived about thirteen years after the pioneer, which is a long time to be late and still win.

47%
Failure rate of market pioneers, against roughly 8% for the early leaders who followed them
Golder and Tellis, 500 brands in 50 categories
10 vs 28
Average market share held by pioneers against early leaders, in percentage points
Golder and Tellis, Journal of Marketing Research
4 / 50
Categories in which the pioneer still held long-term market leadership
Golder and Tellis
95%
Enterprise generative AI pilots producing no measurable profit and loss impact, from more than 300 disclosed initiatives
MIT Project NANDA, July 2025
/ 01

Why the legend survives the evidence

The finding is more than thirty years old and it keeps losing to the story, for a reason worth understanding. Golder and Tellis pointed out that the earlier studies claiming a pioneer advantage drew their samples from surviving firms. If you only interview the companies that are still here, the pioneers look invincible, because the pioneers who died are not answering the phone. Correcting for that single bias turned an established law of marketing into a legend.

The second reason is that we remember categories by their winners and then misremember when the winners arrived. Plenty of companies commonly described as first were not, and the ones that endured tended to share something other than timing: a large vision of a mass market, persistence through a long unprofitable stretch, relentless improvement of the product, and a willingness to commit assets ahead of proof. Tellis and Golder’s later work on enduring leadership names those factors directly. None of them is speed.

PIONEER AGAINST EARLY LEADER 500 BRANDS, 50 CATEGORIES, HISTORICAL ANALYSIS FAILURE RATE 47% 8% AVERAGE MARKET SHARE 10% 28% PIONEER EARLY LEADER
Golder and Tellis, Pioneer Advantage, Journal of Marketing Research. Historical analysis designed to include firms that failed.
/ 02

The argument on the other side

There is a serious case that this time is structurally different, and it deserves to be put properly. The claim is that AI systems improve with use, so a company running them in production accumulates proprietary operational data that a later entrant cannot buy. On that reading the gap widens rather than closes, which reverses the usual pattern where a follower gets to learn from the leader’s expensive mistakes. The Signal’s primer on answer engine optimization is one place that compounding argument gets tested in practice.

The claim is coherent and the evidence for it is currently thin. MIT’s Project NANDA study, built from more than 300 publicly disclosed enterprise AI initiatives, 52 structured interviews and 153 survey responses collected in the first half of 2025, found that 95% of generative AI pilots produced no measurable profit and loss impact despite $30bn to $40bn of spending. If early movement compounded automatically, that number would look different. What the researchers found instead was that task-level gains failed to become systemic unless somebody redesigned the work around them. Data compounds when an operating system exists to compound it. Otherwise it accumulates, which is not the same thing.

Being early buys you the right to learn something expensive. Whether that becomes an advantage depends entirely on whether anyone at the company is set up to learn it.

/ 03

What actually transfers

If speed is not the durable asset, what is? The Ehrenberg-Bass Institute’s work on mental availability, developed by Jenni Romaniuk and Byron Sharp, offers the most practical answer available to a mid-market company. Buyers do not hold a ranked list of suppliers. They hold associations between specific situations and specific names. The situations are category entry points: the moment the equipment fails, the moment the auditor asks a question, the moment a contract comes up for renewal. A brand linked to more of those moments gets thought of more often, and being thought of is the whole game. That is the argument of why a company needs an opinion, from the other direction.

That asset is built slowly and it does not care when you entered the category. It cares how consistently you have been present at the moments that matter, and how easy you are to bring to mind when one of them occurs. A late entrant who owns three entry points completely will beat a pioneer who owns none of them clearly, which is roughly what Golder and Tellis found and Ehrenberg-Bass explains.

The machine layer has not changed this so much as made it legible. When a buyer asks an assistant who handles a particular situation, the answer is assembled from what the internet has consistently associated with that situation. Being the answer is the same discipline it always was: be specific, be consistent, be present at the moment the question forms. What is new is that the association is now written down somewhere and can be read back.

WHAT ACTUALLY COMPOUNDS SITUATIONS THAT BRING A NAME TO MIND THE NAME THE AUDIT ARRIVES THE SYSTEM FAILS RENEWAL IS DUE A BOARD ASKS WHY GROWTH STALLS TIMING OF ENTRY IS NOT ONE OF THESE
Category entry points framework, Romaniuk and Sharp, Ehrenberg-Bass Institute for Marketing Science.
/ 04

Four questions for a leadership team arguing about speed

/ 01

First at what, exactly

Being first to a technology is not the same as being first to a market. Most claimed firsts are the former, and buyers have never once paid a premium for it.

/ 02

Can we still be here in five years

The enduring leaders in the research shared persistence through an unprofitable stretch. If the plan requires the category to reward you inside eighteen months, you are not early, you are exposed.

/ 03

Which three situations do we own

Name the moments that should make a buyer think of you. If nobody in the room can name three without hedging, the timing debate is a distraction from a positioning problem.

/ 04

What are we learning that a follower cannot buy

If moving early produces knowledge that stays in one person’s head, it produces nothing. Compounding requires somebody to write it down and change how the work runs.

/ 05

The part nobody wants to hear

The reason the first-mover argument keeps winning boardrooms is that it is the only version of strategy that can be executed in a quarter. Being best takes years and cannot be announced. Being first can be decided on a Tuesday, funded on a Wednesday and put in a press release by Friday, which makes it enormously attractive to a leadership team that needs to demonstrate motion. Gartner’s 2026 survey found CMOs putting 15.3% of a flat budget into AI while only 30% said they were ready to scale it. That is not a first-mover strategy. That is a deadline.

What separates the companies that turn early movement into a durable position from the ones that simply move early is not appetite for risk. It is whether anything they learn survives the person who learned it. Most organisations run experiments and store the results in the memory of whoever ran them, then reorganise, and the knowledge leaves in a resignation letter. The advantage evaporates quietly and everybody concludes the market was harder than expected.

So the honest answer to the question is that being first is worth very little on its own, being best is worth a great deal and takes longer than anyone wants, and the thing that actually converts one into the other is an operating system for learning that almost nobody builds, because building it is nobody’s job and shipping the thing is everybody’s.

Being first is a date; being best is a practice, and only one of them is still true about you in five years.

SOURCES: GOLDER AND TELLIS, PIONEER ADVANTAGE, JOURNAL OF MARKETING RESEARCH, 500 BRANDS IN 50 CATEGORIES · TELLIS AND GOLDER, WILL AND VISION · MIT PROJECT NANDA, THE GENAI DIVIDE, JULY 2025 · EHRENBERG-BASS INSTITUTE, ROMANIUK AND SHARP, CATEGORY ENTRY POINTS · GARTNER 2026 CMO SPEND SURVEY, 401 LEADERS, MAY 2026
DFJ
David Fox Jensen
Founder & Principal · The Sentient Marketer

20+ years leading marketing for 150+ brands. Now runs a principal-led fractional CMO and AI consulting practice for mid-market companies. Writes The Signal weekly on what the machines are learning about your business. Machines for scale, never for soul.

Subscribe

Wondering what the machines say about you?