Where it comes from.
Samuel Beckett wrote Worstward Ho in 1983, six years before he died. It is a short, austere prose work, fewer than 8,000 words, preoccupied with persistence and reduction. Beckett by that point had spent decades stripping language down to see what, if anything, was load-bearing underneath.
The famous phrase appears about a third of the way in, in a passage that reads, in full: "Ever tried. Ever failed. No matter. Try again. Fail again. Fail better." Six fragments. Twelve words. The most quoted thing Beckett ever wrote.
It is also, almost without exception, misunderstood.
What it isn't.
You will find the phrase on Silicon Valley office walls, on athletic-brand merchandise, and in the LinkedIn bios of people who are trying to look more thoughtful than they are. It is treated as a slogan for resilience, a permission slip to fail with cheerful confidence and try again.
This is the opposite of what Beckett meant.
The full passage in Worstward Ho is not optimistic. The narrator is not encouraging the reader. The narrator is grimly resigned to the inevitability of failure, and is suggesting, with very dark humour, that the only ambition left is to fail with slightly more precision next time. Not to succeed. To fail better.
"Fail better" is not a path to success. It is a more honest way to fail. That difference matters.
This is harder to put on a poster. It is also, for anyone actually working in a field where iteration is the work, the only reading that is useful.
Why this is the only honest position in marketing.
Marketing is the discipline that pretends most aggressively not to fail. Case studies are curated for the win. Reports are filtered to the metrics that moved. The campaigns that did not work disappear from the slide deck before they reach the board.
Agencies, in particular, have a structural incentive to never tell a client the truth about what didn't perform, because the contract renewal depends on the appearance of competence.
The problem is that this denial does not make the failure go away. It just moves it. The campaign that flopped this quarter is replaced by an identical campaign next quarter, run by the same team, hoping for a different result.
The budget is consumed. The leadership is told that "engagement was strong." Nothing is learned.
I have spent twelve years inside marketing teams and watching marketing teams from across the table. I have not seen a single successful marketing strategy that did not pass through several rounds of clearly identifiable failure first.
The successful strategies are not the ones that didn't fail. They are the ones that admitted to it loudly and changed direction faster.
That is what "fail better" means as a working method. It does not mean fail with optimism. It means fail with rigour.
It means design experiments that can disprove themselves. It means treating an underperforming campaign as a data point, not as an embarrassment to be buried under a different campaign.
Three principles I work by.
Hypotheses, not campaigns.
A campaign asserts. A hypothesis predicts and is willing to be wrong.
Every meaningful piece of marketing work I commit to a client starts as a written hypothesis with a specific, quantified prediction. "I believe that switching from outcome-focused messaging to objection-led messaging on the pricing page will improve trial-to-paid conversion by 15 percent within six weeks." That sentence can be falsified. That is the point.
If I cannot write a falsifiable sentence about a piece of work, I do not know what I am doing. I have learned, painfully, to stop doing work I cannot write a hypothesis for.
Brutal measurement.
Most marketing measurement is a soft hug. Self-reported attribution. Vanity dashboards. Reports that exclude the channels that didn't work.
The first thing I rebuild on most client engagements is the reporting layer, not because the numbers are wrong, but because the numbers have been arranged to be kind.
Brutal measurement is unkind on purpose. It compares against pre-registered predictions. It includes the experiments that failed.
It does not allow a campaign to be retroactively justified by a metric the campaign was not designed to influence. It is unpleasant. It is the only way to know what is actually working.
Intentional iteration.
Iteration is not "I'll try again next quarter." Iteration is a structured response to evidence.
After a hypothesis is tested, the result feeds directly into the next hypothesis. If the prediction was wrong, the next experiment is narrower. If the prediction was right, the next experiment scales the conditions until it breaks. The cycle has cadence and discipline.
Most marketing teams confuse iteration with churn. Real iteration produces an increasingly precise model of what works for this product, this audience, this market, at this time. Churn produces a calendar full of activity and a strategy that doesn't change.
Permission to fail in public.
I named the practice Fail Again. Fail Better. because the name is a commitment. It tells clients, before they sign anything, that the work will include things that do not work. It tells me, every time I am tempted to dress up an outcome, that I am working against the brand.
It also gives a team permission. The largest unspoken cost of marketing dysfunction is the energy spent hiding things from each other, campaigns that underperformed, hypotheses that were wrong, vendors that were not delivering.
When the working agreement of the team is that failure is the unit of analysis, that energy comes back.
This is what Beckett, in his bleak way, was pointing at. Not optimism. Not a slogan. A more honest way to fail, which, in the end, is the only way anything new gets built.
If this is the kind of working relationship you want, the audit is the place to start. If you would rather not, that is also a useful piece of information about fit.
Anže Frantar · Ljubljana