Seth Godin on product launches: build an audience before the big reveal
A product can technically work and still have nobody ready to use it. Plan how people will try it, learn it and return after the launch announcement.
Test a complete first use with a relevant group, including the support required to make it useful.
A launch is more than a finished feature
Godin points out that some products depend on community adoption and an audience willing to try them. An engineering definition of a minimum product may miss those conditions. Our application is to plan the first useful experience and the group that will take part in it.
This does not require building a large following before every experiment. It requires being explicit about how the product becomes useful. A solo calculator and a team collaboration tool have different starting conditions. The second may fail if only one person in a team joins.
Choose a complete first job
Imagine a founder developing a handover tool for small creative agencies. The pilot job is not “explore the dashboard”. It is handing a live but non-confidential project brief from an account manager to a designer and checking whether the recipient can begin work.
Recruit people whose current process makes that job relevant. Explain what exists, what is still experimental and what effort you are asking them to contribute. Give them a real way to decline. Friends who politely admire the interface may not reveal the problems an actual team will face.
Use realistic sample data when sharing real project information would be inappropriate. The pilot can still test whether a recipient understands responsibilities and next steps.
Write a pilot agreement in plain language
The agreement is a working brief: who participates, which job they will try, what help is available, what you want to learn and when you will review the experience. Separate the cost of the pilot from any later commercial offer.
For an illustrative two-week trial with four agency teams, the founder might attend the first handover and then observe whether the next one happens without assistance. The difference matters. A product that works only while its creator narrates every step has not yet demonstrated independent usefulness.
Measure a return, not just a registration
Choose an action that reflects the product’s job. A completed handover is more informative than an account being created. A second handover initiated by the team is more informative again. Record the support required alongside those actions.
Ask participants what they would use instead if the tool disappeared. A spreadsheet, a chat thread or doing nothing is a real alternative. Do not restrict the comparison to products in the category you hope to join.
Small pilot groups cannot provide reliable market-size estimates. They can expose confusing steps, missing capabilities and false assumptions. Use the results for those decisions and keep wider demand claims separate.
Decide what happens after the launch
Set a review date before the first announcement. Decide which findings would justify a larger pilot, which would require a redesign and which would make you stop. Otherwise launch-day enthusiasm can become a substitute for learning.
If you proceed, explain the product using what users actually did and valued. Avoid turning four experimental teams into an unsupported claim of market leadership. An accurate story gives the next group a clearer reason to try the product.
Try it in your business
Define one complete first job for a new offer. Name the participants, the support required and one action showing repeat use. Write three conditions for expanding the pilot and one condition that would make you stop or redesign it.
Sources & editorial note
- Seth Godin: When “minimal viable product” doesn’t work
NEXT MBA editorial analysis, prepared with AI assistance from the public sources above. The applications, scenarios and exercises are NEXT MBA examples, not quotations or transcripts. Product recommendations are ours; they do not imply the speaker’s endorsement of those products. Sources checked on 1 October 2026.