Rules
Product launch marketing mistakes: the traps and the way around them
Product launch marketing mistakes come from naming a market before proof exists and treating launch day as the finish line rather than the start.
What to take away
The costliest launch mistake is a positioning claim that outruns the evidence behind it. Launch day is the start of market learning, not the end of the work.
- Theranos claimed its Edison analyzer could run many tests from one finger prick; the claim collapsed in 2015 and Elizabeth Holmes was convicted of fraud in 2022. Every downstream asset inherits the gap.
- Measure qualified adoption after release, not download or registration counts. MoviePass passed 3 million subscribers on a $9.95 monthly plan, then shut down in September 2019.
- Give prevention, detection and correction different owners. One person holding all three will close the ticket before the public asset is fixed.
- Keep a dated decision record with the claim, the evidence, the version and the reviewer, so a later correction can be traced to its source.
Launch mistakes by name: outrunning proof, untested segments, proxy metrics
Positioning that outruns proof. Theranos raised money on claims that its Edison device could run hundreds of tests from a finger prick. The Wall Street Journal reported in October 2015 that the machines did not deliver. Founder Elizabeth Holmes was convicted of fraud in January 2022 and sentenced that November to 135 months.
Launch mistakes and fixes
Mistake
- Positioning outruns proof
- Rank claims by evidence
- Launch day as finish
- Own first 90 days
- Market named early
- Test segment in pilot
- Proxy metrics
- Track qualified adoption
- Agency headcount
- Ask category launches
- Localized claims
- Check each language
Fix
- Positioning outruns proof
- Cut unsupported claims
- Launch day as finish
- Correct early
- Market named early
- Sell to real problem
- Proxy metrics
- Watch cost and quality
- Agency headcount
- Ask what was measured
- Localized claims
- Meet ad standards
Volkswagen marketed "Clean Diesel" cars until a 2015 EPA violation notice, then settled US matters reported above $30 billion. Rank every claim by strength of evidence before release, and cut the ones you cannot show.
Launch day as the end of the work. Samsung's Galaxy Fold reached reviewers in April 2019, and several units failed at the fold. The launch moved to September 2019 with a reinforced display.
Name one owner for the first ninety days of feedback and set a review date before launch.
Announcing before the product exists. Apple announced its AirPower charging mat in September 2017 and canceled it in March 2019 without shipping a unit. Fyre Festival sold tickets from about $1,200 for a 2017 Bahamas weekend with no finished infrastructure on site; organizer Billy McFarland was sentenced to six years in 2018. Teams that pre-announce commit to a date before the engineering is done.
Set the announcement date only after a shippable unit passes its release test.
Naming a market before testing it. Google sold Glass to consumers in 2013 at $1,500 for the Explorer Edition, before rules for wearing a camera in public were settled. Consumer sales ended in 2015, and the product returned in 2017 as an Enterprise Edition.
Juicero priced a $699 press against $7 to $8 juice packs, then shut down in 2017 and refunded buyers after a demonstration that hands could squeeze the same packs.
Quibi raised $1.75 billion for short-form mobile video, opened in April 2020 and closed in October 2020. A segment chosen in a workshop, not in a pilot, sells to buyers whose problem the product only partly solves.
Put the first segment through a paid pilot before you name it in the launch plan.
Proxy metrics dressed as outcomes. MoviePass cut its monthly price to $9.95 in 2017 and passed 3 million subscribers, then shut down in September 2019 and filed for bankruptcy in January 2020. Snapchat sold about 150,000 pairs of Spectacles at $129 and reported an unsold inventory write-down near $40 million.
Signups, downloads and demo requests rise. Qualified adoption after release does not, so the dashboard looks healthy while the business is not.
Track qualified adoption after release on the same dashboard as downloads, and set a threshold that stops spend.
Agency selection that confuses headcount with capability. Peloton's 2019 holiday ad, made by the agency Mekanism, drew a backlash that reportedly cut roughly $1.5 billion from the company's market value within days. A large retainer buys more people, not better positioning. Ask what the agency has launched in your category and what it measured.
Claims that outrun proof across languages. In Canada, a second market for many US launches, the Competition Bureau announced a fine of about C$196.5 million against Volkswagen in December 2020 over misleading emissions advertising. A claim softened in English is often left intact in the localized version, where it may breach advertising standards in that market.
Check every localized page against the English source before launch, and log the reviewer and the date.
Correction is normal: Coca-Cola Classic returned in 79 days
Correction is normal. Coca-Cola replaced its formula in April 1985 and brought back Coca-Cola Classic within 79 days once the market answered.
Launch as public experiment
- Release date starts market learning clock
- Track which buyers convert
- Track which objections repeat
- Track which claims hold
- Treat early numbers as diagnostic evidence
- Expand only on repeating outcome
- Record open questions, not just wins
Treat early numbers as diagnostic evidence, not results. Expansion should depend on an outcome that repeats with the people, rights, capacity and budget you actually have.
Write down what is still unknown. A launch record that lists open questions is more useful than one that lists only wins.
Correction trace: from claim to evidence to asset
Correction trace
| Field | Test | What gets recorded |
|---|---|---|
| Case | Pick one claim from the launch page, such as a battery-life figure or a test method | Population, date, reviewer |
| Method | Follow the claim back to its evidence, then forward to every asset, deck and localized page that repeats it | Inputs, observations, unresolved limits |
| Precedent | Check how the same claim was handled in another market, as in the Clean Diesel case | Regulator, ruling date, penalty or undertaking |
| Outcome | Compare against qualified adoption after release, with cost and quality | Effect on adoption, plus cost and quality movement |
| Escalation | Stop if the case shows launch day treated as the end of the work | Correction owner, safeguard, next review date |
Before you raise spend: claims, ownership, adoption, records
- Every public claim is ranked by the evidence behind it, and unsupported ones are cut.
- A named owner holds the first ninety days of post-release feedback.
- Qualified adoption after release is tracked alongside cost and quality, not instead of them.
- One normal buyer path and one failure path have both been observed.
- Claims, inputs, versions and decisions are preserved with dates.
- A stop rule and a next review date are set before spend increases.
The GOV.UK technology selection guidance and the CISA software acquisition fact sheet inform the vendor and supply-chain checks in a launch record. The W3C information and relationships explanation and the NIST Privacy Framework starting guide cover page structure and privacy risk. None of them is legal clearance for a marketing use.
For launches into Canada, a second market for many US companies, PIPEDA, CASL and Quebec's Bill 96 each set their own requirements. Confirm your obligations with qualified counsel rather than a framework.
Common questions
What is the first decision in a launch?
Name the launch owner, the audience, the outcome you will measure, and the evidence that would stop the launch. That packet, dated and versioned, is what the rest of the work answers to.
How should a launch be reviewed?
Review qualified adoption after release against cost, quality and the exclusions you set. Keep the source limits, the failures and the dated decision record together, so a later reader can see what was known when.
When should a team correct a claim?
As soon as the evidence stops supporting it. Preserve the affected record, fix every asset that repeats the claim, including localized pages, and retest before closing. Volkswagen's Canadian penalty shows what a claim that outlives its evidence can cost.
Where do teams usually go wrong first?
They write the claim before they check the evidence. The launch page, the deck and the localized copy then repeat a number nobody can show.
Our product launch marketing overview covers the sequence that avoids this, and the common launch questions page handles the recurring ones.
For the wider method, see how to improve product launch marketing.
