Rebuilding trust after a confusing launch
The short versionAn AI app grew 40% on new features, then lost trust because nobody explained them well. I built a plan to change how the brand is seen: one story every team tells, real people behind the decisions, and targets to prove it's working.
- Context
- Luma, a fictional fast-growing AI productivity app for small businesses and freelancers.
- My role
- Concept work, done on my own: diagnosis, audience map, message architecture, PR plan and measurement.
- The challenge
- A confusing rollout left users worried about privacy, reporters calling the brand opaque, and teams telling different stories.
- The idea
- Close the gap between what Luma meant and how people saw it, starting inside the company.
- What I built
- An audience map, a three-part message architecture, a four-step plan (align, humanize, reset, explain), decisions on what not to do, and target KPIs set from day one.
The challenge
Luma launched new automation features and its user base grew 40%. The communication around those features didn't keep up.
- Tech reporters started describing Luma as "opaque," "rushed" and "invasive."
- Users read the new AI features as privacy risks.
- Sentiment dropped sharply in social listening.
- Inside the company, teams explained the same decisions in different ways.
The brand insight
The instinct in moments like this is to defend the product and correct people. That usually makes it worse, because it tells people their confusion is their fault.
The product wasn't the problem. The gap between what Luma meant and how people saw it was.
And the gap started inside. If four teams tell four versions of the story, the public hears noise. So the fix starts with the company, not the press release.
Key decisions
Each audience needed something different, so I mapped them first.
Power users
They run their businesses on Luma. They need clarity, predictability and a roadmap.
Early adopters
They shape opinion online. They need plain-language transparency and real experts.
New users
They joined during the spike. They need simple, steady reassurance.
Then one message architecture that PR, product, support and leadership all use. Every announcement answers the same three questions, in the same order:
Why
The user problem this solves, in the user's own words.
How
How Luma uses and protects data, with everyday examples.
Proof
What changed because users spoke up.
- Align: one story inside first.
Audit 18 months of communication for contradictions, run workshops with product, engineering, support and leadership, and roll out Why / How / Proof everywhere.
- Humanize: put faces on the decisions.
An engineer, a researcher and the privacy lead explain why features changed, in their own words, with media coaching first.
- Reset: rebuild relationships with reporters.
One-on-one briefings, plain-language explanations of AI decisions, and direct access to internal experts.
- Explain: Inside Luma.
A monthly long-form transparency series, so Luma explains itself before anyone else does it for them.
A defensive statement from the CEOIt would confirm the "we know best" image.Rolling back the featuresThe product was fine. The explanation wasn't.An ad campaign about privacyPaid reassurance reads as spin. Trust has to be shown, not bought.
How the brand shows up
A brand seen as hiding something can't fix that with a statement. It has to look and act open: real faces, plain words and a regular, honest place to explain itself.
What changed in our automations, and why you asked for it
Plain language, real names, and what we're still working on.
Inside Luma, editorial direction · concept
The people who built it
Photographed and quoted, instead of an anonymous company voice.
Editorial, not ads
Portraits, long-form layouts and white space. It reads like a magazine, not a defense.
Humble and specific
Less "here's why you're wrong," more "here's what we learned, and how we're improving."
How I'd measure it
Targets set at the start, so progress is visible and nobody argues about it later.
What I'd do next
- Build it into every launch. Why / How / Proof becomes the template for every future release, so this doesn't happen again.
- Check the gap every quarter. A short perception survey of users and reporters, compared with what we intend to say.
- Give support a voice in launches. Support hears confusion first. They should review messaging before anything ships.
What this shows about how I work
- Diagnosis first. I separated a perception problem from a product problem before planning anything.
- Systems, not slogans. The message architecture outlasts the crisis.
- Honesty earns trust. Access and plain answers instead of polish.