Empathize–Create–Evangelize
The three-stage loop for developing a product vision: understand deeply, imagine the solved world, then sell it.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 92%
Atawodi's end-to-end process for producing a vision that actually lands. Most teams over-invest in the middle stage (create) and under-invest in the two that determine whether the vision survives: a rigorous, continuous understanding of the problem set, and a patient, multi-audience campaign to socialise the result. Vision work is measured in years, not sprints.
Origin
Ebi Atawodi's own synthesis, refined across Uber (Money/Commerce platform), Netflix, and YouTube (Studio / Creator experience), where she ran a 'Vision 2026' exercise that took roughly 18 months to reach the funding stage.
Core principles
- 01Clarity of the problem is the North Star; everything else moves around it.
- 02Empathy is earned by using the product (dogfood) and your competitors' products (catfood), not by outsourcing understanding to research.
- 03Research updates your mental model of the customer — it does not hand you the answer.
- 04If all the research could be dumped into an AI and it could spit out the PRD, you have not done your job.
- 05Evangelism takes patience — expect naysayers, and expect the vision to reach planning a year or more after you set it.
How to run it
- 1
Empathize — do 'understand work'
Deliberately experience the product as a user, including the flows you never touch. Keep multiple accounts, run through onboarding, and use adjacent and competitor products systematically. Build intuition by articulating what you love about the products you love.
Pro tip Dogfood (your product) plus catfood (competitors' and adjacent products) — Atawodi kept multiple accounts on YouTube, Instagram and TikTok, and used partner payment apps while at Uber.
Watch out Do not reflexively reach for 'let's go test it' — leaning on research to substitute for judgment is how PMs end up program-managing instead of product-managing.
- 2
Empathize — maintain a living problem list
Run the 'Top 10 Things You Should Know' living document (see the separate framework) so the problem set is always current, stack-ranked and shared across PM, engineering, design, research and data.
Pro tip Test the shared clarity by asking your engineering and design partners' reports for the top five problems — if the answers match, you've done your job.
- 3
Create — describe the world where the problems are solved
Take the consolidated problems and answer: if we solved these, what does the world look like? Express it via one of the three vision artefacts — the Once Upon a Time story, the future headline, or the visual mock.
Pro tip Block a no-meeting day and lock the core group in a room: Atawodi has produced a full vision in a single day of whiteboarding, then made it pretty afterwards.
- 4
Evangelize — run the concentric circles
Socialise the vision outward: core team, then stakeholders and adjacent teams, then leadership as high as you can reach. Write up the workshop output as a living, comment-open document.
Pro tip Because stakeholders contributed their own Top 10 in the insights stage, 'how did you arrive at this?' objections evaporate — humans need to know they were heard.
Watch out Many people give up here. The naysayers are inevitable; the vision only pays off if it keeps being repeated until it enters the planning cycle.
- 5
Hold the vision steady; add fidelity, don't rewrite
A vision should be evergreen for at least three years. Rather than producing a new vision each year, increase its fidelity over time — a sketch becomes a hi-fi mock becomes a sizzle reel.
Watch out If you have to redo the vision every year, you did not do the work in the first place.
In the wild
Atawodi set a vision at YouTube called 'Vision 2026'. She spent the following period evangelising it through Studio leads (the PM/eng/design triads), her PM weekly, stakeholders and leadership, keeping the write-up open for comments.
→ About a year and a half later, the vision reached the planning cycle and the big rocks that get there began to be funded; other teams now cite specific slides from it in their own plans.
When generative AI arrived, Atawodi challenged her team: what is the value you add that I could not get by putting all this research into an AI and asking it for the big insight?
→ Reframed the team's job around curated judgment and product sense rather than research relay.
Common mistakes
Skipping to 'create' without the empathize work
Atawodi says it baffles her how many PMs ramble when asked what problems keep them up at night. Without a crisp, visceral problem list, the vision has no anchor and nothing to be evangelised against.
Treating user research as the answer machine
Research should update your mental model of the customer, and that model informs what to build. PMs who believe they should have no opinions and only report what data says have abdicated the job.
Declaring the vision done once the deck is presented
Evangelism is the longest stage. Presenting once and moving on means the vision never percolates far enough to influence planning or funding.
Is it for you?
Best for
Product leaders owning a multi-year surface or platform who need a repeatable process for producing a vision that survives to funding.
Not ideal for
Short-horizon or micro-visions where a single mock of what next year's feature looks like is sufficient.
From the transcript
“one is what I call empathize the second is create so we spend a lot of time talking about create the middle piece and then…”
“he says you need to do understand work and what is understand work it's crazy to mean in the number of PMS who never go…”
“dog food meaning using your own product absolutely a must”
“if I could put all the research into Bard or chat GPT and it could spit out a PRD then you haven't done your job”
“I think if you have to repeat the vision every year you have not created a good”
From the episode
Crafting a compelling product vision
Ebi Atawodi (YouTube, Netflix, Uber)