Annual Day Deck Guide
Deck: annual-day-high-performance-culture.pptx
Theme: From Average Team to High-Performance Product Team.
Core message: We are a small company, and that can be an advantage if we combine customer closeness, AI leverage, engineering discipline, quality, security, performance, and fast learning.
Flow
| Slides | Purpose |
|---|---|
| 1-5 | Establish urgency: small teams can now build serious products; software/hardware/AI trends are accelerating. |
| 6-8 | Name the average-team trap and introduce the high-performance flywheel. |
| 9-11 | Build the right product: product mindset, design thinking, customer problem, outcome, experiment. |
| 12-16 | Build it the right way: quality, fast releases, DORA, DevSecOps, team-owned quality. |
| 17-20 | Team lead loop, OKRs, minimum process, AI as amplifier. |
| 21-23 | Everyone's role and senior developer participation. |
| 24-27 | Interactive game, reflection questions, commitment, closing quotes. |
Senior Developer Participation
Use three senior developers as "field voices" rather than long lecturers.
Senior Developer 1: Good Engineering In A Small Company
Timing: 5-7 minutes after slide 12.
Talking angle:
- Small PRs are not bureaucracy; they are how we reduce risk.
- CI is the team's shared safety net.
- Code review is knowledge sharing, not permission seeking.
- Simple design wins because future us is also a customer.
Suggested prompt:
Tell us one practice that makes your daily development faster and safer, and one practice we should stop normalizing.
Senior Developer 2: Quality Is Not QA's Job Alone
Timing: 5-7 minutes after slide 16.
Talking angle:
- QA should help the team think about risk earlier.
- Developers own testability.
- Product/stakeholders own clarity.
- Operations owns signals from real usage and incidents.
- Performance and observability must be designed before production pain.
Suggested prompt:
Share one example where quality would have improved if we had discussed risk earlier.
Senior Developer 3: AI Helps, But Understanding Still Matters
Timing: 5-7 minutes after slide 20.
Talking angle:
- AI is useful for exploration, tests, refactoring ideas, documentation, and debugging.
- AI-generated code still needs ownership, review, tests, and security thinking.
- The rule: do not submit code you cannot explain.
Suggested prompt:
Show one practical AI use that saves time, and one risk we must guard against.
Exercises
Exercise 1: Request To Outcome
Slide: 11.
Duration: 8 minutes.
Setup:
- Split participants into mixed groups.
- Give the request: "Build a dashboard for managers."
- Ask teams to convert it into user, problem, outcome, metric, and smallest experiment.
Good answer:
Managers struggle to identify staffing risks before weekly planning. We believe a focused capacity dashboard can reduce planning time from 2 hours to 30 minutes for pilot managers. Before building the full dashboard, we will prototype the top three decisions managers need to make and test it with five managers.
Debrief question:
What changed when we started with the user problem instead of the feature?
Exercise 2: Product Incident Game
Slide: 24.
Duration: 12 minutes.
Scenario:
We shipped the requested feature. Users are confused. Performance is slow. Sales demo in 2 hours.
Roles:
- Product/stakeholder: Was the problem validated?
- Design/UX: Is the workflow understandable?
- Developer: What technical complexity or shortcut may be involved?
- QA: What risk was missed?
- IT/Ops: Can we monitor, rollback, or stabilize?
- Security: Is there any trust or data issue?
Good answer:
First, reduce customer impact. If needed, use a feature flag or rollback. Second, communicate clearly to sales/support. Third, review whether the feature solved the right problem. Fourth, add performance, usability, and risk checks to our definition of done. Finally, convert the incident into one product learning and one engineering improvement.
Debrief question:
If this happened in our real team, where would the delay or confusion come from?
Commitment Exercise
Slide: 25.
Duration: 10 minutes.
Each team writes:
- One behavior to start.
- One behavior to stop.
- One metric to watch.
- One improvement experiment.
- One owner.
- Review date: July 25.
Good examples:
- Start: Every feature begins with problem, user, outcome, and metric.
- Stop: Merging large PRs without tests or review context.
- Measure: Lead time for changes and escaped defects.
- Experiment: 30 days of small PRs plus daily CI health check.
Planned Humor
Use lightly, not as stand-up comedy.
- "Closing tickets is not the same as opening customer value."
- "A meeting without a decision, learning, or coordination purpose is just a calendar-shaped snack."
- "Security at the end is like checking the seatbelt after the accident."
- "QA should not be the department of late surprises."
- "Motivation alone has a short battery life."
- "Minimum process does not mean process diet where we remove vitamins and keep sugar."
Serious Lines
- "A small team can compete with a large company when it learns faster than the large company can coordinate."
- "AI will not save a weak engineering culture. It will expose it faster."
- "Quality, security, and performance are not separate departments. They are part of how we keep promises."
- "Team leads should not manage noise. They should manage the learning loop."
- "The goal is not to become process-heavy. The goal is to make the right behaviors repeatable."
Quotes And Proverbs
These are intentionally short for slide use.
- "Measure twice, cut once." Old proverb.
- "The best way to predict the future is to invent it." Commonly attributed to Alan Kay.
- "Trust, but verify." Proverbial management wisdom.
- "Slow is smooth. Smooth is fast." Training proverb.
Fact Sources
- DORA 2025, State of AI-assisted Software Development: https://dora.dev/dora-report-2025/
- DORA home and software delivery research: https://dora.dev/
- Y Combinator Requests for Startups: https://www.ycombinator.com/rfs
- Menlo Ventures, 2025 State of Generative AI in the Enterprise: https://menlovc.com/perspective/2025-the-state-of-generative-ai-in-the-enterprise/
- Gartner Top Strategic Technology Trends for 2026: https://www.gartner.com/en/articles/top-technology-trends-2026
- SVPG, The Foundation of Product: https://www.svpg.com/the-foundation-of-product/
- SVPG, Empowered Product Teams: https://www.svpg.com/empowered-product-teams/
- SVPG, Transformed: https://www.svpg.com/books/transformed-moving-to-the-product-operating-model/
- IDEO Design Thinking: https://designthinking.ideo.com/introduction
- Nielsen Norman Group UX articles: https://www.nngroup.com/articles/
- NIST Secure Software Development Framework: https://csrc.nist.gov/pubs/sp/800/218/final
- OWASP Top 10 for LLM Applications: https://owasp.org/www-project-top-10-for-large-language-model-applications
- AWS DevOps guidance on frequent production deployments: https://docs.aws.amazon.com/wellarchitected/latest/devops-guidance/dl.cd.1-deploy-changes-to-production-frequently.html
- Netflix TechBlog, Spinnaker and continuous delivery: https://techblog.netflix.com/2015/11/global-continuous-delivery-with.html
- Netflix Spinnaker report note: https://netflixtechblog.com/multi-cloud-continuous-delivery-with-spinnaker-report-now-available-6040ba83b765
- Meta Engineering, rapid release at massive scale: https://engineering.fb.com/2017/08/31/web/rapid-release-at-massive-scale/
- Thoughtworks, The Case for Continuous Delivery: https://www.thoughtworks.com/insights/blog/case-continuous-delivery
- Flickr/Velocity 09, 10+ Deploys Per Day: https://www.youtube.com/watch?v=LdOe18KhtT4
Design Notes
- The deck uses original diagrams and visual metaphors rather than copied external graphics.
- External claims are represented as short cited facts, not screenshots.
- Text is intentionally minimal so the live speech can carry the story.
- The generated deck is editable because it is built with PowerPoint shapes, not exported images.