[🇧🇷] Altamente recomendo ler os artigos diretamente da fonte:
[🇬🇧] I highly recomend reading the articles straight form the source:
O Amodiovalerio Verde fez uma apresentação na PendomoniumX Munich 2025 e fez uma apresentação que me chamou a atenção por conta desse slide:
Também vale a pena ler:
Beyond Dashboards: Why Your Beautiful Dashboards Might Be Making You DumberTL;DR (for bullet-point enthusiasts)The Illusion of ClarityPrinciple 1: Avoid the Data DelusionTL;DR (for bullet-point lovers)Principle 2: Adopt a Data-Informed ApproachTL;DR (for people who believe reading full paragraphs is optional):Final ThoughtPrinciple 3: Choose What to MeasureTL;DR (for teams still adding metrics like it’s a hobby):Principle 4: Use Frameworks as Filters, Not BlueprintsTL;DR (For Leaders Scanning Before Their Next Meeting)Principle 5: Focus on Adoption, Not Just DeliveryTL;DR (For Leaders Reading This Between Two Strategy Calls)Principle 6: Know Your Tool Stack’s BoundariesTL;DR (for the confident scroller):Principle 7: Build Layered Dashboards to Scale ThinkingTL;DR (For Those Who Have a Board Meeting in Ten Minutes)Layer 1: The Outcome Layer (The Telescope)Layer 2: The Driver Layer (The Levers)Layer 3: The Deep Dive Layer (The Microscope)Principle 8: Manage Multi-Product Portfolios SeparatelyTL;DR (for the impatient but curious)1. Treat Each Product as Its Own System2. Separate Leading from Lagging Indicators3. Use Layered Dashboards Across the Portfolio4. Create Portfolio-Level Conversations, Not Just ReportsPrinciple 9: Reconcile Metric Definitions Before AnalysisTL;DR (for those whose last meeting was a debate about definitions):The Solution: A Metric Dictionary, the Operating System for ClarityHow to Get StartedPrinciple 10: Build Thinking Systems, Not Reporting SystemsPrinciple 11: Turn AI into a Judgment Multiplier
Beyond Dashboards: Why Your Beautiful Dashboards Might Be Making You Dumber
TL;DR (for bullet-point enthusiasts)
- Dashboards are not decisions.
- AI won’t replace judgment – it exposes it.
- This is the intro to an 11-principle series on data, decisions, and AI.
- Yes, it’s long. Clarity is worth your time.
- Slides coming later for your team debates.
The Illusion of Clarity
- We track everything and decide nothing.
- Features shipped, no one used them.
- Metrics moved, no one knew why.
- Roadmaps full. Strategy hollow.
What’s coming next?
Here are the 11 principles we’ll explore:
- Avoid the Data Delusion Statistically significant. Strategically irrelevant. That’s the trap.
- Adopt a Data-Informed Approach Being data-driven is like staring into the fridge hoping a meal appears.
- Choose What to Measure A metric without a decision is expensive noise.
- Use Frameworks as Filters, Not Blueprints Frameworks don’t decide for you – they stop you from staring into the void.
- Focus on Adoption, Not Just Delivery Shipping is a cost. Adoption is the asset.
- Know Your Tool Stack’s Boundaries You don’t have one truth. You have a stack of partial truths.
- Build Layered Dashboards to Scale Thinking One-size-fits-all dashboards fit no one. Especially your executives.
- Manage Multi-Product Portfolios Separately Blended metrics create Franken-metrics. Useful to no one.
- Reconcile Metric Definitions Before Analysis If teams argue about numbers, they’re arguing about definitions.
- Build Thinking Systems, Not Reporting Systems Dashboards aren’t the goal. Better decisions are.
- Turn AI into a Judgment Multiplier AI multiplies judgment. Without judgment, there’s nothing to multiply.
Principle 1: Avoid the Data Delusion
TL;DR (for bullet-point lovers)
- Avoid mistaking busywork for meaningful progress by questioning if data is creating an illusion of clarity.
- Data without human judgment is just noise; it should be an input for strategy, not a substitute for it.
- Use AI to sharpen your questions and find real problems, not to automate trivial tasks that accelerate waste.
- Focus on whether experiments fundamentally change your direction, not on small, strategically irrelevant wins.
- Shift your team from celebrating data to making decisions by asking what you will do differently with the information.
The real danger is that AI is exceptionally good at making motion look like progress.
…
However, when used with intent, AI becomes a powerful tool for augmenting judgment, not replacing it.
Principle 2: Adopt a Data-Informed Approach
TL;DR (for people who believe reading full paragraphs is optional):
- Being “data-driven” is a trap. It builds passive teams who wait for numbers to give them permission to think.
- Data-informed teams lead with hypotheses, use data to pressure-test thinking, and leave judgment where it belongs: with humans.
- “What does the data say?” is the wrong question. Start with: “What are we trying to learn?”
- AI doesn’t have opinions. If you don’t have a hypothesis, AI won’t help you; it will overwhelm you.
- Shift your mindset: data is not the answer. It’s the sparring partner. You’re the one supposed to think.
Final Thought
- “Data-driven” teams look busy.
- “Data-informed” teams make decisions.
- Dashboards track history. Judgment shapes it.
Principle 3: Choose What to Measure
TL;DR (for teams still adding metrics like it’s a hobby):
- Every metric has a cost. Not money, but something worse: attention. Metrics consume focus, fuel debate, and create cognitive load.
- Track only what informs decisions. Interesting numbers don’t drive action. Vanity metrics waste leadership energy.
- AI will scale whatever signals you feed it. Garbage in? Smarter-looking garbage out. Choose signals that matter.
- Think cockpit, not buffet. Dashboards should steer, not decorate. Track fewer, sharper, decision-driving metrics.
- Use the 6-question checklist before adding any metric. Every number must earn its place.
We measure what’s easy, not what’s useful.
[Me lembrou o princípio ”What gets measured, gets managed”]
Before adding a metric, force this conversation:
- What strategic goal does this support? If unclear, it doesn’t belong.
- What decision will this inform? No decision? Remove.
- What action will we take if this changes? If the answer is "nothing," stop tracking.
- What behavior does tracking this reinforce? Metrics shape incentives. Careful what you count.
- What are we stopping to make space for this? Adding without subtracting is building a landfill.
- Who owns this metric? No owner? No accountability. No point.
Dashboards aren’t reports. They’re steering wheels.
To avoid vanity metrics:
- Start with intent.
- Define the signal.
- Track only what informs action.
Dashboards don’t exist to display data. They exist to help you decide.
Principle 4: Use Frameworks as Filters, Not Blueprints
TL;DR (For Leaders Scanning Before Their Next Meeting)
- Frameworks don’t make decisions. They focus attention.
- Used well, they sharpen clarity. Used poorly, they paralyze teams.
- AI doesn’t solve the problem. It multiplies it, generating frameworks without judgment or context.
- Leadership isn’t about choosing the cleverest model. It’s about enforcing discipline.
- One decision. One primary framework. To enforce clarity, start with one primary framework per decision. The goal is to choose a single, dominant lens for any given problem.
- Stacking models creates noise. Choosing the right lens for the right problem creates clarity.
- Frameworks are useful. Leadership is mandatory.
A framework’s job is simple: Focus attention, highlight signals, and provide a temporary lens for the conversation.
Quick refresher:
- AARRR? Great for growth loops. Useless in understanding user motivations.
- HEART? Good for UX monitoring. Tells you nothing about business impact.
- OKRs? Align execution. But doesn’t account for core product health.
- North Star Metrics? Focus attention, but can focus you on the wrong thing.
- JTBD? Helps you understand needs, but offers no prioritization.
Are we optimizing for clarity or for complexity?
Frameworks don’t prevent bad decisions. They just make bad decisions look methodical.
Framework overload is often a sign of missing focus from leadership.
Someone must choose focus. That’s leadership.
Leadership Checklist: How to Use Frameworks Properly
- Choose One Primary Framework Per Decision. For any single objective, select the one framework that frames the problem best.
- Declare the Boundaries. What does this framework ignore? Make it explicit.
- Name the Decision. What choice is this framework helping to make?
- Challenge the Fit. Why this framework? Why now? Default to rejecting it.
- Lead. Frameworks focus. Leaders decide.
Your customers don’t care what framework you used. They care what you delivered.
Principle 5: Focus on Adoption, Not Just Delivery
TL;DR (For Leaders Reading This Between Two Strategy Calls)
- Shipping is overhead. Adoption is the asset.
- In B2B SaaS, removing features is rarely easily feasible. Prevention might be your only scalable strategy.
- In B2C SaaS, unused features lead to silent churn. Users remove themselves.
- AI won’t tell you what success looks like. It can help surface adoption signals but not define value for your customers.
- Your product is a system for driving outcomes, not a catalogue of releases.
- Shift the conversation from: “What did we ship?” to “What is delivering value?”
The backlog is full. The roadmap is full. Velocity is high. Features are shipping.
And yet… leadership starts asking questions nobody can answer:
- “Are customers using the last three features we shipped?”
- “Which features generate the most value?”
- “What about the ones we built last year?”
Because every feature your teams ship, that your customers don’t use, adds silent operational cost:
- Support tickets (“How do I use this thing?”)
- Training materials (“Here’s how to ignore that setting.”)
- UX clutter (confusing interfaces that hurt adoption of valuable features)
- Maintenance burden (keeping code alive just because someone, somewhere, might use it)
A feature can be technically correct but strategically irrelevant.
Product teams need to think like portfolio managers, not feature brokers.
- What features generate actual value?
- Which ones degrade UX clarity?
- Which features are silent liabilities?
Ask your teams: “Of what we’ve already shipped, what isn’t delivering value? And why?”
That’s where leadership begins.
So the challenge is simple, but not easy:
- Declare adoption as a strategic metric.
- Track it relentlessly.
- Treat non-adoption as debt.
- Prevent before you need to remove.
- Let AI scale your observation, but never your strategy.
If you’re not tracking adoption after launch, you’re not managing a product. You’re managing a feature factory.
Final Reflection: Product is a System. Adoption is Proof.
Your product isn’t a collection of features. It’s a system designed to drive outcomes. Every feature either strengthens that system… or weakens it. Every shipped feature is a strategic decision. Every adopted feature is a strategic victory. Everything else? It’s just overhead.
Principle 6: Know Your Tool Stack’s Boundaries
TL;DR (for the confident scroller):
- Your operational data lives in specialized tools. That’s a reality, not a failure.
- The goal isn’t one dashboard to rule them all. It’s a coordinated system where each tool plays its role.
- Every tool has a purpose and blind spots. Your CRM knows the deal, your analytics knows the click, but neither knows the whole story.
- AI can find patterns across tools, but only if you teach it the boundaries of each data source first.
- Stop duct-taping dashboards together. Start building a system of federated clarity.
You wanted answers. You got complexity.
The myth isn’t that there should be one source of truth.
The myth is that there’s one truth.
… the visual symptom of a deeper issue: a lack of agreement on metric definitions and authoritative sources.
A Quick Story (Because Metaphors Are Sticky)
[🇧🇷] Muito boa história. Vá ler no artigo original.
[🇬🇧] Really good. go to his article to read!
Principle 7: Build Layered Dashboards to Scale Thinking
TL;DR (For Those Who Have a Board Meeting in Ten Minutes)
- One-size-fits-all dashboards are a myth. Trying to serve executives, team leads, and analysts with a single view creates noise for everyone and clarity for no one.
- Build in three layers: Structure your reporting to match how decisions are made.
- The Outcome Layer for executives (telescope),
- the Driver Layer for product teams (levers),
- and the Deep Dive Layer for analysts (microscope).
- AI enriches layers, it doesn’t flatten them. Use AI to summarize deep-dive data for the outcome layer or to flag anomalies in the driver layer. But without clean, structured layers, AI just automates the confusion.
- Your goal is to scale thinking, not just reporting. A good dashboard system provides the right altitude of information to the right person, enabling better, faster decisions at every level.
This isn’t a design problem. It’s a failure to recognize that to make decisions.
- Executives need a telescope. They need to see the destination and know if they’re on course.
- Team leads need levers. They need to see the inputs that are driving the executive-level outcomes.
- Analysts need a microscope. They need to get into the raw data to diagnose why the levers are moving.
Layer 1: The Outcome Layer (The Telescope)
- Audience: Executives, the Board, C-Suite.
- The Question It Answers: "Are we winning or losing?"
- What It Shows: One, maybe two, headline metrics that represent overall business health. Think Net Revenue Retention, Total Active Accounts, or a unified Customer Health Index.
- Purpose: This layer is for a 30-second assessment of performance. There is no drill-down. It’s a statement, not a conversation.
Layer 2: The Driver Layer (The Levers)
- Audience: Product Leadership, Team Leads, GTM Leaders.
- The Question It Answers: "Why are we winning or losing?"
- What It Shows: The 3-5 key input metrics that directly influence the Layer 1 outcome. If the outcome is Active Accounts, the drivers are metrics like New User Activation Rate, Usage Frequency of core features, and Account Churn Rate.
- Purpose: This is the dashboard where strategy is debated and resources are allocated.
Layer 3: The Deep Dive Layer (The Microscope)
- Audience: Product Managers, Analysts, Engineers, Designers.
- The Question It Answers: "Where exactly is the problem or opportunity?"
- What It Shows: Flexible, granular views of the data.
- Purpose: This layer allows for segmentation by user, region, feature, or any other relevant dimension. It’s a workspace for investigation, not a report for presentation.
Your organization doesn’t need more charts. It needs more clarity.
Stop building dashboards that try to be everything to everyone. Start building a system that helps everyone think better.
Principle 8: Manage Multi-Product Portfolios Separately
TL;DR (for the impatient but curious)
- In many enterprise SaaS portfolios, blended metrics across multiple products create what I call Franken-metrics: stitched together, hard to interpret, and rarely actionable.
- Managing a suite of products with a single metric set is like trying to steer a fleet of ships using one compass.
- AI can make the problem worse by creating beautifully wrong roll-ups, or better if you feed it the right distinctions.
- Treat each product as its own system: with separate health metrics, adoption curves, and decision logic.
- Portfolio-level clarity comes from synthesis, not aggregation.
These forces drive it:
- Portfolio Pressure: Leaders often seek a simple, unified story, a neat slide with arrows going up. Yet simplicity at the top can unintentionally create distortion at the bottom.
- Shared OKRs: When goals are defined at the portfolio level without nuance, teams optimize for the metric instead of the product reality. If “total DAU” is the target, everyone pushes for logins, even if value delivered per product is wildly uneven.
- Tool Limitations Analytics stacks often roll up numbers by default. That cohort churn dashboard? It’s aggregating across products unless you explicitly slice. Your BI tool wasn’t designed to scream: “Stop blending apples and oranges.”
- AI Hype Ironically, the more sophisticated our analytics become, the more tempting it is to produce single, impressive-looking charts that bury nuance in layers of machine learning-powered smoothing.
If blended portfolio dashboards create more delusion than clarity, what’s the alternative? You don’t need more dashboards. You need better thinking systems for portfolios.
Here are four practical shifts to make now.
1. Treat Each Product as Its Own System
Every product deserves:
- Its own success metrics (adoption, engagement, retention, satisfaction).
- Its own health signals (support volume, NPS, churn risk).
- Its own ROI view (investment vs. outcome).
Think of it like managing a sports team. You don’t average every player’s performance into one number and call it a day. You track each role. A striker’s value isn’t measured the same way as a goalkeeper’s.
2. Separate Leading from Lagging Indicators
Portfolio dashboards often skew toward lagging indicators (ARR, churn, gross margin). By the time those numbers move, it’s too late.
Instead:
- Pair each lagging metric with a leading one.
- For adoption: pair DAU with “time-to-first-value.”
- For churn: pair renewal rate with “support response time” or “feature stickiness.”
3. Use Layered Dashboards Across the Portfolio
This builds on Principle 7. Executives need a telescope. Teams need a microscope.
- Team-level: Feature adoption, task completion, bug reports.
- Product-level: Retention curves, revenue contribution, NPS by segment.
- Portfolio-level: A synthesis, not an average, showing which products are carrying, lagging, or at risk.
Stop pretending one dashboard can do all three jobs. It can’t.
4. Create Portfolio-Level Conversations, Not Just Reports
Dashboards don’t make decisions. People do. Instead of presenting a green portfolio slide once a quarter, create a recurring Portfolio Review Rhythm:
- Monthly product health reviews: each product lead presents their own key signals.
- Quarterly synthesis sessions: leaders align on cross-product dependencies, risks, and capital allocation.
- Explicit discussion of trade-offs: “We will invest more in Product B even though Product A is carrying revenue right now, because we need B to grow for long-term resilience.”
Blended numbers can’t have that conversation. Humans must.
To test whether your portfolio reporting is driving clarity instead of camouflage, ask your teams three hard questions:
- Can each product team explain their performance without noise from other products? If not, you’re already in Franken-metric territory.
- Can you describe how users move between your product and others in the portfolio? Understanding these flows is critical to seeing where adoption lifts or stalls across the suite.
- Is our portfolio reporting built for true clarity or for convenience? Dashboards that make life easier for executives but obscure the truth are liabilities, not assets.
Principle 9: Reconcile Metric Definitions Before Analysis
TL;DR (for those whose last meeting was a debate about definitions):
- If teams argue about numbers, it’s usually not math they’re debating but definitions. Vague definitions aren't a data issue; they're a systems failure.
- Inconsistent metric definitions (like “Active User”) create conflicting truths, erode trust, and stall decisions. Your MAU might be four metrics in a trench coat pretending to be one.
- AI usually makes it worse, unless you’ve already solved the definition problem upstream. Feed it inconsistent data and you’ll get confident-sounding nonsense at scale.
- What you need is a Metric Dictionary, one source of truth with each metric's name, source, formula, and owner.
The Most Expensive Meeting in Your Company
It’s not the meeting with the highest-paid people that costs the most; it’s the one where smart, well-intentioned people waste an hour arguing about the meaning of a number because it was never defined.
You can’t build strategy on a number that means different things to different teams. This isn’t a data problem, but a leadership problem disguised as a spreadsheet debate.
The real failure is allowing a preventable problem, imprecise language, to undermine your most critical business processes.
The Solution: A Metric Dictionary, the Operating System for Clarity
Here’s how it fits:
- Metrics are raw measures of performance.
- KPIs are the critical subset used to track business health.
- OKRs are strategic goals often tied to one or more KPIs.
So, what goes into a robust Metric Dictionary?
- Precise Name: Not a vague term like "Engagement," but something unambiguous like "Weekly Core Action Completion Rate". The name itself should communicate the meaning.
- The Owner: The single authority accountable for accuracy and evolution, often a functional leader (e.g., VP Product) or Product Operations. A metric without an owner is an orphan, and no one will trust it.
- The Data Source: Where does the raw data come from? Which system is the record of truth? (e.g., BI warehouse, analytics tool, CRM). Be explicit.
- The Formula: The exact calculation, written with software‑level precision. Include all inclusions, exclusions, and time windows. Example: A user who completes at least one core action (X, Y, or Z) in a rolling 7‑day period. Excludes internal users and suspended accounts.
Clarity in metrics is not a technical initiative, it is an act of leadership.
How to Get Started
If you’re building a Metric Dictionary from scratch, keep it simple:
- Identify your top 10 crown jewel metrics used in board decks, executive reviews, or funding conversations.
- Assign a single accountable owner for each.
- Define the precise name, source, and formula.
- Publish in a visible, shared space (e.g., Confluence, Wiki, Notion, or your BI tool).
- Make it a living document. Revisit definitions quarterly as your strategy evolves.
Think of it as a garden: it requires constant care.
- Weeding: Rogue metrics will sprout in unaudited dashboards and siloed reports. Governance, manual or AI‑assisted, means finding and removing these before they choke out the shared language.
- Pruning: As strategy evolves, some metrics lose relevance. A clear process to deprecate and archive them prevents the dictionary from becoming a graveyard of forgotten KPIs.
- Sunsetting: Outdated KPIs left in dashboards create confusion and erode trust. Retire them with a defined process: identify, explain the rationale, document the history, and remove them from live use.
Agreement isn’t always required, but shared definitions are.
Principle 10: Build Thinking Systems, Not Reporting Systems
Principle 11: Turn AI into a Judgment Multiplier
—
Here are the key takeaways from the article “Beyond Dashboards: Why Your Beautiful Dashboards Might Be Making You Dumber” as found on your current Notion page:
- Dashboards Are Not Decisions: Dashboards provide information, but judgment and decisive action come from humans, not from dashboards or AI. Simply tracking metrics isn't the same as making smart decisions.
- The Illusion of Clarity: Data and dashboards can give a false sense of understanding and progress. Many teams track everything but don’t act meaningfully on what they learn—features are shipped without adoption, and metrics move without anyone knowing why.
- Principle 1: Avoid the Data Delusion: Don't mistake data collection and busywork for real progress. Data should be a tool for strategic thinking and decision-making, not an end in itself. Use AI to help find important questions, not to automate trivial tasks.
- Principle 2: Adopt a Data-Informed Approach: Being purely “data-driven” can make teams passive—waiting for the data to “tell them” what to do. Instead, use data to challenge and refine existing hypotheses, but keep human judgment at the center.
- Principle 3: Choose What to Measure — Less Is More: Every metric you track consumes attention. Only track what will inform decisions and actions. Vanity metrics and unnecessary numbers distract and drain leadership energy.
- Principle 4: Frameworks as Filters, Not Blueprints: Frameworks (like OKRs, AARRR, HEART, etc.) are tools to focus your attention, not to provide rote solutions for every problem. Overreliance on frameworks can cause confusion; leadership is about choosing what matters.
- Principle 5: Focus on Adoption, Not Just Delivery: Shipping features isn’t the end goal—adoption is. Unused features add operational costs and can degrade the user experience. Success should be defined by what delivers value to users.
- Principle 6: Know Your Tool Stack’s Boundaries: No single tool can provide the full picture. Each tool (CRM, analytics, etc.) has its place and limits. Don’t try to create one “dashboard to rule them all”; instead, build an integrated system and be clear about what each source covers.
- Principle 7: Layered Dashboards for Layered Thinking:
- Outcome Layer (Telescope): For executives, providing a quick, top-level health check (e.g., Net Revenue Retention).
- Driver Layer (Levers): For product leads/teams, to identify which levers or input metrics drive outcomes.
- Deep Dive Layer (Microscope): For analysts, enabling in-depth exploration and diagnosis of underlying issues.
- Principle 8: Manage Multi-Product Portfolios Separately: Don’t blend metrics from different products. Treat each as its own system with distinct health and adoption metrics to avoid “Franken-metrics” that provide little value.
- Adoption Is the Proof, Not Shipping: Treat adoption as a strategic metric. Track it, learn from it, and understand that non-adoption is a form of debt.
- AI Multiplies, It Doesn’t Substitute Judgment: AI can help scale analysis and find patterns but cannot replace the human context, intent, and judgment needed for meaningful decisions.
- Framework and Metric Clarity Is Leadership: Frameworks, tools, and data are only as good as their application. Leadership means making thoughtful choices about focus, what you measure, and what you act upon.
Overall Message: Build dashboard systems to support better thinking—not just prettier reporting. Focus on actionable insights, clarity, adoption, and judgment rather than an overload of metrics or frameworks. Your product is a system designed to drive outcomes, and every metric, feature, or tool should reinforce that system’s clarity and value.