TL;DR: Product Analytics Setup and Interpretation for startup growth
Product Analytics Setup and Interpretation helps you see what users actually do, so you can fix drop-off, find what leads to retention, and make faster product decisions without guessing.
• Start small: track the few user actions that prove first value, then measure funnel drop-off, time to value, retention, and paid conversion.
• Set clean event names, user properties, identity rules, and simple dashboards first; messy tracking creates false confidence.
• Read charts with context: segment users, check releases or campaigns, and pair behavior data with interviews or session review. A solid product analytics guide and these product analytics examples show how this works in real teams.
• For early-stage founders, the goal is not more dashboards. It is clearer choices about what to fix, build, or cut based on real usage patterns.
If you want fewer wasted sprints and better retention, use this article as your 30-day plan and set up your first clean tracking system now.
Check out startup news that you might like:
Local SEO News | June, 2026 (STARTUP EDITION)
Product Analytics Setup and Interpretation is the process of tracking what users actually do inside your product, then turning that behavior into decisions about growth, retention, and product direction. For startups, it is how you replace founder guesswork with observable patterns before you waste months building the wrong thing.
Why this matters is simple. A bootstrapped founder does not get many expensive mistakes. I say this as Violetta Bonenkamp, a European founder who has built across deeptech, edtech, no-code systems, and AI tooling. When you run lean, every feature, every experiment, and every onboarding step must earn its place. Product analytics helps you see where users stall, what they repeat, and what actually leads to activation, retention, and revenue.
Why the topic is important for startups: startups live under uncertainty, and uncertainty gets expensive fast. Unlike surface-level reporting from ad platforms or vanity totals from app stores, product analytics lets you inspect in-product behavior at the event, user, funnel, and cohort level, which makes it especially useful when you are still trying to prove demand, tighten onboarding, and shape the product.
Key Takeaway
- How Product Analytics Setup and Interpretation affects startup growth, retention, and decision speed
- How to set up event tracking, user properties, funnels, cohorts, and dashboards without turning your stack into chaos
- Which founder mistakes create bad data and false confidence
- Which practical frameworks lean teams can use in 2026
Why does product analytics matter so much for startups right now?
The hard truth is that many founders still confuse traffic with traction. They celebrate visits, signups, downloads, or social buzz while the product itself quietly leaks users. The startup problem is rarely “nobody heard of us.” More often, it is “people came once and never formed a habit.” Product analytics exists to expose that gap.
Recent reporting around digital visibility also shows something useful for founders. A 2026 study covered by Business Insider on AI citation versus Google ranking argued that visibility now depends on corroborated signals across sources, not just one channel. The startup lesson is bigger than search. If your internal product data, your user interviews, your reviews, and your sales calls do not tell the same story, your decision quality drops. Consensus beats isolated vanity signals.
Here is why. Product analytics helps answer the questions that actually move a business:
- Which action predicts retention?
- Which user segment activates fastest?
- Which onboarding step causes abandonment?
- Which feature is noisy but unimportant?
- Which behavior shows product-market pull instead of curiosity?
And yes, startups with limited cash benefit the most. When I built systems in no-code environments and game-based learning products, the point was never to collect more data for the sake of data. The point was to create behavioral evidence. A founder should treat a startup like a strategic game. Not a game of vibes. A game of signals.
This is also where product analytics connects naturally with product-market fit. You cannot claim product-market fit because a few users said nice things. You need repeated behavior, retained cohorts, and clear usage patterns from the right segment.
What is product analytics setup and interpretation, exactly?
Let’s define terms clearly so there is no confusion.
Core concept #1: Event tracking
Definition: An event is a recorded user action inside your product. It can be “Signed Up,” “Created Project,” “Uploaded File,” “Completed Lesson,” or “Invited Teammate.”
Why it matters for startups: events turn product usage into measurable behavior. Without event tracking, you are left with assumptions, anecdotal chat messages, and founder memory.
Real-world example: in an educational product, page views tell you almost nothing about learning progress. A better event model would track “Started Quest,” “Finished Quest,” “Asked Mentor Question,” and “Returned Within 7 Days.” Those actions are much closer to habit formation.
Related terms: event schema, naming convention, property, timestamp, anonymous user, identified user.
Core concept #2: Funnel analysis
Definition: A funnel shows how many users move through a sequence of actions, such as Visit Landing Page → Sign Up → Verify Email → Create First Project → Invite Team.
Why it matters for startups: a startup often dies from leakage between steps, not from lack of ambition. Funnels show where the drop happens.
Real-world example: if 40 percent of new users create an account but only 8 percent connect their first data source, your issue is not acquisition. It is activation friction.
Related terms: conversion rate, activation, abandonment, step completion, time-to-convert.
Core concept #3: Cohorts and retention
Definition: A cohort is a group of users who share a characteristic, often signup week, acquisition source, plan type, or first use case. Retention shows whether those groups keep coming back.
Why it matters for startups: averages hide the truth. One strong cohort can mask five weak ones. Cohort analysis tells you whether the product is truly getting better over time.
Real-world example: imagine paid users from founder-led demos return weekly, but self-serve users disappear after day three. That tells you onboarding and product education need work before you spend more on acquisition.
Related terms: retention curve, stickiness, returning users, churn, cohort comparison.
Core concept #4: Interpretation
Definition: Interpretation means turning product data into an explanation and a decision. Not just reading charts, but understanding what changed, why, and what should happen next.
Why it matters for startups: tracking without interpretation creates dashboard theater. Teams stare at charts, say interesting things, and do nothing useful.
Real-world example: if a funnel improves after a UI change, the explanation might not be the new design. It could be better traffic quality, a promo campaign, or a bug fix. Interpretation means checking context before declaring victory.
Related terms: causation, correlation, experiment, segment analysis, qualitative context.
Which metrics should a startup track first?
Most startups track too much too early. That creates noise, engineering burden, and false urgency. Your first analytics layer should answer one simple question: are users reaching value fast enough to come back?
Start with these foundational metrics:
- Activation rate: the share of new users who complete the first meaningful action
- Time to value: how long it takes a new user to reach that first meaningful action
- Day 1, Day 7, Day 30 retention: whether users come back
- Feature adoption: whether high-value features are actually used
- Conversion to paid: for freemium or trial models
- Expansion behavior: invites, additional seats, extra projects, repeat usage
- Churn indicators: inactivity, failed setup, abandoned onboarding, support-heavy segments
Next steps. If your product has a sales-assisted motion, pair product data with pipeline numbers. That is where a sales metrics dashboard becomes useful. Product and sales should not live in separate realities.
How do you set up product analytics step by step?
This section is the practical part. Keep it lean. A good setup is clear, consistent, and boring. Boring is good. Boring data is trustworthy.
Phase 1: Assessment and planning
Step 1.1: Audit your current state
- List every tool currently sending product data
- Check whether marketing analytics and product analytics are mixed together badly
- Review which events already exist and whether they are named consistently
- Identify duplicate events, missing timestamps, and broken user IDs
- Map where consent, privacy, and data storage are handled
If you are very early, the audit may reveal almost nothing. That is fine. Starting from zero is often easier than fixing months of random tracking.
Step 1.2: Define your analytics questions before your tools
Founders love asking which tool to buy. Wrong first question. Start with decision questions.
- What behavior means a new user got value?
- Which actions predict retention?
- Which steps in onboarding create friction?
- Which segment has the highest repeat usage?
- Which feature deserves more development time, and which one deserves deletion?
This should also connect with feature prioritization frameworks. If a feature cannot be tied to measurable user behavior, it should struggle to win engineering time.
Step 1.3: Define your event taxonomy
Create a naming system before engineers start sending events. A clean taxonomy prevents chaos later.
- Use clear verbs, such as Signed Up, Created Workspace, Completed Lesson, Invited Teammate
- Avoid vague names like Clicked Button or Engaged
- Add event properties that matter, such as plan type, source, device, feature name, user role
- Write definitions for every event in one shared document
- Decide which events are business-critical and which are nice to have
Tools for this phase
- Amplitude for product analytics
- Mixpanel for funnels, cohorts, and retention
- PostHog for product analytics with more technical flexibility
- Segment or RudderStack for data routing
- A shared event dictionary in Notion, Airtable, or a plain spreadsheet
Phase 2: Foundation building
Step 2.1: Choose your setup model
You usually have three options:
- All-in-one product analytics tool for speed and simplicity
- Event routing plus analytics tool for cleaner data governance
- Warehouse-first setup for larger teams with strong technical capacity
Bootstrapped teams should usually start with the first or second option. Do not build a giant warehouse architecture just because a venture-backed SaaS did it on LinkedIn.
Step 2.2: Set up identity and user properties
This is where many teams silently corrupt their own data.
- Separate anonymous visitors from identified users
- Merge identities carefully after signup or login
- Track account-level and user-level properties separately in B2B products
- Keep plan, industry, team size, and acquisition source consistent
- Write rules for when properties can change and when they should stay fixed
If this sounds too technical, remember the bigger point. Good analytics is a language system. My linguistics background made this obvious very early. If two teams use the same word differently, the system lies. Event naming is pragmatics for software.
Step 2.3: Set up your first dashboards
- New user activation dashboard
- Onboarding funnel dashboard
- Retention and cohort dashboard
- Feature adoption dashboard
- Revenue conversion dashboard
Keep each dashboard tied to one decision area. Mixed dashboards become decorative wallpaper.
Implementation checklist
- Documented event taxonomy
- Clear owner for product data
- Tracking installed on web or app flows
- Identity rules checked
- Consent and privacy rules checked
- First dashboards live
- Baseline numbers recorded
Phase 3: Testing, reading, and scaling
Step 3.1: Test your data before trusting it
- Run your own journeys as test users
- Check whether each event fires once, not three times
- Confirm timestamps and user properties arrive correctly
- Compare analytics logs with database records where possible
- Ask support and product people whether charts match reality
This sounds obvious, yet many teams skip it and then make product calls based on broken events.
Step 3.2: Build weekly review loops
Analytics should feed action. Set a weekly cadence:
- What changed this week?
- Which segment improved or declined?
- Which release might explain it?
- What is the likely reason?
- What experiment or fix do we run next?
And pair this with user testing feedback loops because behavior alone rarely explains motivation. Data shows what happened. Human conversation helps explain why.
How should founders interpret product analytics without fooling themselves?
This is the part many articles skip. Setup is easy compared with interpretation. Charts do not speak for themselves. People force stories onto them.
Here is a founder-safe reading model I like:
- Observe: what changed?
- Segment: who changed, exactly?
- Contextualize: what product release, campaign, pricing change, bug, or seasonality event happened?
- Triangulate: does support, sales, interviews, or session review confirm the pattern?
- Decide: what action follows?
- Re-check: did the action improve the same metric for the same segment?
That fourth step matters a lot. In startup teams, the most dangerous phrase is, “The data clearly says…” No, it usually does not. It suggests. You still need judgment.
If you need qualitative input on a budget, pair your analytics reviews with user research on budget. Even five short interviews can save you from a very expensive wrong explanation.
What are the best practices for product analytics in 2026?
Practice #1: Track fewer events, but track the right ones
What it is: focusing on value-linked behavior instead of collecting every click.
Why it works: too many events create reporting clutter and interpretation errors. High-signal tracking keeps attention on activation, retention, and revenue paths.
- Define the one action that proves first value
- Track the actions leading to that moment
- Track the repeat actions that predict habit
Common pitfall: tracking cosmetic interactions because they are easy to instrument.
How to avoid it: ask whether losing this event would change a product decision. If not, it may not deserve space.
Metrics to track: activation rate, time to value, repeat usage.
Practice #2: Build your event dictionary before your dashboards
What it is: a shared document that defines every event, property, and business rule.
Why it works: teams interpret data better when the language is stable.
- Name every event clearly
- Define when it fires
- Record which team owns it
Common pitfall: engineers, marketers, and founders all using different meanings for “active user” or “completed setup.”
How to avoid it: define every business term in plain language and review quarterly.
Metrics to track: tracking completeness, duplicate event rate, dictionary coverage.
Practice #3: Pair quantitative data with user evidence
What it is: reading charts alongside interviews, support tickets, session recordings, and review text.
Why it works: numbers show pattern; human language shows motivation and friction.
- Review your top drop-off segment
- Watch a small set of sessions or read support threads
- Interview a few users from that segment
Common pitfall: treating analytics as complete truth.
How to avoid it: require at least one qualitative check before major product changes.
Metrics to track: funnel drop-off, support topic frequency, interview-confirmed friction themes.
This idea also shows up outside standard product tools. Coverage in Practical Ecommerce on Sellyze review analysis described systems that classify large volumes of customer reviews to reveal product clues. Different context, same lesson. Behavior data gets stronger when paired with language data.
Practice #4: Use cohorts, not averages, to judge product progress
What it is: analyzing user groups by signup date, plan, acquisition source, geography, or use case.
Why it works: averages flatten reality. Cohorts reveal whether changes actually improved outcomes for the users you care about.
- Choose one cohort logic that matches your business model
- Compare retention across time
- Link changes to releases, channels, or onboarding shifts
Common pitfall: claiming “retention is stable” while your best segment props up everyone else.
How to avoid it: review at least three cohort cuts every month.
Metrics to track: cohort retention, expansion behavior, segment conversion.
What mistakes do founders make with product analytics?
Mistake #1: Tracking vanity behavior instead of value behavior
Why founders make this mistake: vanity numbers feel comforting. They go up quickly.
The impact: teams celebrate noise while retention quietly weakens.
- Define one value event for each product line
- Remove low-signal events from executive dashboards
- Review whether each tracked event connects to a decision
If you already made this mistake: freeze new tracking, clean your event list, and rebuild dashboards around activation and retention first.
Mistake #2: Letting each team define metrics differently
Why founders make this mistake: speed. Early teams often move so fast they never standardize language.
The impact: meetings turn into semantic arguments, and trust in data falls.
- Create one metric glossary
- Assign one owner per metric family
- Review definitions with product, engineering, growth, and sales together
If you already made this mistake: choose one source of truth, rewrite your definitions, and accept a short painful clean-up period.
Mistake #3: Reading correlation as causation
Why founders make this mistake: pattern hunger. Humans want a neat story fast.
The impact: teams ship the wrong fixes and misread what really changed user behavior.
- Check release notes, bugs, campaign changes, and segment shifts before making claims
- Run controlled experiments where possible
- Compare with qualitative evidence
If you already made this mistake: revisit the chart, re-segment the data, and document alternative explanations.
Mistake #4: Waiting too long to instrument product analytics
Why founders make this mistake: they think analytics is something for later-stage companies.
The impact: they lose the earliest learning window, which is often where product direction is still flexible.
- Install a lightweight setup early
- Track only the key value path first
- Add detail as the product matures
If you already made this mistake: start now. A late simple setup beats a perfect imaginary one.
How do you measure success with product analytics?
Foundational metrics to track first
- Activation rate
- Time to first value
- New user funnel completion
- Weekly retention
- Feature adoption of one or two high-value actions
- Trial-to-paid conversion
Advanced metrics to add after three months
- Retention by segment and acquisition source
- Expansion revenue behavior
- Account-level usage depth for B2B
- Power-user frequency thresholds
- Leading churn signals
- Time between repeated value events
What should your metrics dashboard include?
- Real-time overview of activation and retention
- Trend views by day, week, and month
- Cohort comparison
- Alerts for sudden drops or spikes
- Segment filters for plan, source, device, and user role
- Space for notes on releases, experiments, and incidents
Do not forget annotation. A chart without context ages badly. Six months later, nobody remembers whether a spike came from a feature launch or a tracking bug.
How should product analytics change by startup stage?
Pre-seed and seed stage
Your reality: limited cash, unclear demand, fast changes, small team.
Approach:
- Track the core activation path only
- Instrument one retention measure
- Pair data with founder-led interviews every week
What to prioritize: time to value and repeat use.
What to defer: fancy warehouse architecture and dozens of feature-level charts.
Estimated requirement: a few focused days to plan, then lightweight setup.
Success looks like: you can explain why users stay or leave with more than guesswork.
Series A stage
Your reality: growth pressure rises, teams split by function, self-serve and sales may overlap.
Approach:
- Standardize metric definitions
- Track account and user behavior separately if you sell B2B
- Build weekly reviews across product, growth, and sales
What to prioritize: segment retention, expansion behavior, and onboarding friction.
What to defer: edge-case event coverage that no team reads.
Estimated requirement: one clear owner plus team training.
Success looks like: teams make product decisions from shared definitions, not departmental opinions.
Series B and later
Your reality: more channels, more teams, more product surfaces, more reporting pressure.
Approach:
- Move toward stronger governance of event schemas
- Connect product usage with revenue, support load, and account health
- Introduce advanced cohorting and predictive churn analysis carefully
What to prioritize: cross-functional clarity and account-level health.
What to defer: pet dashboards built for internal politics.
Estimated requirement: dedicated analytics ownership and stronger data discipline.
Success looks like: product analytics influences pricing, expansion, retention, and product portfolio decisions.
What outside signals can improve interpretation?
Strong founders do not read product analytics in isolation. They compare it with external signals and operational evidence.
- Customer support themes
- Sales objections
- Review platforms and public sentiment
- Competitor review mining
- Search demand shifts
- Release notes and bug logs
There is also a wider visibility lesson from search and publisher ecosystems. Coverage in Practical Ecommerce on Google preferred sources points to how source preference shapes what users see. Inside a product, something similar happens. Users also form preferred paths. They trust certain flows, ignore others, and return to familiar value routes. Your analytics should identify those preferred routes, then remove friction around them.
What is a practical 30-day action plan for founders?
Week 1: Research and alignment
- Choose one product goal: activation, retention, or paid conversion
- Write down the exact user action that represents first value
- Review current tracking and event naming
- Assign one analytics owner
Week 2: Event design
- Create your event dictionary
- Define user and account properties
- Map your onboarding funnel
- Choose your analytics tool
Week 3: Setup and QA
- Install tracking for the core value path
- Test all events manually
- Build one activation dashboard and one retention dashboard
- Record your baseline numbers
Week 4: Interpretation and action
- Review the first week of clean data
- Segment by source, plan, or user type
- Interview a few users from weak segments
- Choose one product change based on the evidence
Glossary of product analytics terms
Activation: the point where a new user completes the first meaningful action that suggests value has been reached.
Event: a recorded user action inside a digital product.
Event property: extra information attached to an event, such as device type, plan, or feature name.
Funnel: a sequence of actions used to measure conversion through a process.
Cohort: a group of users who share a common trait, often signup period or acquisition source.
Retention: the share of users who return and continue using the product over time.
Time to value: the time it takes a new user to reach their first meaningful success in the product.
Churn: when users stop using the product or cancel their account.
Segment: a subgroup of users defined by shared characteristics or behavior.
Event taxonomy: the structured naming system for events and properties.
Key takeaways
- Product Analytics Setup and Interpretation is a founder survival skill because startups need behavioral evidence, not hopeful storytelling.
- A good setup follows a clean path: questions first, event taxonomy second, tooling third, dashboards fourth, interpretation always.
- Seed-stage teams should stay narrow and focus on activation, time to value, and retention before building giant reporting stacks.
- Interpretation depends on context, segmentation, and qualitative confirmation, not just charts.
- Startups that read product behavior well make faster product calls, cut wasted feature work, and improve retention with less drama.
Final thought. I have little patience for analytics theater. Founders do not need more dashboards to feel sophisticated. They need a system that makes the next decision clearer. If your setup does not change product choices, cut scope until it does. Analytics should make your startup sharper, not busier.
People Also Ask:
What is product analytics setup and interpretation?
Product analytics setup and interpretation means two connected parts of working with product data. Setup is the process of deciding what user actions to track, defining events and properties, installing tracking, and checking that the data is accurate. Interpretation is the process of reading that data to understand how people use a product, where they drop off, what features they return to, and what changes may improve results.
How to set up product analytics?
To set up product analytics, start by defining what questions you want answered, such as activation, retention, or feature usage. Then create a tracking plan with events, user properties, and naming rules. After that, add tracking to the product, test the event data, build reports or dashboards, and review the results regularly so the data stays clean and useful.
What does product analytics measure?
Product analytics measures how people interact with a digital product. This can include sign-ups, feature clicks, session activity, funnels, retention, conversion, churn, and paths users take before completing or abandoning a task. The goal is to understand behavior inside the product rather than just traffic coming from outside sources.
Why is product analytics important?
Product analytics matters because it shows what users actually do, not just what teams think they do. It helps product teams find friction points, understand feature adoption, compare user groups, and make better product decisions. It can also help connect product changes to business outcomes like retention, conversion, and revenue.
What tools are used for product analytics?
Common product analytics tools include Amplitude, Mixpanel, Heap, Pendo, Fullstory, and Glassbox. These tools help teams collect event data, view funnels, study retention, segment users, and analyze feature usage. Some also include session replay, dashboards, and experiment analysis.
What are the main metrics in product analytics?
Common product analytics metrics include active users, retention rate, churn rate, activation rate, conversion rate, feature usage, session frequency, and time to value. Teams may also track funnel completion, cohort behavior, and revenue-related metrics depending on the product and business model.
How do you interpret product analytics data?
Interpreting product analytics data starts with a clear question, such as why users abandon a signup flow or why retention fell. From there, compare segments, review funnels, look at trends over time, and check whether changes happened after a release, campaign, or experiment. Good interpretation focuses on context, sample size, and whether the pattern points to a real product issue.
What is the difference between product analytics and data analytics?
Product analytics focuses on how users behave inside a digital product, such as feature usage, retention, and paths through an app or website. Data analytics is a broader term that can include finance, operations, marketing, sales, and other business data. Product analytics is one part of the wider analytics field.
What are the 4 types of analysis?
The four common types of analysis are descriptive, diagnostic, predictive, and prescriptive. Descriptive analysis explains what happened. Diagnostic analysis looks at why it happened. Predictive analysis estimates what may happen next. Prescriptive analysis suggests what action to take based on the data.
What are the 5 steps of analysis?
A simple five-step analysis process is: define the question, collect the data, clean and organize the data, analyze the results, and communicate what the findings mean. In product analytics, this often means starting with a product question, checking event quality, studying user behavior, and turning the findings into product changes.
FAQ
How do you know when your startup is ready for deeper product analytics instrumentation?
You are ready when simple traffic and signup numbers stop explaining user outcomes. If you cannot answer why users activate, stall, or churn, deeper instrumentation is justified. Start with one key journey, one activation event, and one retention view before expanding to more advanced product analytics setup.
Should B2B startups track account-level behavior differently from user-level behavior?
Yes. In B2B SaaS, one active champion can hide weak team-wide adoption. Track both user actions and account health signals such as seat growth, multi-user engagement, admin setup completion, and repeated workspace activity. This makes product analytics interpretation more useful for expansion and churn prevention.
What is the best way to handle feature launches in product analytics reporting?
Tag every major release with a launch date, affected segment, and expected behavioral outcome. Then compare pre-launch and post-launch cohorts instead of staring at blended averages. This gives you cleaner feature adoption analysis and helps separate true improvement from seasonality, campaign noise, or tracking errors.
How can founders use product analytics to improve pricing decisions?
Watch which usage patterns correlate with upgrades, downgrades, or stalled conversions. If free users hit value but never pay, your packaging may be wrong. If heavy users expand quickly, pricing may be too flat. Pair product behavior with billing data to make pricing decisions from evidence, not instinct.
What role do session recordings and qualitative tools play alongside event data?
They help explain why a drop-off happens, not just where it happens. If funnel metrics show friction, session review and interviews reveal confusion, hesitation, or broken expectations. A practical product analytics guide can help teams combine behavioral tracking with qualitative evidence.
How often should a startup review product analytics dashboards?
Weekly is usually best for early-stage teams. Daily reviews often create panic around normal variation, while monthly reviews are too slow for fast product iteration. Keep one recurring review focused on changes, likely causes, and next actions so product analytics reporting actually improves decision speed.
Can product analytics help with customer support and success workflows?
Absolutely. Support-heavy patterns often appear before churn becomes visible in revenue reports. Track repeated failed actions, abandoned setup steps, and low-usage accounts, then route those signals to support or success. This turns product analytics into an early warning system rather than a retrospective reporting tool.
How should mobile apps approach product analytics setup differently from web products?
Mobile products need extra attention to app version, device type, OS behavior, notification interactions, and offline-to-online sync issues. Event quality can break across releases faster than on web. For mobile product analytics setup, version control and QA after every update are especially important.
What signs show that your analytics stack is becoming too complex?
Warning signs include duplicate dashboards, conflicting definitions, multiple tools answering the same question, and engineering time spent maintaining low-value events. If the system creates more confusion than clarity, simplify. Many teams benefit from reviewing broader AI automations for startups to streamline repetitive reporting work.
How can product analytics support board updates or investor conversations?
Use it to show behavioral proof behind growth claims. Instead of reporting only top-line signups or revenue, show activation improvements, retention by cohort, and usage depth in your best segment. Investors trust startups more when product analytics interpretation connects product changes to durable user behavior.
