AI Agents1 min
AI agents as workflow, not as chat: The lesson for growth
Because growth agents must have inputs, tools, memory and clear limits.
Read article
INSIGHTS & UPDATES
Latest
GinnyTech articles
Practical guides on tracking, dashboards, experiments, data engineering, and AI applied to marketing.
AI Agents1 min
AI and experimentation: What changes when the variant is not deterministic
The AI experiments require new metrics, controls and evaluation logics.
Read article
SMEs1 min
Growth engineering with AI for SMEs: Small systems, great discipline
How to apply growth engineering principles in small businesses without huge teams.
Read article
Experimentation1 min
Backlog experiments: How not to turn it into a cemetery of ideas
A practical method to order, cut and decide on growth experiments.
Read article
Quality1 min
Bug Bash Growth: Test Experience, Data and Decisions
A modern bash bug doesn't only check if the UI works: check if the system measures well.
Read article
Career1 min
Career from growth engineer: Skills that remain even when changing tools
A concrete path to grow as a growth engineer in a world of AI, data and product.
Read article
Leadership1 min
Communicate the Impact: How to tell an experiment without selling smoke
A method to explain growth results to CEO, marketing, product and engineering.
Read article
Product Analytics1 min
Coorti e retintion: Growth is measured over time, not in launch
Because cohorts are essential to understanding whether a change creates lasting value.
Read article
Analytics1 min
Dashboards that decide: From monitoring to next action
How to design growth dashboards that reduce ambiguous instead of adding graphs.
Read article
Governance1 min
Data contract in growth: Small agreements that avoid great chaos
Because data contracts are crucial when experiments, dashboards and AI agents depend on the same events.
Read article
Product Engineering1 min
Design doc for experiments: Write before releasing
Because a short technical document can save experiments from ambiguous, bugs and unnecessary discussions.
Read article
Marketing Analytics1 min
Lifecycle email in growth: Messages that respond to behavior
How to transform automatic emails into part of the product learning system.
Read article
Experimentation1 min
Product Experiences: They are not races, they are questions
A more mature way of seeing experiments: not winners and losers, but operational learning.
Read article
Ethics1 min
Ethical Personalization: Where Growth Stops Helping and Starts Pushing
How to distinguish useful personalization from manipulation, especially with AI and advanced segmentation.
Read article
Analytics1 min
Events that matter: The minimum telemetry for growth
A practical guide to designing product events useful for experiments, retention and operational decisions.
Read article
Experimentation1 min
Fake door test: Validate the question without fooling people
How to use fake door tests in a useful, ethical and measurable way to evaluate new functionality.
Read article
Product Engineering1 min
Feature flags for growth: Release without betting everything
Because feature flags are product infrastructure, not just convenient technique.
Read article
Leadership1 min
Trust in Growth: How to prevent testing from appearing as laboratory games
Stakeholder confidence is built with governance, transparency and repeatable decisions.
Read article
Governance1 min
Governance for Growth AI: Simple Rules Before Difficult Problems
A checklist to govern agents, experiments and automations without blocking innovation.
Read article
AI Agents1 min
Human-in-the-loop: When the AI growth must ask permission
A simple model to decide where human review is needed in workflow agents.
Read article
Data Modeling1 min
User ID and Account: The hidden node of the B2B metrics
Because growth B2B requires models that distinguish person, workspace and company.
Read article
Experimentation1 min
How to write growth assumptions that don't seem to want
Templates and examples to transform vague ideas into experimental and measurable hypotheses.
Read article
Growth Engineering1 min
Learning loop: The real competitive advantage of growth engineering
How to build loops where use, data, experiments and product improve each other.
Read article
AI Agents1 min
AI agents memory: What to remember to grow without accumulating risk
How to design useful memory for growth AI agents without turning it into a dangerous archive.
Read article
Metrics1 min
North star metric: Compass or slide decoration?
How to choose a guide metric that truly guides product, data and experimentation.
Read article
Metrics1 min
Metrics guardrail: Growing up without breaking trust, margin or quality
How to choose protection metrics to avoid short-sighted optimizations.
Read article
Product Analytics1 min
Modeling the Activation: The moment when the product becomes real
How to define and shape activation without falling into vanity meters or arbitrary thresholds.
Read article
Data Modeling1 min
OLTP and OLAP in the growth: Because the production database is not enough
A journalistic and practical explanation of the difference between operational data and analytical data for growth.
Read article
Product1 min
Onboarding data-driven: Bring the User to the first value, not to the first tour
How to design onboarding using events, frictions and experimentation instead of infinite tutorials.
Read article
AI Agents1 min
Observabilita per agenti AI: Log utili, non registrazioni infinite
How to monitor AI agents in production with events, evaluations and alerts that are really operable.
Read article
Data Engineering1 min
Product Observability: Seeing Growth As It Happens
How to use log, metrics and mental tracking to see if a product is really generating value.
Read article
Statistics1 min
Peeking: Look too soon and convince yourself too quickly
Because checking the results every hour can turn a valid experiment into a factory of false winners.
Read article
Data Engineering1 min
Growth data pipeline: From click to decision
How to design lightweight but reliable pipeline to turn raw signals into product decisions.
Read article
Business1 min
Pricing as a growth system: Not only price, but signal
How to read pricing as part of product, experiments and positioning.
Read article
Privacy1 min
Privacy in the growth AI: Measure without turning everything into surveillance
A practical guide to balancing analytics, AI, log and user respect.
Read article
AI Agents1 min
RAG for growth: Reliable sources before brilliant answers
How to design retrieves and sources for AI agents that support product and marketing decisions.
Read article
Experimentation1 min
Randomization: The boring part that decides whether the test is worth
Because weak randomization makes even beautiful dashboards and great champions useless.
Read article
Team1 min
Experiment Respectives: Where Growth Becomes Knowledge
A light process to turn every test into reusable learning.
Read article
Product Engineering1 min
Gradual Rollout: The Adult Way of Saying ♪ We're Not Safe ♪
How to release growth changes for segments, thresholds and stop signals.
Read article
Team1 min
The growth engineer is not a full stack with metrics
What really distinguishes a growth engineer: code, product, data, experiments and responsibility for the result.
Read article
Business1 min
Sales feedback in the growth B2B: qualitative signals, quantitative decisions
How to use feedback sales without turning every anecdote into roadmap.
Read article
Statistics1 min
Sample size and MDE: How big should a test be?
How to reason on sample size, minimum detectable effect and duration without falling into blind formulas.
Read article
Analytics1 min
Growth Scorecard: A page to decide, not to impress
How to build a readable scorecard that connects experiments, metrics and next actions.
Read article
Product Analytics1 min
Growth segments: The medium and comfortable, but often lies
How to use segmentation to read experiments and metrics without getting lost in useless micro-clusters.
Read article
Growth Engineering1 min
Growth engineering: Because systems grow, not campaigns
A practical article on how to move from isolated campaigns to measurable growth systems, inspired by the principles of Growth Engineering.
Read article
Customer Experience1 min
Customer support as a growth sensor
Because tickets, chats and complaints are valuable product data if you model well.
Read article
Team1 min
Cross-functional growth team: Why it's not enough to add an analyst
How to compose a team growth with engineering, product, design, analytics and go-to-market.
Read article
Experimentation1 min
Triggering in A/B tests: Measure only those who could be affected
A practical guide to tripping: when to include a user in an experiment and when not.
Read article
Business1 min
Unit economics in experiments: Uplift without margin and growth
How to connect product and marketing tests to CAC, LTV, payback and margin.
Read article
Team1 min
Team growth standup: Less state, more friction removed
How to make the standup a useful time for controlled experiments, data and releases.
Read article
AI Agents1 min
Agents AI: Because they are not just smarter chatbots
An AI agent does not just answer: he observes, decides, uses tools and produces effects. Here's how to evaluate it with an engineering and human mindset.
Read article
Innovation1 min
AI 2027 in Italian: Scenario, graphs and what it really means
A reasoned summary in Italian of AI 2027: timeline, computations, risks of alignment, geopolitics and practical lessons for those who work with data and AI.
Read article
Data Engineering1 min
Observation for AI agents: See what happens inside the system
AI agents in production need logs, metrics and tracks. Without observation, you don't know if they are helping or creating invisible problems.
Read article
Data Engineering1 min
Events and telemetry for AI agents: Measuring actions
A practical taxonomy to measure AI agents: inputs, sources, tools, decisions, approvals and results. Growth starts from reliable signals.
Read article
Data Engineering1 min
AI data pipelines: From raw signal to useful decision
AI agents need solid pipes: collection, cleaning, storage, recovery and monitoring. Without pipeline, the agent works in the dark.
Read article
AI Agents1 min
The memory of AI agents: What to remember, what to forget
Memory makes an AI agent more useful, but also more risky. A practical guide to designing operational memory, context and privacy.
Read article
AI Agents1 min
RAG and growth engineering: How to give reliable sources to AI agents
The retrieved augmented generation allows AI agents to work on real documents. But careless for sources, versions and metrics becomes fragile.
Read article
Experimentation1 min
Experiments for products with AI agents: Test before automating
Each AI agent should be born from a hypothesis: which clutch reduces, which behavior improves, which risk controls. Here's how to set up useful experiments.
Read article
Product Engineering1 min
AI agent flag features: Release with control
The features flag allow you to release AI agents for segments, use cases and risk levels. They are a safety belt for intelligent products.
Read article
Experimentation1 min
Mettrici guardrail for AI agents: Growing up without breaking trust
An AI agent can improve one metric and worsen others. Guardrail meters protect quality, security, privacy and user confidence.
Read article
AI Agents1 min
Human-in-the-loop: Where Agent AI has to stop and ask for help
Human control does not slow AI agents down: it makes them reliable. That's when to use revision, approval and shared responsibility.
Read article
Product1 min
AI agents in Onboarding: Helping the User without submerging him
An AI agent can make onboarding clearer, personal and measurable. But it has to reduce friction, not add another level of complexity.
Read article
Marketing Analytics1 min
Agents AI in operational marketing: From ideas to measurable tests
AI agents can help marketing switch from generic brainstorming to measurable hypotheses, segments, experiments and decisions.
Read article
AI Agents1 min
AI agents to qualify lead: Speed without losing context
An AI agent can help sales and marketing qualify lead, but it must explain why score and respect human and commercial signals.
Read article
Customer Experience1 min
AI agents in customer support: Solve without hiding the problem
The AI agents in the support work when they reduce waiting and improve quality. If they close tickets too soon, they are just moving the problem.
Read article
Analytics1 min
Speaking Dashboards: When Agent AI explains the data
An AI agent can turn static dashboards into operational conversations, but it must distinguish insight, assumptions and decisions clearly.
Read article
Data Engineering1 min
AI agents and data quality: Checks before decision
AI agents can help you find problems in your data, but they cannot save a system without ownership, controls and shared definitions.
Read article
Experimentation1 min
A/B test for AI agents: Compare experiences, not just answers
Testing an AI agent requires more attention than a normal UI change: tripgering, guardrail, segments and output quality matter as much as conversion.
Read article
Experimentation1 min
A/A test for agentic platforms: Check randomization
An A/A test shows whether randomization, logging and pipeline work before launching real experiments on AI agents.
Read article
Experimentation1 min
Sample ratio mismatch in AI agents: The signal that the test is broken
If the groups of an experiment do not respect the expected proportions, the result can be unusable. That's why it happens with AI agents.
Read article
Experimentation1 min
Triggering of AI experiments: When to start a test
Triggering decides who enters an experiment. With AI agents it is central to avoid dirty analysis and wrong conclusions.
Read article
Experimentation1 min
Fake door for AI agents: Test the question before building
Before developing a complete AI agent, you can measure whether users really want it with an ethical, clear and respectful fake door.
Read article
Experimentation1 min
Reverse experiments and AI agents: Remove to understand what counts
Sometimes the best way to understand the value of an AI agent is to temporarily remove it from a segment and measure what really gets worse.
Read article
Business1 min
Printable for services with AI agents: Paying the value, not magic
When you sell AI agents, the price must reflect outcomes, risk, integration and support. It is not enough to calculate the cost of the model.
Read article
Ethics1 min
Personalization with AI agents: Useful only if you respect the User
AI agents can customize experiences, messages and paths, but without limits they risk becoming invasive or manipulative.
Read article
Analytics1 min
Churn and AI agents: Take action before the client escapes
AI agents can help identify risky customers, but value arises when explaining possible human signs, causes and actions.
Read article
Growth1 min
Retention loop with AI agents: From signal to action
An AI agent can feed retention loops if it helps the user to get recurring results. But valueless notifications and nudges burn trust.
Read article
Growth1 min
Growth loop for AI agents: When Use improves system
An AI agent can become part of a growth loop if each use produces data, learning and value that improve the next experience.
Read article
Product1 min
AI agents and product-market fit: Do not confuse curiosity with adoption
AI agents attract attention, but the product-market fit is seen by repeated use, concrete value and willingness to change habits.
Read article
Business1 min
SME AI agents: Useful automation without complexity
SMEs do not need complicated stacks to start with AI agents. Clear processes, reliable minimum data and human control are needed.
Read article
Product Engineering1 min
Designing streams for AI agents: From task to system
An AI agent works well when the flow is clear: input, context, tools, decisions, confirmation, output and measurement.
Read article
Security1 min
Instruments and permissions of AI agents: Give power without giving everything
An AI agent becomes useful when it can use tools. It becomes risky when those permissions have no limits, audits and approvals.
Read article
Security1 min
Security in active workflows: Protecting data, actions and users
AI agents introduce new risks: prompt injection, abused tools, sensitive data and unverified outputs. Security should be designed in the flow.
Read article
Privacy1 min
Privacy and log of AI agents: Measure without storing too much
AI agents need log to be controllable, but those logs may contain sensitive data. We need minimization, retention and governance.
Read article
Ethics1 min
Ethics of AI agents in growth: Speed with responsibility
AI agents can optimize conversions and retention, but responsible growth protects choice, clarity and trust of users.
Read article
Ethics1 min
Dark pattern and AI: When Optimization Becomes Manipulation
AI agents can generate very effective persuasive patterns. The team must recognize when optimization exceeds the ethical boundary.
Read article
Ethics1 min
Fairness in AI Agents: Average can hide who gets worse
An AI agent can improve the mean metric and worsen the experience of some segments. It requires group analysis, accessibility and context.
Read article
Analytics1 min
Scorecard for AI Agents: A page to see if they are working
A well-made scorecard brings together lenses, metrics, guardrAIls, segments, quality and decisions. It is used to manage AI agents without getting lost in detail.
Read article
Team1 min
Respectives for AI experiments: Transforming each test into knowledge
After an experiment with AI agents, the retrospective serves to save what the team has learned on data, users, prompts, tools and risks.
Read article
Team1 min
Growth engineering team in Era AI: Roles, rhythms and responsibilities
AI agents change the work of the growth teams, but do not eliminate skills. They serve engineers, PMs, analysts, designers and shared ownership.
Read article
Team1 min
The role of the growth engineer with AI: Building systems that learn
The growth engineer in the AI era works on experiments, data, features flag, observation, safety and products that learn from real signals.
Read article
Team1 min
PM, designer and engineer with AI agents: A closer collaboration
AI agents cross product, interface and infrastructure. To build them well, PM, designer and engineer must work on the same flow.
Read article
Analytics1 min
Documenting the Impact of Agents AI: From the result to the story
If an AI agent works, the team must know what changed: metrics, context, limits, decisions and next steps.
Read article
Product Engineering1 min
Bug bash for AI agents: Test experience, data and behaviour
Before the rollout, an AI agent must be tried by more people: not only for technical bugs, but for sources, tone, limits, permissions and strange cases.
Read article
Product Engineering1 min
Gradual Rollout of AI Agents: From the internal team to the production
Launching an AI agent for everyone in one day is risky. A gradual rollout allows you to learn, correct and increase autonomy with real data.
Read article
Data Engineering1 min
Monitor AI agents in production: Signals and alarms
An AI agent in production must be monitored as a live system: quality, costs, drift, errors, feedback and guardrail.
Read article
AI Agents1 min
Evaluating AI Agent Outputs: Correct, Useful, Secure
The quality of an AI agent is not measured only with correctness. An output must be useful, contextual, safe and operable.
Read article
Product1 min
The prompt as a product interface: Design well
In products with AI agents, the prompt is not only technical text. It is part of the interface, behavior and governance.
Read article
Data Engineering1 min
Data modelling for AI agents: Clean and reusable context
AI agents produce better insights when users, events, sources, tasks and results are modelled consistently.
Read article
Data Engineering1 min
OLTP and OLAP for AI agents: Operational and analytical data
AI agents work best when you distinguish operating and analytical systems. OLTP captures events, OLAP allows questions and decisions.
Read article
Growth1 min
The future of growth engineering with AI agents
AI does not eliminate growth engineering. It makes it more strategic: observation, experiments, ethics and human judgment become the real advantage.
Read article
Best Practices1 min
A/B test statistical errors: How to run reliable tests
90% of A/B tests in marketing are not statistically valid. Discover sample size, peeking, multiple testing and Minimum Detectable Effect for robust testing.
Read article
Data Science1 min
Predicting university drop-out: Italian data analysis
15-20% of Italian students leave after the first year. Predictive models identify risks. Practical guide with ISTAT data and predictive analytics.
Read article
Metrics1 min
Affinity index: Calculation and use of marketing analytics
Learn how to calculate the Affinity Index, the CPM in-target, and why media planning without affinity and how to fish in the dark. Formula, SQL, case study e-commerce.
Read article
Innovation1 min
AI marketing analytics 2026: What really works
Predictive audiences, data-driven attribution, MMM and creative testing. What works in 2026, what is hype, how to amplify (not replace) human work.
Read article
Best Practices1 min
Data Analysis Traps: Bias and common errors
Co-related vs causality, survival bias, Simpson's paradox, base rate neglect. Cognitive biases that cost millions to each company and how to avoid them.
Read article
Tutorial1 min
Boxplot data analysis: How to read the chart
Boxplot Guide: how to read it, create it in Python, when to use it. Learn how to see what the media hides in your business.
Read article
Data Science1 min
Cluster analysis customer segmentation: K-means and DBSCAN
Client segmenting with cluster analysis: K-means, DBSCAN, RFM, silhouette score. Python code, e-commerce study cases, when using which algorithm.
Read article
Data Science1 min
Conjoint analysis: pricing and product based on real data
How to use the joint analysis to discover the real value of product attributes and optimize prices and features with 250+ consumer data.
Read article
Best Practices1 min
Effective Dashboard: Ignored to decision in 7 steps
Because 70% of dashboards are never used and how to create a decision-driven dashboard that the team really consults every day.
Read article
Strategy1 min
Data-driven italy 2026: Analytical maturity of Italian SMEs
Analysis of the state of data-driven culture in Italian companies with ISTAT, DESIGN Index data and EU comparisons. How SMEs can start today.
Read article
Data Science1 min
Analytics in cycling: Power, cadence and performance with data
As professional cycling uses data to optimize performance. Business-wide cycling KPIs: efficiency, clustering and feedback loops.
Read article
Metrics1 min
E-commerce analytics: The 10 KPI metrics that generate revenue
Practical guide to essential metrics for e-commerce: CR, AOV, CAC, LTV, retention. Framework acquisition→conversion→retention with SQL is case study.
Read article
Data Science1 min
Sales Forecast: From Moving Average to Arima/prophet with Python
Three statistical methods to predict sales: MA, linear regression and SARIM. Seasonality management, measurement accuracy with MAPE/MAE.
Read article
Framework1 min
Growth loop vs funnel: Why linear funnel no longer scale
Structural difference between traditional funnel and growth loop. Viral, content, paid and network effect loop with K-Factor, cycle time and metrics measurement.
Read article
Framework1 min
HADI cycles: Framework growth hypothesis testing
The HADI cycle (Hypothesis, Action, Data, Insights) for growth analytics. How to test hypotheses, accelerate decisions and scale experiments with statistical rigour.
Read article
Metrics1 min
Increality testing: Measure real effect campaigns
How to measure the increase in marketing campaigns beyond attribution. Geo-lift testing, incremental testing and iROAS to discover the real ROI.
Read article
Innovation1 min
Innovation Italian University: Analytics and digital
How to bring data-driven management and analytics to universities. Learning analytics, digital transformation and successful cases Bologna, Politecnico Milano.
Read article
Strategy1 min
Perceptive maps: Positioning analysis competitor
How to build perceptive maps to understand the positioning of the brand. Complete guide: survey, PCA, MDS, strategic analysis with Python and Italian case studies.
Read article
Framework1 min
Kano method: Priority product requirements
Kano's model to classify features in Must-be, Performance, Attractive. How to use it for product prioritization and maximize customer satisfaction.
Read article
Management1 min
Millennium management: Generation y data-driven
As the Millenniums are transforming Italian management with data, transparency and speed. People analytics, real-time dashboards, data-driven cultures.
Read article
Metrics1 min
NPS limits and criticisms: Metric Alternative
Critical analysis of the Net Promoter Score. Scientific limits of NPS, comparison with CSAT/CES, driver analysis and alternative customer loyalty metrics.
Read article
Strategy1 min
Quantitative pricing: Mathematical models for fixing prices
Three scientific methods (Van Westendorp, Gabor-Granger, A/B test) to test prices and optimize profits with real data instead of intuition.
Read article
Innovation1 min
Privacy-first analytics: GDPR and tracking without cookies
Practical guide to cookieless tracking: server-side tagging, Allow Mode v2, first-party data and privacy-compliant alternatives to GA4.
Read article
Framework1 min
Agile market research: Fast and low-cost methods
How to do market research in 1-2 weeks instead of 6 months. 7 concrete methods: interviews, smoke tests, surveys, social listening.
Read article
Tutorial1 min
Statistically valid online surveys: Practical guide
How to design correct online surveys: sampling, measuring scales, biases in questions, sample size calculation, chi-square analysis.
Read article
Tutorial1 min
SQL for marketing: 15 query templates ready to use
Query SQL practices for marketers: revenue per channel, CAC, retention, ROAS, cohort analysis, window functions, CTEs per attribute and CLV.
Read article
Strategy1 min
Unit economics per startup: CAC, LTV, payback period
How to calculate CAC and LTV correctly, interpret the LTV/CAC ratio, optimize the payback period and benchmark by sector (SaaS, e-commerce, marketplace).
Read article
Want to apply these ideas to your project?
Book a short call: we will look at tracking, data, and dashboards, then identify the first useful intervention.
Book a 15 minute call