For CMOs, VPs and marketing, CX and sales professionals
I teach marketers to use AI and first principles to turn data into better decisions.
Data science methods, market intelligence frameworks and AI-enabled analysis. For the marketing, CX and sales leaders who have always driven the growth and want the evidence to prove it.
Looking for a traditional digital marketing growth solution? See the Digital Marketing Architecture Roadmap
Four levels of AI in marketing
They are not better and worse ways of working. They answer different questions, and each one depends on the one before it. Most investment so far has gone into the first two, which is where the tooling was. The analytical value sits in the last two.
-
I
Assistance
Drafting, summarising, ideation, desk research. One task at a time, with you in the loop on each one.
Saves time -
II
Automation
The process runs, not the task. The monthly report, the competitor digest, the data pull. Built once, repeats without you. This is infrastructure, and it is what makes the levels above it affordable.
Creates capacity -
III
Diagnosis
Why did conversion fall, why did that segment leave, what actually drove the lift. Cohorts, correlation, regression, controlled tests.
Explains outcomes -
IV
Intelligence
Patterns nobody thought to look for. Assumptions nobody tested. What the market is likely to do next, and what you should do about it.
Changes decisions
AI is a fast junior market intelligence analyst
It does the intelligence work a junior analyst would do, in minutes instead of weeks. It does not decide what is worth knowing.
What it does for you
- Pulls, cleans and reshapes the data
- Writes the code you used to wait on
- Runs the method: cohorts, segments, tests, models
- Drafts the chart and the summary
- Repeats it every month, unprompted
The part that used to need a hire, a vendor, or six weeks.
What it still needs from you
- The question worth asking in the first place
- Context about your market it cannot know
- Whether the sample is representative
- A judgement on whether the finding is causal or coincidence
- The decision, and accountability for it
Brief it badly and it returns something confident and wrong, on time.
That is how a market intelligence function gets built without headcount: you supply the judgement, the analyst supplies the throughput, and a documented method governs how work is briefed and checked. That method is what I teach.
Case studies: problems solved and impact
Seven business questions I was given, the methods I applied, and what changed as a result. Every outcome below is work I led.
Is our marketing actually incremental, or are we paying for demand we already had?
- The question
- Platform-reported conversions said paid was working. Platforms are not disinterested witnesses. Was any of it incremental?
- The intelligence
- Campaign and conversion data across channels, restructured into holdout and exposed groups rather than read off the platform dashboard.
- The framework
- Deliberate control group design using geo and audience holdouts, measuring incrementality instead of accepting last-click attribution.
- The insight
- Reported return and incremental return were different numbers, and the gap varied sharply by channel. The budget split optimised on reported performance was wrong.
- The decision
- A continuous experimentation programme across channels, landing pages and the activation journey, reallocating on measured return.
- The outcome
- Cost-effectiveness improved 35% while scaling acquisition 8x at flat budget.
Which customers are actually leaving, and which segment is quietly growing?
- The question
- Headline churn was stable. But a stable average can hide one segment collapsing while another compounds.
- The intelligence
- Cohort-level retention and behavioural usage data across 90,000+ customers on direct and reseller channels.
- The framework
- Cohort velocity analysis. Not just how many leave, but how fast, from which segment, and which way the mix is moving.
- The insight
- The aggregate was concealing two opposite movements. The growing segment and the leaking one had entirely different economics.
- The decision
- Acquisition and journey investment redirected toward the compounding segment; friction fixed where the leak was.
- The outcome
- Conversion from non-paying to paying lifted 15% → 42–60%, contributing to 58% customer growth and 16.5% revenue expansion.
Where should we compete when we can't win everywhere?
- The question
- A challenger brand can't win every segment at once. Which parts of the category were actually growing, and which of those were winnable?
- The intelligence
- USTMA industry shipment data decomposed by rim size and segment, against our own sell-through and the shelf we held at each retail partner.
- The framework
- Category segmentation crossed with growth rate and our own presence, separating where the market was moving from where we happened to be strong.
- The insight
- Growth was concentrating in specific size bands where our portfolio was under-represented. The category wasn't flat. It was migrating, and our mix was anchored to where it used to be.
- The decision
- Refocus portfolio and pricing on the migrating segments, and take the category story to the retailer rather than a price list.
- The outcome
- 50% market share growth over three years in the targeted segments, and a shift from supplier to strategic vendor relationship. The buyer began using our analysis to plan the category.
How do we prove a promotion actually worked?
- The question
- Trade and promotional spend was being judged on sales during the promotion. How much of that would have happened anyway?
- The intelligence
- Point-of-sale data at store level, with matched control stores selected on pre-period behaviour rather than convenience.
- The framework
- Test-and-learn design in the APT tradition: matched control groups, a clean pre-period, and a measured lift rather than a reported total.
- The insight
- Some of the best-performing promotions were the least incremental. They were being credited with demand that already existed.
- The decision
- Promotional investment reallocated by brand and region on measured incremental return.
- The outcome
- An ROI framework adopted for evaluating commercial spend, presented to regional VPs and executive leadership.
How do we size a category that doesn't exist yet?
- The question
- Should Budweiser launch a non-alcoholic beer in Canada, and if so, who buys it and on what occasion?
- The intelligence
- International markets where NA beer had already scaled, domestic category and occasion data, and demographic shift signals.
- The framework
- Demand-space and category entry point analysis, mapping occasions where the need existed but the category didn't yet serve it.
- The insight
- The opportunity wasn't substitution by existing beer drinkers. It was a set of occasions where beer had been ruled out entirely. Demand the category had never addressed.
- The decision
- Launch positioned against the occasion rather than against alcoholic beer.
- The outcome
- Budweiser Zero brought to the Canadian market.
Where is our industry heading, and are we positioned for it?
- The question
- Where was the Canadian beer industry going, and where should the portfolio sit to meet it?
- The intelligence
- A Bain & Company industry-mapping framework built for the US market, rebuilt against Canadian category, channel and premiumisation data.
- The framework
- Treat a structurally similar market that is further along as a leading indicator, then test which dynamics transfer and which are artefacts of the source market.
- The insight
- A full industry map separating what was structurally growing from what was cyclically strong, and where the portfolio was over-indexed to the wrong half.
- The decision
- Directional growth priorities set at portfolio level rather than defended brand by brand.
- The outcome
- The map supported Labatt's acquisition of Mill Street Brewery, with the rapid growth of the premium and import segments as the evidence behind the case.
Where do we grow without cannibalising what we already have?
- The question
- With roughly 2,400 locations open, where was there genuine unmet demand, and where would a new site simply take sales from an existing one?
- The intelligence
- Trade-area demand potential, regional saturation, and observed sales-transfer patterns across the existing national network.
- The framework
- A predictive expansion model paired with a cannibalisation model. Every candidate site scored on what it would create and what it would take.
- The insight
- Net contribution, not gross forecast, was the only defensible basis for a site decision, and it reordered the pipeline materially.
- The decision
- Capital deployed against net incremental potential, advised to the CFO and steering committee.
- The outcome
- Forecast accuracy improved from 40% to 70%, and the model became the basis for the five-year national growth strategy.
Who is Daniel Zaitz
Economist turned marketer, bringing data science to marketing with AI.
I was the market intelligence lead behind a USD 300M consumer portfolio at one of Latin America's fastest-growing consumer brand companies. I built the market intelligence function from zero at a tire manufacturer with CAD 800M in annual revenue, modelled where a 2,400-restaurant chain should grow next at Tim Hortons, Canada's flagship quick-service brand, and ran marketing for a telecommunications B2B SaaS business at USD 12M ARR. Different industries, and the same pattern in every one: the data existed, the intelligence did not always.
Experience
B2B SaaS & Telecom
VoIP.ms
Director of Marketing & Demand Generation
4 years
Manufacturing
Sailun Tires America
Manager of Market Intelligence
4 years
CPG & Beverage
Coty · AB InBev
Manager of Sales & Marketing Intelligence
5 years
Retail & QSR
Tim Hortons / RBI
Senior Manager, Real Estate Analytics
1 year
Insurance
Liberty Mutual
Growth Strategy Analyst
1 year
Economic Consulting
LCA Consultores
Consultant
2 years
Consulting clients
Hunter Douglas · Garda Cash Services · Telecom Exchange · Fit It Out
Education
BSc Economics
FEA‑USP, University of São Paulo
Top-three economics faculty in Brazil. Heavy on econometrics, statistics and quantitative methods.
MBA, Marketing
Schulich School of Business, York University
Specialization in Marketing.
Five ways in. Start wherever you are.
Free. Start here and take whatever is useful.
- YouTube channel
- Knowledge Hub: resources and templates
- Notes and analysis on social
Then $300/yr. Launch price ends 31 October 2026.
- A 30-minute discovery call
- 10+ workflows customised to your industry and data*
- Delivered within a week
- Method and validation notes on every one
One-off · includes the Library
- Everything in the Library
- In-depth discovery of your problem
- A recommended path: which workflows to build, and the training plan to run them
Scoped to team size & objectives
- Your team trained on the method
- Workflows customised to your data
- Built during the engagement, not after
- Your team owns and runs them
Scoped to the business
- An AI-driven market intelligence capability built with you
- AI workflows, training and governance
- Your team trained on the method
- Ongoing support for a minimum of six months
* After a 30-minute intake call I send you the recommended workflows adjusted to your context. See an example.
The workflows: choose ten, customised to you
A 30-minute intake call
What you sell, what data you hold, what decision is stuck.
I build your ten
Customised to your industry, your competitors and your data.
Delivered within a week
Ready to run, with the method and its limits written down.
What plugs in
Public data plus whatever you already have. No warehouse, no new subscriptions, no engineering ticket.
What you actually receive
A folder you open in Claude. Your context, your workflows, your outputs.
Every workflow ships with its method and its limits written down, so whoever runs it next quarter knows what the number means. Expand any folder below.
▸zaitz-intelligence/
README.mdStart here. What you have and how to run it.
▸context/What Claude needs to know about you
company.mdWho you are, what you sell, how you make money
market.mdYour category, segments and competitors, named
data-sources.mdWhat data exists, where it lives, what it can and cannot answer
glossary.mdYour terms defined once, so every output uses them consistently
▸workflows/The ten you chose
▸share-of-search/
SKILL.mdThe instructions Claude follows, step by step
method.mdWhy this works, where it breaks, how to validate the result
inputs/competitors.csvYour brand set, editable
outputs/Where each run lands, dated
competitive-monitor/Same four files, different question
win-loss-analysis/
category-entry-points/
customer-voice-mining/
...and the rest of your ten
▸reports/What the workflows produce
2026-08-market-brief.mdThe monthly read, assembled from the workflows above
2026-07-market-brief.md
HOW-TO-RUN.mdPlain instructions for someone who has never done this
Includes an assessment of applicability. Before anything gets built, I tell you which of your choices actually work with the data you have. Sometimes the honest answer is fewer than ten.
Which workflows do you need?
Three questions. You get a high-level recommendation of the market intelligence workflows that fit your situation, and an honest note on any that your data won't support yet.
What a market intelligence analyst actually knows how to do
The workflows are these skills, packaged. That is the whole difference between a workflow and a prompt: a prompt produces an answer, a method tells you whether the answer holds.
The skill
What it actually does
Where it shows up
Segmentation & clustering
Groups customers by how they behave, not by how someone described them in a workshop.
Segmentation rebuild · ICP refinement
Sampling & significance
Knows when a result is real and when it is forty people and a hunch. Runs underneath everything else.
Every workflow
Causal inference & test design
Separates what you caused from what would have happened anyway. Holdouts, matched controls, clean pre-periods.
Incrementality test design · Promotion lift
Elasticity & sensitivity
Quantifies how demand responds to price, promotion and mix, instead of arguing about it.
Pricing study · Discount audit
Time series & forecasting
Decomposes trend, seasonality and noise, so you plan ahead of a peak rather than reacting to it.
Demand forecast · Share of search
Triangulation
Builds a defensible number from incomplete sources when no single dataset has the answer.
Market sizing · Dealer sell-through
Category & competitive structuring
Cuts a market into segments that actually behave differently, then finds where the growth is moving.
Category analysis · Industry mapping
Qualitative synthesis at scale
Turns thousands of calls, tickets and reviews into themes with the evidence attached.
Customer voice mining · Win/loss
The return isn't one analysis. It's the next hundred decisions.
↺ and the learning sharpens the decision after that
Questions people ask before starting
What is market intelligence?
Market intelligence is the capability of knowing what is happening in your market and what to do about it. It combines external signals such as category movement, competitor activity, demand and search behaviour with internal evidence from customer, product and campaign data, applies a method to test what is real, and ends in a decision. It is a function, not a report.
What is the difference between market intelligence and marketing?
Marketing decides and executes. Market intelligence supplies the evidence those decisions rest on. Marketing asks what we should do; market intelligence answers what is actually true about our market, our customers and our position. In most companies marketing exists and market intelligence does not, so decisions default to intuition.
Why is market intelligence important?
Because the cost of a wrong bet on a launch, a price or a market entry is far higher than the cost of knowing first. Companies that compete on category understanding staff it accordingly. P&G runs a Consumer & Market Knowledge organisation of roughly 800 people, and its insights leader reports to the CEO. That is what treating intelligence as a first-class function looks like.
If market intelligence is so important, why don't most companies use it?
Because until recently it required specialists. A dedicated analyst, syndicated data subscriptions, and weeks per study. Below a certain size no company could justify the headcount, so the work simply did not get done. It was never a question of value. It was a question of access.
How has AI changed market intelligence?
It removed the execution barrier. Pulling and cleaning data, writing the code, running the method and drafting the output no longer need a specialist queue. What remains is judgement: which question matters, whether the sample is representative, whether a correlation is causal. AI made the analysis cheap and the methodology more valuable.
Do you also work with traditional marketing workflows?
Yes, and most teams start there. Reporting, campaign analysis, briefing and content all improve with the same approach. The difference is that the goal is not only doing existing work faster. It is using the same data to find things nobody has looked for yet.