Your Data Is Your Advantage
ABOUT TADPULL
From the Harvard classroom to the boardroom, we’ve been helping clients build AI systems since 2013, the tools have gotten a lot better.
WHAT WE KEEP SEEING
Three Patterns, in the Classroom and in the Data
We teach marketing leaders at Harvard and we run the same frameworks with operators every week. The same three things come up, almost regardless of company size.
01 Confident Answers on Messy Data
The output looks polished either way. That is the part that makes it dangerous.
02 Pilots with No Agreed Business Metric
Everyone can describe what was built. Nobody can say what it moved.
03 Teams Using a Fraction of the Tools
Not a licensing problem. Nobody taught the team how any of it actually works.
Underneath all three is the same gap, and it is not a technology gap.
WHAT WE BELIEVE
Which means the model is not the advantage. What a company cannot rent is its own data and its own context. Managed as an asset and paired with the judgement of its people, that is the differentiator. Managed carelessly, the same data becomes a liability, feeding wrong answers into real decisions.
Models Get Better for Everyone at Once
01 | Name the Target Variable First
Agents optimize whatever you point them at. We anchor on customer lifetime value. because it is the one number marketing, finance, and the board can all agree on.
02 | Treat Context as an Asset
Clean customer data, plus written-down know-how (brand standards, playbooks, definitions) that people and agents can both read.
03 | Redesign the Work, Then Add the Agent
Bolting a tool onto an old process rarely moves the needle. We map which tasks agents own, which they draft for review, and which stay with people.
WHY THIS IS HARD TO BUY
Most companies can find someone to teach AI, someone to write the strategy, or someone to build the tool. Rarely do those three come from the same team, and the gaps between them are exactly where AI programs stall.
Teach It, Plan It, Build It
A Trainer
Teaches the vocabulary. Leaves before anything is built.
A Build Shop
Ships the tool. Nobody on your team can run it after.
A Strategy Firm
Writes the roadmap. Never tests it against your data.
Tadpull
All three, one team, working off a unified data layer. The roadmap gets tested against reality in weeks, not debated for quarters.
WHO WE ARE
The frameworks we teach are the ones we run with clients, and the client work flows back into the cases and the research.
We Live in Two Worlds on Purpose
Jake Cook, Harvard Business School Lecturer, Cofounder
Jake teaches digital marketing and AI at Harvard, cofounded Tadpull in 2013, and brings a physics and mathematics background to roughly twenty years of practice. He has written cases with Dane Vahey, who leads B2B marketing at OpenAI, and with Dr. Peter Fader and Dr. Daniel McCarthy, two of the leading researchers in quantitative marketing and customer lifetime value.
Eulalie Cook, Ph.D., Cofounder
Eulalie runs Tadpull day to day. Her doctoral research was in consumer choice using mixed methods, and she leads the customer interviews that explain the why behind the numbers.
Agents can summarize what customers did. Someone still has to talk with them.
Tadpull is a commerce intelligence company. Our platform, the Tadpull Data Pond, unifies customer, order, and catalog data, and we are an approved partner of both OpenAI and Anthropic.
BOZEMAN, MONTANA, SINCE 2013
Where the Name Came From
A tadpole transforms to live on land and in water, which felt like decent inspiration given the pace of change in digital. We spelled it with “pull” because the best brands pull the right customers into their orbit rather than push messages at everyone.
The mark says the same thing twice: the outer arrow is continuing growth, the inner one is upward progress.
Since 2013 the aim has not changed: bring the scientific method, and the best thinking from technology, finance, and design, to marketing.
HOW WE THINK
The 5Cs of Commerce
01 CAPITAL
Money goes into the business with a return it has to produce, on a clock somebody else is holding. Every decision downstream is judged against that.
02 CULTURE
A team deploys that capital, and no model survives a team that will not use it. We always begin with a small sprint inside the culture, letting the data set a realistic range of opportunity.
03 CUSTOMERS
Our north star is the free cash flow each customer produces over a lifetime, forecast by decile using actuarial and machine learning methods. A small share of customers nearly always drives a large share of profit, and Eulalie’s qualitative research explains what draws them in.
04 CAMPAIGNS
How do the best customers find us, engage, and stay? A Black Friday promotion can set a sales record and still bring in customers who never return, a pattern only longitudinal data reveals.
05 CATALOG
Product and ERP data often hold the best returns of AI, showing how products turn into cash and surfacing product development ideas from the best customers.
In short: Deploy Capital within your Culture to reach the right Customers with the right Campaigns and the right Catalog.
HOW WE START
A Small Sprint, Before Anything Else
We like to start small: prove the value, take risk off the table, find a few wins people can see, and build from there. Trust compounds faster than a rollout does.
01 SPRINT
A short engagement inside your culture, letting your own data set a realistic range of opportunity.
02 ROADMAP
Which parts of the marketing value chain are ready for agents, which need human judgement, and in what order to move.
03 BUILD
Real pilots on your data, run alongside your team, so your people own the system rather than renting it from us.
That approach tends to turn into a long relationship. Many of our clients have worked with us for three to ten years.
WHAT IT ADDS UP TO
6% to 33%
We Measure Success in Years
Mystery Ranch e-commerce as a share of the business, over a decade together, through acquisition by YETI.
76%
Mack’s Prairie Wings revenue growth across a seven-year engagement, before its acquisition by Bass Pro Shops.
10+ Years
Jackson Hole Mountain Resort
An audit is the shortest path to knowing what is actually there: what your customer data can support today, where the value is hiding, and what the first visible win should be.