How to build an AI-ready data foundation in one week

Day 030 minutesJust you

Is it time?

Quick test. What was your revenue last month? Now ask your head of sales, and then whoever does your books.

Did you get the same number? If you hesitated, you're in good company. Numbers come up in every conversation, and there's always something slightly off about them: a report someone rebuilds by hand every Monday, two dashboards that disagree.

Maybe you've connected your CRM and Google Analytics to ChatGPT or Claude. The answers came back fast and confident, and they still didn't match what finance says.

The problem was never a lack of data. It's that nobody agreed what it means. Tick the ones that sound familiar.

0

Tick what sounds like your company.

Your homework before Day 1

Three small tasks. Open each one for the details.

1Write down five questions you struggled to answer last week

Questions you asked, or were asked, and couldn't answer quickly, or didn't fully trust the answer. Make each one specific.

Too vague"How is marketing doing?"

Specific"Which channel brought the most customers who ordered again within 90 days, over the last six months?"

Too vague"Are our sales reps performing?"

Specific"How many deals did each rep win last quarter, and how long did each one take from first call to signed?"

2List the tools where your numbers live

CRM, e-shop, payments, ads, analytics, accounting. Next to each, write who looks after it.

3Book one afternoon with the people who own the numbers

Usually two or three people: whoever leads the company, whoever runs sales or marketing, and whoever handles the money.

Exercise · 5 minutes Turn a vague question into one worth answering With AI, anyone can get an answer in seconds. What sets you apart is asking the right question. Open the exercise and sharpen one of yours. Try the exerciseHide the exercise

Start with a vague question

1
Start with the decision
What will you do differently depending on the answer?
2
Name the number
Which single number would make that decision for you?
3
Add who, what and when
Which customers, which products, which time period?
4
Define the words
Any word two people could read differently needs one meaning.
5
Get your specific question
Moodree combines your answers into one question you can bring to Day 1.

Before

After

To decide
The number
For
Defined as

You walk away with

Five sharp questions, a list of your tools, and an afternoon in the calendar. That's all you need to start.

Days 1–21–2 daysThe people who use the numbers

The business interview

Day 1 is about one thing: agreeing what your business needs to know, and what every word in those questions means. Nothing technical happens.

It's the hardest part of the week and the one that matters most. It's often the first time a team says out loud what a "customer" is, and finds out they've been counting differently for a year.

Who's in the roomOne session, or two to three smaller ones

Most small companies

One session

About 4 hours · 2–4 people

Whoever leads the company, whoever runs sales or marketing, and whoever handles the money. Everyone hears every definition, so it gets settled once.

Bigger, or very different teams

Two or three smaller sessions

About 2 hours each · plus a short wrap-up

Split by area, for example sales and CRM, marketing and e-shop, finance and operations. The wrap-up settles the words that overlap, like "customer".

Tip: invite the people who use the numbers to make decisions, not only the people who put the reports together.

The interview in four parts

Each part answers one question. Part 2 is the one most teams skip, and it decides what gets built first.

1
20 min · Your questions"What can't we answer today?"
You leave with
A list of specific questions
2
20 min · Their value"What would we gain with this answer, and what is not having it costing us?"
You leave with
A gain and a cost next to every question, so you know which to answer first
3
60–90 min · Your words"What exactly do we mean?"
You leave with
One definition for every word in those questions
4
20 min · Your tools and people"Where does each answer live, and who uses it?"
You leave with
Sources, known gaps, and who needs what, how often

Cheat sheet: the most common definitionsEight words every team defines differently, and the questions that settle them

Customer

Does a free trial count? Someone who paid once, two years ago?

Revenue

Before or after discounts? With or without VAT? What happens to refunds?

Won deal

Signed, paid, or live? What about deals signed but still testing?

Active

Logged in, ordered or paid? In the last 7, 30 or 90 days?

Lead source

The first touch or the last one? Who fills it in, and what if it's empty?

The same customer

How do we know a contact in the CRM and a payment in billing are one person?

Which tool wins

If the CRM and accounting disagree, which one is right?

Recent days

When is yesterday's number final? Some tools keep updating it for days.

Why this matters

A data foundation is only as good as the business rules behind it. Every number it produces follows a decision someone made in this room, so those decisions have to be clear, agreed and written down.

1

Agree, don't assume. Settle each definition out loud, together. If two people mean different things by "customer", the numbers will never match.

2

Log what isn't decided. If something isn't confirmed, or people disagree, point it out and write it down as an open question. It gets settled before it's built, never guessed.

3

Ask for examples, not opinions. Open last month's report together and point at the numbers.

Not sure where to start? Get help with your business interview

We've run more than 50 of them. In 30 minutes we'll help you plan yours: who to invite, which questions to bring, and what to do next. If you'd like, we can also run the interview with your team, or review your notes and KPI matrix afterwards.

Get a free brainstorming session

This session is completely free, with no commitment.

What it produces: your KPI matrix

Every number your questions need, what it means, how it's broken down, where it lives, and who uses it. This is the blueprint for the build.

KPI matrixExample

KPI	Definition	Broken down by	Source	Who · how often
MarketingCost per new customer	Ad spend ÷ new paying customers	ChannelCampaignMonth	Google Ads, Meta Ads, Shopify	Marketing · weekly
Detail card · every KPI gets one Formula Ad spend in a month ÷ customers whose first paid order is in that month Decision it supports Where next month's ad budget goes Without it Budget follows the cheapest clicks, not the best customers Final after 3 days, once ad platforms stop re-attributing
MarketingRepeat rate	Share of new customers who place a second paid order within 90 days	ChannelFirst productMonth	Shopify	CEO · monthly
SalesWin rate	Won deals ÷ all closed deals. Won means signed and first invoice paid.	Sales repLead sourceQuarter	Pipedrive	Head of sales · weekly
SalesSales cycle	Days from first call to won	Sales repDeal size	Pipedrive	Head of sales · monthly
FinanceNet revenue	What customers paid, after discounts and refunds, without VAT	ProductChannelMonth	Shopify, accounting	Finance · monthly
SalesLead source	To confirmFirst touch or last touch?	Channel	Pipedrive, GA4	Marketing · weekly

The prompt we use: turn your interview notes into this matrixPaste it into Claude or ChatGPT with your notes. It returns this matrix, a detail card per KPI, open questions and suggestions

You are an experienced analytics engineer. Turn our business interview into a KPI matrix that a data team could build from without asking us again.

INPUT
Interview notes or transcript (below).

WORK THROUGH IT IN THIS ORDER
1. Questions. List every business question we want answered, in our words. Rank them by value: what we would do differently with the answer, and what not having it costs us today.
2. KPIs. For each question, list the KPIs needed to answer it. Don't invent KPIs nobody asked for.
3. Definitions. For each KPI, settle the details that usually cause arguments:
- revenue: before or after discounts, refunds and VAT; which currency; order date or payment date
- customer: does a trial count; is it a person or a company
- won / active / churned: the exact event and time window
- attribution: first touch or last touch
- ratios: sum the top and the bottom, never average percentages
- conversion: "this week" (flow) or "this group over time" (cohort)
4. Joins. If a KPI needs data from two tools, name the ID that links them. If there isn't one, write "No shared key: matching will be approximate".
5. Freshness. Say when the number is final. Ad platforms keep changing conversions for about 3 days; email engagement for about 7.
6. Gaps. If something is unclear or two people disagreed, don't guess. Write "TO CONFIRM:" followed by the exact question we need to settle.

OUTPUT
A. KPI matrix, one table:
KPI | Definition (plain language, for a new colleague) | Broken down by | Source tool | Who · how often
B. A detail card for each KPI:
- Formula, in words: top ÷ bottom, and the time period
- Grain, in words, e.g. "campaign × day"
- Decision it supports
- Without it: what we lose today
- Final after: how many days until the number stops changing
- TO CONFIRM: open questions, if any
C. Open questions: every TO CONFIRM in one list, with who should answer it.
D. Worth considering: up to 5 KPIs that companies like ours usually track and we didn't mention. One line each on why. Keep them separate from the matrix.

Interview notes:
[paste your notes or the meeting transcript here]

You walk away with

Your questions ranked by what they're worth, and a KPI matrix everyone agreed on. It ends the "whose number is right?" debate for good.

Day 22–4 hours, alongside the interviewYou or your technical person, guided by Moodree

Connect your tools

Pick your tools from Moodree's ready-made connectors and give read-only access. Moodree guides you through every step, and your data starts arriving the same day.

One-time setup: how it works

Where does my data live?: In your own Google Cloud account, not Moodree's.
Who owns it?: You do. If you ever leave, everything stays with you.
What do I need to do?: Create two accounts once, Google Cloud and GitHub. Moodree walks you through it.

Want the full technical picture first? Book a walkthrough of the setup.

1Pick your sources
From ready-made connectors.
2Give read-only access
Moodree shows you where to click.
3Data starts arriving
With as much history as each tool keeps.

WorkspaceConnect in a few clicks, read-only.

ConnectorAsk for a source, and it's live.

Checked every time it arrives

Every night's run is logged step by step, so nothing fails quietly.

Did the data arrive?Did every connector run?Do the row counts look normal?Did anything get renamed?

You walk away with

A few clicks, and all your data is in one place you own. Refreshed every night, checked every time it arrives. For most teams, it's the first time their data has been looked after like this.

Days 3–51–2 days of check-insMoodree, with you checking in

Moodree builds your foundation

Moodree turns the raw data into numbers you can trust, using the KPI matrix from Day 1. Your part is to keep an eye on it: answer the odd question and approve the changes as they come in.

Raw copyCleanedBuilding blocksYour rulesReady to use

How the build works, the kitchen versionFive steps from raw data to numbers you can trust

Picture a restaurant kitchen. Raw ingredients arrive, get washed and prepped, go through a recipe, and end up on a plate made for the person ordering.

DeliveryA daily copy

Raw data from each tool, untouched. Personal details locked away separately.

WashingCleaned

Duplicates removed, formats fixed. Broken rows set aside, not lost.

PrepBuilding blocks

One shared version of customers, products, dates and campaigns.

RecipeYour rules

The KPI matrix from Day 1, turned into code. Revenue means one thing.

On the plateReady to use

Tables shaped for each person: you, marketing, sales.

The full buildThe whole build in two minutes: connect, plan, build, explore and use.

You walk away with

A fully built foundation: every layer built, checked and described, ready for your first questions.

Talk it through with someone who's done it before

Moodree does the heavy lifting, and we're here for everything around it. In 30 minutes we'll look at your build with you and plan what comes next. We can also set up the first build together, review each change before it goes live, check your key numbers with you, and keep an eye on everything once it's running.

Get a free expert session

This session is completely free, with no commitment.

Day 62–4 hoursYou, then your team

Check the numbers, then bring your team in

First, you check the foundation against numbers you already know. This is where trust starts, and where it starts paying off.

Then you open it up to everyone who makes decisions. If they can ask a colleague a question, they can ask Moodree.

First · Check

Last month's revenue against your accounting
Customer and deal counts against your CRM and shop
The one number you've always doubted

Anything that doesn't match gets explained or fixed on the spot.

Then · Share

How to ask, in plain language
How to ask for a new number, source or rule
How approving a change works

Nobody needs SQL, and nobody has to become "the data person".

First questionA first question: how is our website data connected to the CRM? The answer covers what you can answer today, and what you can't yet.

What changes for each roleSales, marketing, finance, and whoever used to pull the numbers

Sales

Waits for the pipeline export before the Monday call.

Asks "which deals have gone quiet for 30 days?" before the call starts.

Marketing

Copies numbers from four ad dashboards into one sheet.

An AI agent reads the foundation every Monday and shifts the ad budget toward the campaigns that bring repeat customers. Marketing reviews the split and approves it.

Finance and operations

Reconciles revenue by hand at the end of every month.

Starts from one revenue number that already matches the books.

Whoever used to pull the numbers

Spends hours every week on exports and "quick questions".

Gets that time back for the work they were hired to do.

You walk away with

A place where your whole team can talk to your data, and trust what it says back.

Answers in seconds, not daysThe question that used to wait for Monday's spreadsheet gets answered while you're still thinking about it.
Numbers you've checked yourselfRevenue and counts reconciled against your own accounting and CRM.
Every number explainedA plain-language meaning for each one, matched to the KPI matrix from Day 1.
Everyone can askQuestions get asked in the meeting, and answered before it ends.

Day 7As long as you likeYou

Ask your five questions

Remember the five questions you wrote down on Day 0? Ask them. In plain language, in your own words.

You get the same answer every time, whoever asks. And if someone in the meeting asks "where does that number come from?", the answer is one click away.

AskAsk a question. Moodree checks your data, then shows the answer and how it got there.

You walk away with

Your next leadership meeting starts with decisions, not with a debate about whose spreadsheet is right.

múdry is Slovak for wise. Say it out loud and you get Moodree. Wise is what we want every growing company to be with its data: knowing which questions to ask, and trusting the answers.

We met building data foundations for clients. Tom was running delivery at a data consultancy, and Tereza had just swapped nuclear physics for data. Project after project, we built the same foundation by hand, and kept seeing the same thing: the companies that needed it most couldn't afford a senior data team.

So we built Moodree. It holds everything those years taught us: which decisions matter, and which shortcuts come back to bite you. We believe every growing company deserves numbers it can trust, not only the ones big enough to hire a data team.

We still talk to every company ourselves, and we'll tell you honestly if it isn't the right time. When it is, it takes a week.

Tom Dědek, sales & direction · Tereza Váradyová, product & delivery

Get your data AI-ready in 7 days

Bring your questions and a list of your tools. We'll go through them with you and plan your seven days.

Get a free planning session

This session is completely free, with no commitment.