martechdevs has been acquired by Growth and Analytics Partners LLC
martechdevs

AI readiness score for your marketing data

21 questions across the seven checks AI tools depend on: tracking, identity, warehouse models, metric definitions, context docs, access and MCP. You get a score out of 100, a score per check and the order to fix the gaps. It takes about five minutes, runs in your browser and asks for no email.

1 of 7

Tracking accuracy

Do you have a written tracking plan that lists every key event and its properties?
When did you last test your key events and conversions end to end?
Do conversion counts in GA4, your ad platforms and your CRM agree for the same week?

2 of 7

Identity resolution

Do you send the same user ID to your analytics, product and CRM tools?
Can you follow one person from first visit to signup and payment?
Are visits before signup joined to the person once they sign up or log in?

3 of 7

Warehouse models

Where do your marketing, CRM and billing data come together?
Are the joins modelled and tested, for example in dbt?
How is the data people report on refreshed?

4 of 7

Metric definitions

Do marketing, sales and finance use the same definition of revenue and of a customer?
Are CAC, MQL, pipeline and churn defined in writing?
Do reports and AI tools compute metrics from one shared definition, such as a semantic layer?

5 of 7

Context docs

Are your ICP, personas and naming conventions written down?
Are campaign playbooks and brand and compliance rules documented?
Can an AI tool read those docs today, as skills, knowledge files or a project?

6 of 7

Access and permissions

If you connected an AI tool today, would it get its own limited login?
Is access set per team and per tool, with an owner for each system?
Could you list every AI tool and connector with access to your CRM, ad accounts and warehouse?

7 of 7

MCP readiness

Are your core tools (CRM, analytics, warehouse) connected to Claude, ChatGPT or your own agents over MCP?
Is that access read-only by default, with writes switched on deliberately?
Could you switch off an AI tool's access in one place today?

0 of 21 answered

How the score works

The seven checks are the ones in our AI readiness checklist, in the order we fix them. Each has three questions worth 0, 1 or 2 points, so every check counts equally and the total is shown out of 100.

  • 70 and over: Ready for AI on most questions. Connect AI read-only first and test it on questions you already know the answers to. The gaps below are what is left.
  • 40 to 69: Partly ready. AI can answer questions inside single tools. Questions that join tools, such as CAC or pipeline by channel, will drift until the gaps below are fixed.
  • Under 40: Not ready yet. AI answers on this stack would be guesses. Fix tracking and identity before connecting more AI tools.

The seven checks

  1. 01

    Tracking accuracy

    AI answers count events that fired twice or never fired at all.

    Good looks like: A written tracking plan, every event tested end to end, and conversion counts that match the CRM, with any gap explained.

  2. 02

    Identity resolution

    One customer shows up as several people, so counts, journeys and revenue per customer are wrong.

    Good looks like: Shared IDs such as user ID, email and device ID, with merge rules that turn each person into one profile.

  3. 03

    Warehouse models

    Every question starts from raw exports, so each answer joins the data a little differently.

    Good looks like: Snowflake or BigQuery with tested dbt models for customers, accounts, campaigns and revenue, refreshed on a schedule.

  4. 04

    Metric definitions

    AI tools pick one of several definitions of CAC, MQL or revenue, and you can't tell which.

    Good looks like: Each metric defined once, written down, and built into a semantic layer that every report and agent uses.

  5. 05

    Context docs

    Agents guess at your ICP, naming rules and playbooks, so their output needs rewriting.

    Good looks like: A marketing knowledge base for AI: definitions, naming conventions, ICP, playbooks, and brand and compliance rules, shipped as skills and knowledge files.

  6. 06

    Access and permissions

    AI tools inherit an admin login, so one bad prompt or connector can read or change too much.

    Good looks like: Access set per team and per tool, a separate login for AI tools, and an owner for each system.

  7. 07

    MCP readiness

    AI tools can't reach your systems, or reach them in ways nobody can switch off.

    Good looks like: MCP servers on the tools that matter, an internal MCP server for your ops teams, and access you can switch off at any time.

An example result

An illustrative 15-person SaaS on HubSpot, GA4 and Stripe, with no warehouse yet and HubSpot connected to Claude by one admin, scores 26 out of 100: not ready yet. Its fix order starts with tracking and identity, then one agreed definition of a customer, well before more connectors. Why AI answers on marketing data go wrong walks through the same example.

  • Tracking accuracy2 of 6
  • Identity resolution2 of 6
  • Warehouse models0 of 6
  • Metric definitions2 of 6
  • Context docs2 of 6
  • Access and permissions1 of 6
  • MCP readiness2 of 6

Questions about the score

What is an AI readiness assessment?

A check of whether your data, definitions and access are ready for AI tools to use. This one covers the seven checks we run in our AI readiness audit: tracking accuracy, identity resolution, warehouse models, metric definitions, context docs, access and permissions, and MCP readiness.

How is the score calculated?

Each of the 21 answers is worth 0, 1 or 2 points, and each check has three questions, so every check counts equally. The total is shown out of 100. A check with 0 to 2 points out of 6 is marked fix first, 3 or 4 partly there, and 5 or 6 ready.

Do you store my answers?

No. The score is worked out in your browser and the answers never leave it. Our analytics records that the tool was started and the final score, without the answers.

What should we fix first?

Tracking and identity, because every model, definition and AI answer is built on them. Your result lists the checks that fall short in that order, with what good looks like and a first step for each.

How is this different from the AI readiness audit?

The score is a self-check you run alone in a few minutes. The free audit is a 30-minute call where we take your stack through the same seven checks with you, then send a fixed-price plan to fix the gaps.

AI readiness guides from our blog

Read all guides

Book a Free Audit

A free audit of your stack and what it needs. No pitch.

Thirty minutes, and you leave with a plan

We look at what you have, what it is costing you and what to do about it first.

  • 30 minutes
  • Video call, link on booking
  • No obligation

What we'll cover

  • Your current stack and where it hurts
  • Quick wins we can ship first
  • An integration roadmap
  • Timeline, scope and next steps

Calendar not loading? Open it in a new tab or email hello@martechdevs.com.