Data governance

Everyone has the data. Why does nothing add up?

Q: What should consent management and tracking governance cover?

At least four things:how each consent option maps to each tag;who can add or change tracking, through GTM permissions and publishing review;how long data is kept: event data in standard GA4 can be kept for 2 or 14 months, and for up to 50 months in 360;whether every change leaves a record you can check later.

Start in a meeting you already know. See why the numbers disagree, and what keeps them agreeing tomorrow.

A FAMILIAR MEETING

“That is not the number I have”

Marketing reports 312 conversions this month. The e-commerce backend says 287. Finance closes the books at 240. Nobody made a mistake, yet the meeting spends its first half hour deciding which number to believe.

Same meeting. Three numbers.

The problem is rarely a miscalculation. That “same number” simply never had its origin explained.

STEP BACK ONE SQUARE

Where was this number first recorded?

The ad platform counts purchases within seven days of a click. GA4 records events fired on the site. Finance counts orders that were actually paid. One word, three paths, three birthplaces. Before they can agree, you need to know where each one came from.

01 / SOURCECan this number be traced back?

02 / MEANINGIs everyone counting the same thing?

03 / OWNERWho finds out first when a number is wrong?

04 / SHELF LIFECan this number still be trusted next month?

SMALL THINGS, EVERY COMPANY

Numbers do not stop agreeing overnight

They drift apart one small thing at a time. Your company probably has a few of these.

Same thing?

The field has been in use for three years. Its definition was never written down

Nobody renamed it, because everyone assumed someone else knew what it meant. One definition changed, six reports quietly changed with it, and nobody was told.

Who notices first?

The handover was done. The reasoning never made it across

They calculated the conversion rate and remembered the exceptions. The handover had the passwords, but not the “why it is counted this way”.

Still true next month?

Nobody dares delete that report, because nobody knows who still reads it

Reports outlive the people who built them. This one still refreshes every week on schedule, long after the numbers stopped meaning what they once did.

One clarification on “consistent”: it does not mean forcing different platforms to show the same number. It means knowing where they differ, and why.

START WHERE IT HURTS

You do not need a big plan first

The most useful starting point is the problem your team already feels.

Conversions and orders never seem to agree

Start with a Data Health Check: origin, meaning, owner and lifespan, four signals that tell you where to begin.

Lots of reports. No clear place to look

Organise Data Studio around the decisions people actually make.

Each team makes sense. The pieces do not fit

Work with a data consultant to align goals, definitions and interpretation.

We know the tools. We need better questions

Bring training back to the work your team actually does.

BEFORE THE ANSWER

Can we trace the source?Do definitions agree?Are the limits clear?

AND NOW, AI

AI only amplifies what is already there

AI can turn data into a fluent answer in seconds. Data with unclear origins and inconsistent definitions gets the same fluent treatment: wrong, but better dressed and harder to catch. Platforms like iKala Nexus let a company connect its internal data to its own AI agents. Before that connection, the data has to be trustworthy. That part is what CloudAD does.

Explore our perspective

WHAT WE CALL IT

Data governance

Put simply: everyone means the same thing by the same words, and still will tomorrow. Numbers with an origin, one definition, an owner and upkeep. That is the ground a decision can stand on.

Align the definitions, then read the numbers.

Start with a Data Health Check Explore our work Read COO perspectives

COMMON QUESTIONS

Compliance and first-party data

We are a financial institution in Taiwan, bound by personal data and compliance rules. Can we still measure and attribute marketing results?

Yes. Decide first which data may be collected, where it is stored and who can see it, then decide how tracking is installed. Three practices are common:

send only de-identified events and parameters to analytics tools, never names or phone numbers;
use consent mode so tags follow each visitor's choice;
when internal data has to be matched, do it inside your own environment and export only aggregated results.

GA4 itself does not log or store IP addresses. Each practice should be checked against Taiwan's Personal Data Protection Act and your internal rules.

Third-party cookies are less and less reliable. How do we build first-party tracking?

Start with the signals you control: consistent event definitions across site and app, first-party identifiers after login, and hashed first-party data sent back with conversions, such as Google Ads enhanced conversions and the Meta Conversions API.

Browser limits are already in place. Safari blocks all third-party cookies by default and deletes cookies created by JavaScript after seven days without user interaction. Design on the assumption that browser signals will be missing, and fill the gap with server-side and first-party data.

What should consent management and tracking governance cover?

At least four things:

how each consent option maps to each tag;
who can add or change tracking, through GTM permissions and publishing review;
how long data is kept: event data in standard GA4 can be kept for 2 or 14 months, and for up to 50 months in 360;
whether every change leaves a record you can check later.