Hashing Is Not Consent: A First-Party Data Checklist for Marketers

Posted 24 September 2026 in Insights

Customer data signals being filtered and transformed before entering a marketing measurement system

First-party data is being asked to do more work.

As browser signals weaken and advertising platforms automate more decisions, marketers are connecting website conversions, CRM outcomes, app events and offline sales back into media systems. Google’s September 2026 measurement update reflects that direction: broader Data Manager integrations, enhanced conversions across more products, stronger diagnostics and new tools for modelling and experiments.

The opportunity is real. Better signals can help platforms match outcomes, optimise towards more useful actions and reduce gaps in reporting.

But there is a dangerous shortcut in the conversation: the idea that hashing customer information makes the whole workflow “privacy safe”.

Hashing is a technical control. It is not consent, transparency, data minimisation, accuracy or proof that the use is appropriate. Those decisions still belong to the organisation collecting and activating the information.

For New Zealand marketers, the right question is not simply, “Can we send this data?” It is, “Should this data enter the workflow, what have people been told, and what decision will it improve?”

What enhanced conversions actually do

Google describes enhanced conversions as a way to supplement existing conversion measurement. First-party customer data—such as an email address—is normalised, hashed using SHA-256 and sent to Google, where it can be matched to signed-in accounts that interacted with advertising.

That can improve conversion matching. It also means customer information is being used in a marketing measurement process.

Hashing changes the representation of the data before transmission. It does not automatically make the underlying activity anonymous, remove the need for an appropriate purpose or answer whether the customer would reasonably expect the use.

This distinction matters because implementation often gets treated as a tag-management job. Marketing requests the feature, analytics configures it, legal reviews a privacy policy, and everyone assumes the platform’s security language closes the issue.

It does not. The organisation remains responsible for the information it collects and uses.

Why this is now a marketing issue, not just a compliance issue

The New Zealand Privacy Act’s principles govern how organisations collect, store, use and share personal information. The Office of the Privacy Commissioner says businesses should collect only information necessary for a lawful purpose, tell people why it is being collected and how it will be used, and consider whether a new use is consistent with what people were originally told.

Since 1 May 2026, IPP3A has also added notification obligations for personal information collected indirectly, subject to the Act’s exceptions.

Customer expectations are moving in the same direction. In the Privacy Commissioner’s 2026 survey, 82% of New Zealanders said they wanted more say in how their information is collected and used. Sixty-seven percent were concerned about agencies and businesses using AI to make decisions about people using personal data.

Those figures do not mean customers oppose useful personalisation or better measurement. They do suggest that invisible data expansion is a poor trust strategy.

If a marketing team wants richer first-party signals, it needs a better brief—not just a better connector.

Five checks before customer data enters the measurement stack

1. Purpose: what decision will improve?

Start with the business decision, not the platform feature.

Are you trying to distinguish qualified leads from form fills? Connect offline sales to acquisition sources? Improve bidding towards profitable orders rather than all transactions? Understand which activity creates incremental demand?

Write the purpose in one sentence. Then identify the minimum information needed to support it.

“Improve advertising performance” is too broad. “Help the bidding system distinguish completed policies from quote starts” is specific enough to assess.

A clear purpose also prevents the common pattern of connecting every available field because it may become useful later. Data availability is not the same as data necessity.

2. Transparency: what was the customer told?

Review the promise made at the point of collection.

Did the form, checkout, account process or privacy statement explain that information may be used for advertising measurement, audience activation or matching with advertising platforms? If the proposed use has changed, is it still consistent with what people were told, or is a new authority or updated notice needed?

Do not hide the practical meaning inside generic language about “improving services”. Customers should not need to understand conversion APIs to understand the purpose.

This is also where marketing, privacy and customer-experience teams should work together. A technically comprehensive notice that nobody can understand is not a good customer communication.

3. Minimisation: what can stay out?

Map the fields before configuring the pipeline.

Separate what is required for matching from what is useful for internal analysis. Do not send a rich CRM record when a narrowly defined event and approved identifier will do. Exclude free-text notes, sensitive attributes and fields collected for unrelated operational purposes unless there is a clearly assessed reason to include them.

Minimisation is not only a privacy principle. It reduces implementation complexity, limits the consequences of mistakes and makes governance easier to explain.

The strongest data pipeline is not the one with the most columns. It is the one with the clearest purpose and the least unnecessary information.

4. Quality: is the outcome signal worth optimising?

A privacy-reviewed pipeline can still be a bad measurement system.

If sales stages are inconsistent, refunds arrive late, test leads contaminate the data or teams use the same status differently, automation will learn from a distorted outcome signal. Better matching then creates more confidence in a weak target.

Define the event, owner, update timing and exclusions. Reconcile platform imports against the operational source. Monitor match rates, but do not treat a higher match rate as proof that marketing caused the outcome.

Platform attribution, marketing mix modelling and controlled experiments answer different questions. Use them together rather than forcing one dashboard to provide a complete causal story.

5. Governance: who approves, monitors and proves value?

Name an accountable owner before launch.

Record the purpose, data fields, source systems, destinations, retention settings, customer notice, access controls and escalation path. Decide who can change the mapping and how a new field or use case gets approved.

Then define the evidence needed to keep the workflow.

That could include improved reconciliation, fewer unattributed sales, more stable bidding or experiment results showing incremental value. Vendor-reported uplifts can help build a hypothesis; they are not a forecast for every advertiser.

Set a review date. Platform products, business processes and customer expectations change. A sound implementation can drift into an inappropriate one if nobody revisits it.

A practical implementation sequence

A sensible first-party measurement project can begin small:

Choose one valuable conversion outcome and one decision it should improve.
Map the current collection notice, operational source and proposed destination.
Remove unnecessary fields and document the approved identifiers.
Validate data quality before sending it to the platform.
Run the change with named marketing, analytics and privacy owners.
Compare measurement quality and business decisions before and after implementation.
Review the workflow when the purpose, platform or data changes.

This sequence may feel slower than switching on another integration. It is usually faster than unwinding a poorly understood data flow later.

Better signals need better judgement

The future of marketing measurement will use more first-party data, more automation, more modelling and more experiments. That makes governance more important, not less.

Hashing is valuable. Secure pipelines are valuable. Better conversion matching is valuable.

None of them answers the question of whether the data should have been used in the first place.

The organisations that earn an advantage from first-party data will not simply collect more of it. They will be clearer about purpose, more disciplined about quality and more credible with customers about what happens next.

Sources

This article provides general marketing guidance, not legal advice.