July 2023, No Extensions: Universal Analytics Gets Its Sunset Date

Google announced on March 16 the date we anticipated without being able to name. Universal Analytics stops processing data on July 1, 2023. Fifteen months from the announcement, the measurement platform that has been the web’s default instrument for a decade goes read-only. The properties themselves retire on a schedule ending in deletion. And, the clause doing the heaviest work, historical data is non-transferable. UA history does not import into GA4, because the models measure with different rulers. You cannot buy last year’s data next year. Start collecting now.

First, the earlier calls, checked honestly. In October 2020 we wrote ‘a formal sunset announcement within roughly two years.’ Seventeen months: close enough. It predicted the runway would ‘feel generous and prove tight.’ The math below is the early verdict. And it advised dual-tagging at once, because ‘the runway is the only year-over-year comparison you will ever have.’

Readers who took that advice are now building their second year of GA4 history. They may skim the rest of this post with the particular satisfaction we try to make routine. Everyone else gets the countdown math, which is unforgiving in one specific place, and the work list, which is longer than the migration wizards suggest.

The countdown math, and the place it bites

The math that makes this month urgent rather than eventually-important: year-over-year comparison requires a year of data, and the year has to come from somewhere.

A team that begins GA4 collection this spring enters the July 2023 cutover with a year-plus of new-model history. The seasonal baselines. The enrollment-cycle comparisons. The giving-season patterns that reporting actually runs on. All present in the ruler the future will be measured in.

A team that defers to next spring enters the cutover with weeks. It faces its first post-UA reporting season with no comparable prior period in the new model. And it cannot purchase the missing history at any price, because it was never collected.

Every month you wait is compare-to history you can never get back. That is the entire case for urgency. It is arithmetic rather than advocacy, and it is why the advice is ‘now, not next quarter’ even though the deadline is fifteen months out. The deadline is when UA stops. The effective deadline for a painless transition was, functionally, some months ago, and the second-best time is this week.

The real work, beyond the tag

• Conversion and event reconstruction, deliberately: the auto-migration tooling maps UA goals to GA4 events mechanically and flatters nobody. The real work is the taxonomy: what interactions matter, named and parameterized, rebuilt as considered configuration. The teams that did the taxonomy exercise are configuring. The rest are re-deriving their measurement vocabulary under a countdown.

• The Ads linkages, which break silently: conversion imports, audience feeds, and bidding signals wired to UA need rebuilding against GA4. The failure mode is quiet: campaigns optimizing against dying signals. The paid-media tier of the migration belongs on the same calendar, this quarter, with its own check pass.

• The history-export decision, with actual warehouse thinking: UA history worth keeping needs exporting before the retirement schedule reaches it. ‘Export’ deserves better than a drawer of CSVs. The reporting views the organization actually reuses (the annual summaries, the campaign benchmarks) extracted deliberately into a queryable home. The warehouse tier for the sophisticated, structured archives for the rest, scoped by what future-you will genuinely ask.

• The stakeholder re-education, starting now: the numbers will not reconcile. Sessions, users, and conversions all differ by design. The dashboard readers deserve the methodology brief before the reports move rather than after. Discrepancy explained in advance is methodology. Discovered afterward, it is a credibility incident.

The sorting, put plainly

The pattern from every forced replatforming applies here with unusual purity, because the deadline hits everyone and the move is required. Early movers get calm and choice. Deferrers get consultancy-market pricing and a scramble. July 2023 will sort the two populations with a precision most deadlines never achieve, because measurement continuity is binary. The history either exists in the new model or it does not.

The market dynamics are already legible and worth naming for budget holders. The GA4 migration industry is assembling. The agencies staffing up, the webinar economy warming. Its pricing will follow the standard scarcity curve into next spring. The teams that convert this quarter’s calm into configured properties, mapped conversions, and briefed stakeholders will spend next year’s peak-demand season watching it from the completed side.

The work is genuinely modest when started early. The dual-tag afternoon, the taxonomy workshops, and the linkage rebuilds, spread across ordinary sprints. It is genuinely miserable when compressed. The same tasks, under deadline, against a consultancy market at peak pricing, with the historical gap already unrecoverable. This paragraph is the boarding call.

The lesson for measurement planning

This closes the question we opened in November 2019. The App + Web beta read as a successor. The 2020 default-status read as a sunset in the drawer. This month the drawer opened on schedule. The sequence proves an old lesson. The plumbing announces what press releases defer, and reading the plumbing buys years of calm prep that reading the announcements cannot.

The same method is already running on the next files. The cookie endgame’s contested mechanisms. The modeled-measurement path GA4 itself walks. The consent-scoped collection the privacy rules govern. The readers who experienced this sunset as configuration (dual-tagged since 2020, taxonomy done, history accruing) are the ongoing proof of the method.

For everyone else, the fifteen-month version of the advice, compressed to its executable core. Create the GA4 property this week. Map the conversions this month. Rebuild the linkages this quarter. Decide the export this year. Brief the stakeholders throughout, in that order, starting with the step that costs an afternoon and buys the history nothing else can. July 1, 2023, no extensions. The date is now on every calendar we can reach.