Analytics & Data
Universal Analytics: Google Rethinks Measurement Around People, Not Pageviews
Web analytics has counted pageviews since the server-log era, and the model is visibly straining. Sessions split across the devices we have spent five years tracking. More and more of what matters is not a page at all. And the customer who researches on a phone, returns on a desktop, and converts by calling remains three strangers and a mystery to the standard toolkit.
At its analytics summit this week, Google admitted the strain formally. Universal Analytics, announced in beta, is a rebuilt measurement foundation rather than a feature release. A new collection protocol. Custom dimensions and metrics. And, biggest of all, the ability to tie a known user’s activity together across devices where your own logins permit it. Three devices, one person. If your reports cannot see that, your budget cannot either.
The pageview era of web analytics is formally ending. Endings of this kind reward the teams that treat them as projects and punish the ones that treat them as toggles. Here is the sorted version. What the new model actually enables, what the move will eventually involve, and what to do during a beta, which is deliberately less than the excitement invites.
What the rebuild actually changes
A measurement protocol, not just a page tag. Universal’s collection layer accepts data from anywhere that can send it. Websites, yes. But also apps, kiosks, and event check-ins. The strategic meaning: your analytics property can become the system of record for digital touchpoints generally, not just browser sessions. Think of the prospect who attends the info session, or the donor who gives by phone. The model finally has somewhere to put reality.
Cross-device users, where identity exists. With a site’s own hashed login ID, Universal can stitch one signed-in person’s phone, tablet, and desktop sessions into one journey. Nothing personal is stored. The three strangers become one prospect. The obvious boundary is the honest one. This works where users sign in, which for most institutional sites means portals, applications, and members. Not anonymous browsing. It is not magic. It is bookkeeping for identities you already hold, and it makes login-gated experiences newly interesting to measurement strategy.
Custom dimensions, the quiet workhorse. The ability to attach your own labels to hits and users, and slice every report by them. Member type. Program of interest. Content category. Campaign cohort. Less glamorous than cross-device stitching. In our judgment, this is the feature most mid-market analytics programs will actually build value on, because it lets measurement speak your own vocabulary.
What migration will eventually require, said early
The beta label governs everything in this section’s tense, but the shape is already visible and worth pricing now.
The tracking code changes. Every property will eventually re-tag. The setup deserves rethinking rather than copying over. Universal’s model rewards deciding what your dimensions, events, and user IDs should be. That is analytics design work, not IT ticket work.
History does not convert. Your years of existing data remain in the old model, and comparisons across the boundary will need care and annotation. The reporting-continuity discipline our measurement series installed for exactly such moments.
And the privacy layer needs adult attention. Cross-device measurement built on your own user IDs touches policies, disclosures, and in some sectors compliance review. Involve those stakeholders at design, not at launch.
None of this is alarming. All of it is a project. The organizations that learned that lesson from the last replatforming will spend a comfortable 2013. The toggle-flippers will call someone like us in 2014.
What this signals about where measurement is heading
Read the announcement through the signal-reading lens we practice, and the direction is legible past the feature list. Google is rebuilding its measurement foundation around people and events, just as device sprawl makes session-and-pageview counting more and more fictional. The same visitor as three strangers is not a reporting nuisance. It is the model failing, and Universal is the admission.
Expect the rest of the industry to follow the reframe. The category’s vocabulary will be ‘users’ and ‘events’ within a few years, on this trend. And expect the identity question to move to the center of measurement strategy. Organizations with signed-in relationships (portals, apps, member systems) will measure journeys their anonymous-traffic peers cannot. That quietly raises the value of every login you operate.
There is also a privacy trend to note with equal weight. Measurement built on your own first-party IDs, disclosed properly, is the durable approach as third-party mechanisms come under pressure. Universal’s design, whatever its beta rough edges, points the same direction responsible measurement was already heading. First-party, consent-aware, built around the relationship rather than the pageview. The plumbing is telling you the strategy again. Read it.
The right approach to a beta, and the homework that pays regardless
The right approach to a beta is straightforward: nobody migrates production measurement onto it yet. Your current setup keeps working, keeps collecting, and remains the system of record until Universal matures. Likely a year or more.
What is worth doing now costs little and transfers entirely. Design new measurement work with the event-and-user model in mind so it ports cleanly later. Inventory the identity question. Where do users sign in? What ID would stitching use? What do our policies say? And begin the unglamorous governance work every measurement transition runs on. Naming rules. A written measurement plan. An owner for the analytics setup.
That homework improved the current setup the last three times we assigned it. It will make the eventual Universal move a scheduling exercise instead of an archaeology project.
Measurement replatformings are announced as features and arrive as migrations. This one has now been announced. Calendar it, adopt nothing in production, and start the design thinking while it is cheap. The beta will firm up on Google’s schedule. Your readiness is on yours.
The pattern holds each time a foundation shifts. The organizations that do the design homework during the announcement era treat the migration era as scheduling.