Digital analytics

Digital analytics is the collection and analysis of behavioral data from websites and mobile apps: which pages and screens people saw, what they clicked, how they moved through a journey, where they stopped, and which changes moved those numbers. It is the measurement layer that most other digital disciplines depend on.

What does digital analytics actually measure?

Events, and the sequences they form.

At the bottom are events: a page view, a click, a scroll, a purchase, a custom event a developer defined. Above that are sessions and users, which group events into visits and visits into people, using identifiers that are increasingly constrained by browser and privacy rules. Above that are the constructs teams actually report on: funnels, ordered steps with a drop rate at each; cohorts, groups defined by when they arrived or what they did; and retention, whether they came back.

The reporting layer is where most of the value and most of the disagreement lives, because a funnel is a model someone built, not a fact the data contained.

What is digital analytics genuinely good at?

Three things, and they are foundational rather than optional.

Sizing a problem. Knowing that a step loses a third of its traffic, and that the step before it loses almost none, tells you where to spend attention. Nothing else does this as cheaply.

Detecting change. Analytics is the early warning system. A conversion rate that moves without an obvious cause is usually the first sign of a bug, a broken integration or a shift in traffic mix.

Settling questions of fact. How many people use the filter. Whether mobile really is worse. Which channel sends visitors who come back. These are answerable and frequently answered wrongly from memory in meetings.

Who sells digital analytics software?

Google Analytics is the default for most of the market by volume. Adobe Analytics and Contentsquare serve the enterprise end; Amplitude and Mixpanel serve product teams and lean toward event-level and cohort analysis; Heap emphasizes automatic capture; Piwik PRO and Matomo target organizations with strict data-residency requirements.

The practical differences are less about which metrics exist and more about how much modeling work sits between raw events and a usable report, and how much of that work your team can sustain.

What are the limits of digital analytics?

It cannot produce a cause. This is a property of the method, not a gap any vendor can close. Analytics records behavior, and a behavior is consistent with many different reasons. A visitor who abandons a form may have hit a confusing field, decided the price was too high, or been interrupted by their doorbell. The data is identical.

The standard responses to this all refine the "what" without reaching the "why." Segmenting tells you which group abandoned. Drilling into the funnel tells you which step. Cohorting tells you when. None of them tell you what was in the visitor's head, and treating a well-segmented number as an explanation is the most common analytical error in the discipline.

Instrumentation decides what is knowable. You can only analyze what someone thought to track, and tracking plans are written before the questions arise. Most analytics investigations stall on a missing event rather than a missing insight.

Identity and consent constrain the data. Cross-device journeys, third-party cookie deprecation, consent rates and ad blockers all mean the dataset is a partial sample of reality, and the size of the gap is rarely known.

Numbers without a reason produce the wrong fix. A team that sees a drop and cannot explain it will fix the most visible candidate, ship it, and see the number stay flat. That cycle is expensive and extremely common.

How Pulse Insights relates to this

We are not a digital analytics platform. We do not replace one, and we do not want to be evaluated as one.

If your question is how many, how often, which segment or has it changed, buy analytics. Every organization needs that layer, and a program that tries to run without it is guessing about size as well as cause.

Our work starts where the analytics report ends. When a number shows a drop, we help establish why it happened by asking the people it is happening to: a short contextual question triggered by the behavior itself, answered in the moment rather than reconstructed from memory later. The answer is a stated reason rather than an inference, which is why it is specific enough to act on.

Two things make that a different job rather than a competing one. The output is a cause, not a metric. And the response can be delivered in the same session, from a library the client's team approved in advance, so the visitor who told you what was wrong gets it addressed rather than contributing to next month's report.

Analytics tells you the step lost a third of its traffic. Asking tells you it was the returns policy. You want both, and they are not substitutes.

Frequently asked questions

What is digital analytics?

The collection and analysis of behavioral data from websites and apps to measure what visitors did, where they stopped, and how those numbers change over time.

What is the difference between digital analytics and web analytics?

Web analytics conventionally means website measurement. Digital analytics covers websites, mobile apps and other digital touchpoints together. In practice the terms are used interchangeably.

Can analytics tell you why visitors abandon?

No. It can narrow where and for whom with considerable precision, but a behavior is consistent with many reasons and the data cannot distinguish between them. Establishing a cause requires observation or asking.

What is the difference between digital analytics and session replay?

Analytics aggregates behavior across many visitors into counts and rates. Session replay reconstructs individual visits so they can be watched. One tells you how big a problem is, the other shows you one instance of it.

Do you still need analytics if you ask visitors directly?

Yes. Asking produces causes on a self-selected subset. Analytics produces reliable magnitudes across everyone. Deciding what to fix needs both the size and the reason.

Related: Session replay and heatmaps · Digital friction · Voice of the customer · Pulse Insights vs Hotjar

Browse every term in the Pulse Insights glossary.