Digital Analytics Solutions

Web analytics has existed since the late 1990s. Forrester published its first scored evaluation of this market in Q3 2025.

Twenty five years is a long wait for a category that sits on nearly every digital team's desk. The delay is not oversight. It is that until recently there was no single market to evaluate, only four adjacent ones that buyers used together and vendors described differently.

Read the vendor list and the convergence is obvious: Acoustic, Adobe, Amplitude, Contentsquare, Fullstory, Glassbox, Google, Mixpanel, Pendo, and Quantum Metric.

Four lineages, one category

Those ten companies did not grow up in the same business.

Adobe and Google arrived from web analytics, the original discipline built around page views, sessions, traffic sources, and conversion funnels. Their heritage is the marketing question: where did visitors come from and did they convert.

Amplitude, Mixpanel, and Pendo arrived from product analytics, built around events, cohorts, retention curves, and feature adoption. Their heritage is the product question: what are users doing inside the application and does it keep them coming back.

Contentsquare, Fullstory, Glassbox, and Quantum Metric arrived from experience analytics and session replay, built around watching what actually happened: where people struggled, rage-clicked, hesitated, and abandoned. Their heritage is the design question: what is broken and why.

Acoustic arrived from marketing technology.

Four different questions, four different data models, four sets of buyers who historically did not talk to each other. Forrester's definition of the category now spans all of it, describing solutions that collect, analyse, visualise, and interpret quantitative and qualitative data from digital channels to improve product usage, experience design, marketing channel effectiveness, or digital performance.

The phrase worth noticing is quantitative and qualitative. Numbers and observed behaviour in one category is exactly the convergence that made a single evaluation possible.

Inside The Forrester Wave: Digital Analytics Solutions, Q3 2025

The evaluation scored ten vendors against thirty seven criteria, thirty covering current offering and seven covering strategy, using questionnaires, strategy briefings, demonstrations, and customer interviews conducted over roughly five months.

Amplitude placed as one of only two Leaders, taking the highest current offering score of any vendor with maximum marks in twenty one of the criteria, including roadmap, real-time and predictive analysis, AI for insights, AI for assistance, and analysis for digital product performance. It was also recognised as a Customer Favorite.

Forrester singled out two specifics. Amplitude is the only evaluated vendor with a dedicated storytelling tool, and it stands out for bidirectional data warehouse integration allowing analysis against warehouse tables directly within the platform. Its positioning is for product-led organisations where product and marketing teams need close alignment and ease of use is a priority.

Quantum Metric placed as a Strong Performer with maximum scores in real-time analysis and alerting and in mobile application deployment across devices and channels, positioned for customer experience, user experience, analytics, or engineering teams at medium to large enterprises addressing digital friction.

The other vendors evaluated were Adobe, Google, Contentsquare, Fullstory, Glassbox, Mixpanel, Pendo, and Acoustic.

Forrester also published a companion Buyer's Guide drawing on twenty two interviews with enterprise buyers, which is generally the more useful document for shortlisting in a converged market.

Why one tool is not enough

Forrester's first market trend from the evaluation is stated plainly: while these platforms are becoming more comprehensive, no single vendor fully addresses all use cases.

The reason is architectural rather than commercial, and it comes down to how each lineage models data.

Web analytics is session-centric. The unit is a visit, and the questions are about traffic, sources, and conversion within that visit. The model handles anonymous visitors well and struggles with the same person across devices and months.

Product analytics is user-centric. The unit is an identified user with a history, and the questions are about retention, cohorts, and behaviour over time. The model handles the logged-in experience well and is weaker on anonymous acquisition.

Experience analytics is session-recording-centric. The unit is a captured interaction, and the questions are diagnostic. It answers why something happened better than any of the others and scales badly as a source of aggregate truth.

A vendor can add capabilities from the other lineages, and all of them have. What is harder is changing the underlying model, which is why a product analytics platform with web reporting bolted on still feels like a product analytics platform.

The practical consequence is that most mature organisations run two of these, and the operational cost is not the second licence. It is reconciliation: two systems reporting different numbers for the same thing, and a quarterly argument about which one is right.

That argument is usually resolvable and usually not resolved, because it requires someone to define the metric authoritatively across both tools, which is governance work nobody owns.

The warehouse question

Amplitude's bidirectional warehouse integration points at the structural question underneath this market.

Historically these tools collected their own data, stored it in their own infrastructure, and answered questions from it. That produced fast analysis and a proprietary copy of behavioural data that was difficult to combine with anything else.

The composable alternative keeps behavioural data in the organisation's warehouse alongside transactions, support history, and everything else, with the analytics tool querying it rather than duplicating it.

Both models have real advantages. The proprietary store is faster, purpose-built, and works without a data engineering team. The warehouse model joins behaviour to outcomes, which is what turns a funnel report into a revenue analysis, and avoids maintaining two versions of the same events.

For anyone in revenue operations the second is usually the more consequential capability, because the questions that matter cross the boundary. Which acquisition behaviours predict expansion. Which product usage patterns precede churn. Which content preceded pipeline. None of those are answerable inside a tool that only holds digital events.

The evaluation question is not whether a vendor claims warehouse integration, since all of them do. It is whether the integration is read-only export, one-way import, or genuinely bidirectional, and whether analysis can run against warehouse tables without first copying them.

Fewer sessions to analyse

The context this category is entering deserves stating, because it changes what these tools can see.

Forrester reports B2B organic traffic declining between ten and forty percent over the past year, with buyers researching via AI roughly one tenth as likely to click through to a website.

Digital analytics measures what happens on your properties. If a growing share of the buying process happens inside an AI assistant summarising your content alongside three competitors, that portion is invisible to every vendor in this evaluation.

The effect is not that analytics becomes less accurate. It is that it becomes less representative. The sessions you can measure are increasingly the later, higher-intent ones, because the earlier research now happens elsewhere. Conversion rates improve, traffic falls, and both numbers are describing a smaller and differently composed sample of the actual audience.

An organisation reading that as improved efficiency is drawing the wrong conclusion, and an organisation reading the traffic decline as a performance failure is drawing a different wrong conclusion.

The honest position is that a measurement gap has opened, and no vendor in this category can close it because the interactions do not touch their instrumentation. What they can do is measure the portion that remains well, which is genuinely valuable and is a smaller portion than it used to be.

Where the AI capability actually helps

Amplitude's maximum scores in AI for insights generation and AI for assistance point at where the practical value sits, and it is worth being specific rather than general.

The genuine constraint in digital analytics has never been data availability. It has been analyst capacity. These platforms generate more questions than anyone has time to answer, and most organisations have a small number of people who know how to interrogate them properly.

AI for assistance addresses the interface problem: letting someone ask a question in language rather than constructing a funnel definition, which widens who can use the tool.

AI for insights generation addresses the attention problem: surfacing anomalies, drivers, and cohort differences that nobody thought to look for.

The second is more valuable and harder to trust. An automatically surfaced insight is a statistical observation, and in a dataset with thousands of dimensions, statistically notable observations occur constantly without meaning anything. A platform surfacing twenty insights a week is producing a stream that needs the same analytical judgement it was meant to replace.

Forrester crediting Amplitude for being the only vendor with a dedicated storytelling tool is a signal about the same bottleneck from the other end. Finding the insight was never the whole job. Getting someone to act on it was, and that has been the unsolved half of digital analytics since the beginning.

Which is the recurring pattern across analytics categories generally. The instrumentation improves, the analysis improves, and the constraint remains whether anyone changes what they do as a result.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.