Customer Analytics Services

One figure from Forrester's current evaluation of this market explains what the category has become.

Nearly ninety percent of Blend's insights engagements include implementation and managed services.

Read that as a statement about the market rather than about one firm. A customer analytics engagement is now, nine times out of ten, not a piece of analysis handed over at the end. It is analysis plus the work of embedding it into something that runs, plus in many cases the ongoing operation of that thing.

The consulting deliverable stopped being the insight. It became the system that acts on it.

What these firms do

Customer analytics services providers apply data science to customer behaviour on behalf of an enterprise that either lacks the internal capability or has it and cannot scale it.

The work spans the customer lifecycle. Acquisition modelling and audience targeting. Segmentation and lifetime value. Propensity and churn prediction. Personalisation and next best action. Pricing and promotion optimisation. Attribution and marketing mix modelling. And increasingly the data engineering underneath all of it, because none of the models work if the customer data is fragmented.

Forrester's Q2 2025 evaluation, authored by Zeid Khater, scored ten providers on current offering, strategy, and customer feedback.

The criteria that ZS took maximum scores in are worth listing precisely because of how unglamorous they are: partner ecosystem, pricing flexibility and transparency, and global delivery strategy among ten in total. Forrester also noted its global delivery scale, with thirty five offices providing cost-effective services through multilingual teams offering round-the-clock support.

Not one of those is an analytical capability. They are operational: can you deliver this at scale, in our markets, at a price we can predict.

That is what a mature services market looks like. Everyone can build a churn model. The differences are in whether the engagement is affordable, staffable, and actually gets used.

Almost everyone gets replaced

The vendor lists across three evaluations are the most striking thing in this category's record.

The Forrester Wave: Customer Analytics Service Providers, Q3 2017, authored by Brandon Purcell, scored ten providers against thirty three criteria: Allant Group, BRIDGEi2i Analytics Solutions, Clarity Insights, CSS Corp, Experfy, Fractal Analytics, Genpact, LatentView Analytics, Mu Sigma, and Tiger Analytics.

The Forrester Wave: Customer Analytics Service Providers, Q3 2021, also by Purcell, scored eleven against thirty criteria: Course5 Intelligence, Evalueserve, EXL, Fractal Analytics, Genpact, Kantar Analytics, Mu Sigma, Quantium, Tiger Analytics, Tredence, and TTEC.

Six of the ten 2017 names are absent four years later. Seven new ones appear.

By the Q2 2025 evaluation, ZS and Blend are prominent, neither of which appeared in either earlier list.

Fractal Analytics is the one name present across all three.

That level of turnover is unusual even for services. It reflects a market with low barriers to entry, where a credible team and a few reference engagements produce a viable firm, and where scale is hard to build because the work is delivered by people rather than by product.

It also means analyst research in this category ages badly as a shortlist and holds up well as a description of what buyers were asking for at the time.

What the buyers were asking for, by era

Read the three evaluations as a sequence and the demand shifts are legible.

The 2017 field is populated by pure-play analytics firms, several of them small and specialist. The framing was transforming large volumes of customer data into value, which was the ambition of that period: organisations had accumulated data and wanted somebody to find something in it.

The 2021 field brings in business process outsourcers, research firms, and larger operations companies. That composition suggests buyers wanting analytics attached to a delivery capability rather than analytics alone.

The 2025 field is dominated by firms that describe themselves in terms of implementation and outcomes. ZS positions around personalisation for large enterprises in specific verticals including healthcare, pharmaceuticals, medical technology, quick-service restaurants, airlines, and retail. Blend positions around embedding analytics directly into client workflows for real-time decision-making, with the ninety percent implementation figure as evidence.

Insight, then insight plus delivery, then delivery with insight inside it.

Why analysis stopped being the product

The reason for that progression is not that analysis got easier, though it did. It is that unimplemented analysis has no value and enterprises worked this out expensively.

A propensity model delivered as a report tells a marketing team something interesting. A propensity model deployed into the campaign platform, scoring in real time, feeding a decision that changes what a customer receives, changes revenue. The gap between those two is engineering and change management, and it is where the entire benefit sits.

Forrester's assessment of Fractal points at the same thing from a different angle, crediting its behavioural science work extending into change management programmes combined with a structured adoption model.

Change management appearing in a customer analytics evaluation is the tell. The constraint on value is not model quality. It is whether the organisation alters what it does.

This mirrors a pattern visible elsewhere in analyst research on adjacent categories. Conversation intelligence customers universally adopted call recording and rated it least valuable, because recording without changed behaviour creates work. Cloud cost tools produce recommendations that nobody with authority acts on. The failure mode is consistent: the analysis arrives, and nothing downstream changes.

Services firms that carry the work into implementation are selling the answer to that failure rather than the analysis itself.

The competitor is the client's own team

There is a structural tension in this category that does not appear in the evaluations.

Every large enterprise buying customer analytics services either has an internal data science team or is building one. That team has opinions about being supplemented, its leadership has budget ambitions of its own, and the political dynamics of an external firm delivering the visible wins are not trivial.

The firms that navigate this well position as capacity and specialisation rather than replacement: the client team owns the core, the provider handles the peaks, the specialist domains, and the engineering the internal team does not want to do.

The ones that navigate it badly end up delivering excellent work that quietly does not get adopted, because the internal team was not invested in it succeeding.

This is also why global delivery scale and pricing transparency score as highly as they do. An arrangement that can flex up and down without renegotiation survives internal politics better than a fixed programme that looks like a permanent replacement for headcount.

What generative AI does to this market

Two opposing pressures are now acting on these firms simultaneously.

The first compresses them. A meaningful share of what junior analytics consultants did, data preparation, exploratory analysis, first-pass modelling, documentation, is exactly what code generation and agentic tooling now accelerate. A services business priced on people-hours and leveraged on junior staff faces the same repricing question as every other consulting category.

The second expands them. Enterprises want personalisation, next best action, and conversational experiences that require capability most do not have, and the demand for someone who can build and run those systems has grown rather than shrunk. Fractal's positioning around multimodal generative AI for next best experience is an example of the expansion.

Which pressure dominates depends on where a firm sits. A provider whose value is analytical throughput is exposed. A provider whose value is domain knowledge, implementation capability, and operating the resulting systems is not, at least not yet.

The vertical specialisation in ZS's positioning is instructive here. Healthcare, pharmaceutical, and medical technology analytics carry regulatory constraints, data sensitivities, and domain complexity that generalist tooling does not resolve, and that specificity is a defensible position in a way that general modelling capability no longer is.

What the churn suggests

Return to the vendor lists. Six of ten replaced between 2017 and 2021. Substantially different again by 2025. One firm present throughout.

For a buyer, that record argues against treating any evaluation in this category as a durable shortlist, and argues for a specific kind of diligence instead.

Ask what happens to the models, the pipelines, and the documentation if the relationship ends. In a market with this much turnover, that question is not hypothetical. A provider that operates the system it built, on infrastructure it controls, using approaches it has not documented, has created a dependency that a change of provider turns into a rebuild.

The same firms making the implementation argument are the ones creating that dependency, which is the honest tension in the ninety percent figure this article opened with. Embedding analytics into workflows is where the value is. It is also how a services engagement becomes structural.

Both things are true, and the difference between a good outcome and an expensive one is usually decided in the contract rather than in the analysis.

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