Customer Service Solutions For B2B

This category was flagged as planned. The flag is out of date. The Forrester Wave: Customer Service Solutions, Q1 2026, went live on 11 March 2026, authored by principal analyst Kate Leggett with contributors Linda Ivy-Rosser, Rich Saunders, and Angela Lozada, and it carries the most radical sentence this market has ever been scored on.

Leggett's verdict: "The role of AI and the CSR flips: AI addresses the majority of the work, while CSRs assist AI."

A naming note before the scorecard. Forrester's title is Customer Service Solutions, full stop. No edition carries a "For B2B" suffix. This article keeps the B2B lens because the Q1 2026 field's fit verdicts skew enterprise and regulated work, and because B2B service is exactly where the flip Leggett describes is hardest to pull off.

The scorecard in motion: The Forrester Wave: Customer Service Solutions, Q1 2026

The evaluation scored twelve providers against thirty one criteria, and the full tier table is public.

Four Leaders. Salesforce, on enterprise scale and the momentum of its Agentforce service suite, with cautions on complex pricing and slower AI-agent value realization. Microsoft, on integration across Azure, Office, and Teams, with cautions on complexity and support in global environments. ServiceNow, on enterprise orchestration and SLA frameworks, with cautions on complexity and cost. And Pegasystems, the standout of the four.

Pega received the highest possible scores in sixteen of the thirty one criteria: customer service desktop, case management and contextual guidance, AI assist tools, customer service operations, coaching and learning, conversational AI, self-service portals, conversation intelligence, process management and automation, answer management, AI agents, global deployment support, vertical offerings, vision, and innovation. Forrester's sentence is the one to remember: Pegasystems "offers all the building blocks for autonomous service." The fit verdict runs to large enterprises with complex, industry regulated processes where customer value management is a top priority. That verdict is the B2B answer in this category, whether or not the B2B suffix exists.

Four Strong Performers. Zendesk, singled out for pretrained intents and real-time quality measurement, with cautions on packaging and scale pricing. Oracle, strong on back-office automation and AI agent tooling, thin on digital experience. Freshworks, with maximum scores in digital channel support, marketplace, and pricing flexibility and transparency, plus the reference numbers every buyer will quote: autonomous AI agents resolving up to eighty percent of queries, a forty percent cut in ticket handling time at Satair, twenty percent better resolution at Big Bus Tours. Creatio, completing the band.

Four Contenders: Intercom, SAP, Zoho, and HubSpot. No Challengers.

The desk that became a supervisory station

Hold the 2026 criteria against the 2019 edition and the product change is visible before any score.

The Q2 2019 Wave, also by Kate Leggett, scored twelve vendors against thirty three criteria. The criteria then were the agent's world: the desktop, case management, knowledge, the queue. The 2026 criteria include AI agents, conversational AI, conversation intelligence, process mining, answer management, and AI assist tools for CSRs. Criteria that did not exist in 2019 now decide the Leaders.

Pega's winning stack shows what the new product actually is. Blueprint turns a natural language workflow description into BPMN diagrams and documentation. Process mining discovers what the processes actually do. The AI agents then run on both. Forrester's "building blocks for autonomous service" is precise: the case management engine orchestrates, the mining layer verifies, the agents execute, and the human's new job description is written by what remains.

Leggett's own buyer guidance names the shift. Look for tightly blended AI and CSR experiences with context-preserving handoffs. Look for measurement and optimization frameworks to manage AI performance drift. Look for agentic execution with orchestration, guardrails, and policy control. None of those three requirements existed as a purchasing criterion seven years ago.

What seven years did to the 2019 field

The continuity between editions is unusually clean: same author, same field size, twelve providers, and a movement chart any buyer can read in five minutes.

Salesforce held its Leader seat from 2019 to 2026. Oracle dropped from Leader to Strong Performer. Microsoft, Pegasystems, and ServiceNow all promoted from Strong Performer to Leader. Zendesk held Strong Performer. Freshworks promoted from Contender to Strong Performer. SAP fell from Strong Performer to Contender.

The rest of the 2019 field is simply gone: Verint, bpm'online, Appian, and SugarCRM are not in the 2026 evaluation. In their place came Creatio, Intercom, Zoho, and HubSpot, none of which existed in the 2019 field.

Same author, same field size, a different product. The vendors that moved up are the ones that rebuilt around the agentic flip. The vendors that moved down are the ones that did not. The vendors that vanished are the ones that waited.

The promise is scored, the value is cautioned

Here is the honest limitation, and the report itself supplies it.

Salesforce, a Leader, carries the caution of slower AI-agent value realization. Zendesk, a Strong Performer, is cautioned on packaging and scale pricing. ServiceNow, a Leader, is cautioned on complexity and cost. The scorecard rewards agentic architectures and then, quietly, warns that the value of those architectures is arriving slower than the demos suggest.

That gap is the whole game for B2B buyers. The exception layer, the cases the AI hands back to humans, is exactly the layer B2B service lives in: complex, regulated, customer-specific, contract-bound. An AI agent can resolve the simple tickets in any industry. A B2B service operation is the remaining percentage, and the current scorecard has more reference evidence for the eighty percent that gets automated than for the twenty percent that must not be.

The workforce question follows from the same gap. Leggett says the CSR now assists the AI, correcting automation and coaching a digital workforce. For a European B2B buyer, that sentence is not a trend. It is a works council agenda item, and the vendor's answer about what happens to the human team is part of the purchase.

Three questions for a B2B shortlist

Who supervises the agents, and what happens when a case comes back? Ask each vendor to design your exception layer on paper: which cases route to humans, who corrects the automation when it errs, who coaches the digital workforce. In regulated B2B work, the exception layer is the product.

Where is the AI drift measured, and who fixes it? Leggett's criteria include measurement frameworks for AI performance drift. Ask for the drift dashboard before you ask for the ROI slide. A vendor that cannot show its own agent fleet's drift data is selling a promise its own operations have not stress-tested.

What does agent value look like at eighteen months, in your industry, at your scale? Salesforce's caution about slower value realization is the report telling you to ask this question. Demand a reference with an agent fleet at production scale in your sector, and ask for the cost per resolved case, not the deflection rate.

The flip has happened on paper. AI addresses the majority of the work, and the human assists. The Q1 2026 Wave is the first scorecard of that world. B2B buyers should read it as a warning with rankings attached: the vendors that score highest on the flip are the vendors betting your exception layer can be made smaller. Whether it can is the one question the report cannot answer for you.

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

A companion piece published since covers this from the adjacent category: Customer Service Solutions, where forrester's newest evaluation split into an even four-four-four-zero tier table, the emptiest bottom rung this series has recorded, which means the ranking has stopped doing the shopping for the buyer.