Medallia

What it is

Medallia is the enterprise experience-management suite built around getting customer feedback into the hands of the people who can act on it: store managers, branch staff, contact-centre supervisors, hotel general managers. It collects surveys, digital feedback, speech, social, and video, runs text and speech analytics over them, and routes the result into role-based reporting, alerts, and case management. Founded in 2001, listed on the NYSE in July 2019, and taken private by Thoma Bravo in October 2021 for $6.4B in cash. Its closest like-for-like is Qualtrics.

The question every buyer is asking in 2026 is not about features. On 22 April 2026 Thoma Bravo agreed to hand Medallia to its lenders in a debt-for-equity swap that wiped out roughly $5.1B of sponsor equity against about $3B of debt; Reuters reported it the next day, by which point two of the lenders were already carrying the loans at 74 and 79 cents on the dollar. The recapitalization was formally announced on 17 June 2026 and closed on 3 August 2026. Medallia is now owned by an investor group led by Blackstone, Apollo, and FS KKR Capital, with its debt reduced by an undisclosed amount and $150M of new capital on the balance sheet. The company is not shutting down, and it is not the same bet you would have made in 2025.

Why it shows up in CX stacks

• It is built for the frontline, not the insights team. Gartner named Medallia a Leader for the fifth consecutive time in the 2026 Magic Quadrant for Voice of the Customer Platforms (announced 12 March 2026, 12 vendors rated), citing Frontline-Ready AI used by close to seven million people weekly. That usage profile is the product: tens of thousands of employees who each see their own location’s feedback, not a dozen analysts in a head-office dashboard.
• Unlimited users are inside the price. Medallia meters on the Experience Data Record (EDR) — all data associated with one interaction between a customer or employee and the company — sold in annual tiers. Its pricing page lists unlimited users, role-based reporting, alerting, and case management as included. For a 40,000-employee retailer, that is the difference between a program the frontline can use and one rationed by seat.
• The generative-AI layer shipped in July. On 14 July 2026 Medallia made Smart Topic Builder, Intelligent Summaries, Smart Response, and Root Cause Assist generally available, with French and German added to the core AI features. The company reports that more than 40% of its top 300 enterprise customers had enabled generative AI by then, a cohort representing over $200M of software ACV.
• It hands insight to agents rather than being one. The 21 January 2026 partnership with Ada pipes Medallia’s omnichannel signal into Ada’s AI agent so it can act on a complaint pattern, following a similar Adobe tie-up in September 2025. If your roadmap puts an AI agent in front of customers, Medallia is positioned as the listening layer that tells it what to fix.

Pricing reality

Nothing is published. The pricing page explains the EDR model and routes to sales; there is no list price, no minimum, no free tier, checked on 18 September 2026.

Vendr’s buyer data puts the median Medallia contract at $13,338/yr across 32 purchases, ranging from $6,739 to $112,044, last updated February 2026. Read that carefully: 32 data points against 317 for Qualtrics on the same platform means Vendr sees Medallia’s smaller, single-program deals, not the enterprise programs that span contact-centre speech and dozens of business units. Treat the Vendr range as the floor for one journey on one channel, not a benchmark for a frontline rollout.

The meter is what moves the bill. Survey-only programs generate modest EDR counts; adding contact-centre speech or digital-behaviour signal multiplies records without adding a single user. Get the EDR definition per channel, the included annual volume, the overage rate, and a renewal-uplift ceiling in writing.

Best for

CX program owners at multi-location, frontline-heavy enterprises — hotel groups, retail chains, retail banks, telcos, and hospital systems — where the job is putting each location’s feedback in front of its own manager within hours and tracking whether they closed the loop. Medallia reports six of the top ten hospitality brands on its generative-AI features within twelve months of launch, which is the segment where its design fits most tightly.

Do not buy it if your feedback lives in support tickets and product channels and the work is reading it, not collecting it — that is an analysis-layer purchase, and the frontline tooling you would be paying for has no one to use it. Do not buy it for a single NPS program at a company under roughly 1,000 employees. And do not buy it if you need agents querying feedback over MCP this quarter: there is no first-party MCP server.

Versus the alternatives

Qualtrics is the other incumbent by install base, the highest-rated Leader in the same Magic Quadrant, and the one with an official MCP server over its Public API. It absorbed Press Ganey Forsta on 18 May 2026, so it is running its own integration. Pick Qualtrics when the contract has to cover research and employee experience alongside CX, or when agent access matters now; pick Medallia when frontline action across hundreds of locations is the job.

Sprinklr is the third remaining Leader and the pick when social and care channels carry your signal and you already run it for social — one vendor fewer, one connector fewer.

Enterpret is the fastest-growing AI-native entrant taking budget off the analysis layer. Pick it when the feedback already exists and the job is classifying it for product and support. Chattermill does the same job where reviews and app stores carry the signal, and AskNicely is the cheaper route when NPS plus frontline coaching is the whole requirement.

If none fit, look at Gartner’s two Challengers, Alchemer and Pisano. Alchemer publishes self-serve plans for small teams, which neither incumbent does, and a one-journey pilot on either tells you whether you need an enterprise program before you sign for one.

Watch-outs

• The owners are lenders, and the product commitment is a press-release line. CEO Mark Bishof has said Medallia will invest over $500M in products and services “over the next few years.” That is a statement, not a contract term, and analysts quoted by CMSWire in May 2026 warned that lender control can favour cost-cutting over R&D; Gartner’s Maria Marino said renewal challenges were already surfacing among Medallia clients. Guard: do not sign three years on the strength of the announcement. Take a one-year initial term, or a three-year term with termination for convenience at month 12 if named roadmap items are not GA by a stated date. Add a data-export clause covering raw responses, transcripts, and hierarchy data in a documented format.
• Announced is not shipped. Insights Assistant, the natural-language query feature shown at Experience ‘26 in February 2026, was absent from the 14 July GA list that included Smart Topic Builder. Guard: build the business case on the July GA set only, run the demo on your own data, and write anything else into the order form as a dated milestone with a price adjustment if it slips.
• The EDR meter punishes success. The year the program works is the year you add speech, chat, and digital signal, and every one of those raises EDR volume. Guard: model year two at double year-one EDRs, and negotiate the overage rate at first contact, when walking away is still an option.
• Agent access is DIY. Programmatic access runs through the GraphQL Query API and the Import API over OAuth 2.0 client credentials; the MCP servers that exist are third-party wrappers, including Zapier’s, which put another vendor in the path of your customer data. Guard: if you want Claude or Cursor reading feedback, build a thin internal MCP server over the Query API on a read-only data-access role, and keep PII detection switched on before any transcript reaches a model.
• Your negotiating position is unusually strong — use it. A vendor that has just changed owners and has analysts publicly flagging renewal risk wants multi-year logos. Guard: price Qualtrics or Sprinklr in the same cycle, even if you expect to stay, and ask which product teams are fully funded for the next 18 months. The answer, or the refusal to give one, is worth more than the discount.