evaluagent review, pricing and verdict

Contact-centre QA platform that auto-scores every human and AI agent conversation against your own scorecards, with prices published; founded in 2012.

Best for
QA manager
Not for
Speech analytics researcher

Should you use evaluagent?

Pick evaluagent if you want AI scoring on every conversation, coaching that agents engage with, and a price you can see before the demo. It also scores your chatbots on the same scorecard. Skip it if you need research-grade speech analytics. It is a Leader on G2's Summer 2026 QA Grid.

One scorecard for human agents and AI agents, with fabrication detection for the bots and separate seat and per-conversation pricing.

Scorecards, calibration history and agent habits built inside evaluagent are tedious to move, though its published seat prices also make it easy for a rival to undercut in a quote.

Also for

  • Contact-centre team leader
  • QA managers in in-house contact centres
Show 3 more
  • Support operations leads adding AI agents
  • Compliance teams at UK financial services firms
  • Team leaders who coach agents from QA results

Main trade-off

Analytics depth stops short of a full speech analytics platform.

evaluagent pricing

PlanMonthly priceWhat it includes
AutoQM & Improvement$35/moVoice transcription and multi-language support, Custom scorecards, Automated QA scoring on 100% of conversations, Coaching workflows and performance dashboards, evaluagent Context Engine, SSO, MFA and role-based access
AutoQM + Conversation Intelligence$65/moEverything in AutoQM & Improvement, Reason for contact and intent, Spotlight and summarisation, Insight topics and theme detection, Sentiment analytics, Predicted xNPS, xResolution and xCSAT
Pay per conversation (AI agents)CustomPer-conversation rate built around your volume, Same features as the seat plans, For AI agents and variable workloads

AutoQM & Improvement from $35 per user a month; AutoQM + Conversation Intelligence from $65 per user a month. AI agents are priced per conversation instead of per seat, and volume and longer-term discounts are negotiated with sales.

Before you commit

  • AI agent conversations priced separately per conversation

What evaluagent does

evaluagent is contact-centre quality assurance software from the UK, founded in 2012. It scores every conversation against a team's own scorecard, explains qualitative marks, detects invented answers from AI agents and adds sentiment and predicted satisfaction on its upper plan. G2's Summer 2026 Grid placed it fourth of 57 QA products.

  • AutoQM

    Automated QA scoring on 100% of conversations against custom scorecards, with manual review and calibration sessions alongside.

  • SmartScore line items

    The AI scores qualitative scorecard items, such as empathy, and records its reasoning for each mark.

  • Fabrication detection

    Flags AI agent answers that hallucinate or invent information, part of its AI Agent Observability module.

  • Predicted customer metrics

    Estimates satisfaction, resolution, NPS and customer vulnerability for every conversation on the Conversation Intelligence plan.

Show all 6 features
  • Auto-publish and auto-fail rules

    Routes scored conversations automatically, so a compliance breach reaches a supervisor without waiting for a reviewer.

  • Seats for people, conversations for bots

    Human agents are licensed per seat and AI agents per conversation reviewed, on the same feature set.

Works with

  • Genesys Cloud
  • Salesforce
  • Zendesk
  • Intercom
  • Amazon Connect

Capabilities

  • Contact center qa
  • Auto qa
  • AI agent observability
  • Conversation analytics
  • Agent coaching

Strengths and limits

Strengths

  • Both seat plans show a starting price on the pricing page, while Observe.AI, Level AI and CallMiner publish none.
  • On G2's Summer 2026 data, its ease-of-setup score is 93 percent against a category average of 90.
  • Its homepage lists SOC 2 Type II, ISO 27001:2022, Cyber Essentials Plus, GDPR and HIPAA, with EU or US hosting.

Limits

  • Reviewers summarised by G2 Learn (read 22 November 2026) say AI transcription occasionally needs manual checking and call uploads sometimes stall.
  • The published prices are starting points: billing period and minimum seats are not shown, and AI agent pricing is a custom per-conversation rate.
  • Analytics are aimed at QA teams, not at research teams mining millions of calls.

How Hokai rates it

  • Ease of use8.0 / 10
  • Features8.0 / 10
  • Value for money8.0 / 10
  • Support8.0 / 10

evaluagent alternatives

ToolPrice fromFree planRuns onRating
evaluagent This page$35/moNoWeb4.0
ScorebuddyCXPick evaluagent if you want published per-seat prices and AI agents scored per conversation; pick ScorebuddyCX if you want a learning system built into the QA tool.QuoteNoWeb3.9
CallMinerPick evaluagent if a QA team runs the programme; pick CallMiner if analysts need categories across every channel for compliance and research.QuoteNoWeb3.7
Observe.AIPick evaluagent for QA-first scoring with visible prices on any stack; pick Observe.AI if you also want voice AI agents and automated coaching plans.QuoteNoWeb3.8

evaluagent facts

Works offline
No
Minimum skill level
Basic
Price band
Mid-range
Maker founded
2012
Compliance
SOC 2 Type II · ISO/IEC 27001:2022 · Cyber Essentials Plus · GDPR · HIPAA
Total funding
$20M growth investment from PeakSpan Capital (2022)

Data and privacy

Proprietary

Questions about evaluagent

How much does evaluagent cost per agent?

Seats cost from $35 a user per month on AutoQM & Improvement, or from $65 with Conversation Intelligence added. Bots are charged by the conversation, at a rate set for your volume. The billing period is not stated, and large teams or longer contracts get discounts.

Can I try evaluagent before buying?

There is no free plan. evaluagent offers a proof of concept after a successful demo, run on your own conversations and scorecards.

Why does evaluagent price AI agents per conversation?

Its pricing FAQ says human agents have a fixed capacity, so seats make sense, while AI agents handle unpredictable volume. Paying per conversation means paying only for what you need reviewed.

What is xVulnerability in evaluagent?

It is one of the predicted metrics on the Conversation Intelligence plan, listed beside xNPS, xResolution and xCSAT. evaluagent scores it from each conversation, so a vulnerable customer can be spotted without a survey.

Which contact-centre platforms does evaluagent connect to?

For calls, its pricing page lists Aircall, Amazon Connect, Five9, Genesys Cloud, NICE CXone, RingCentral, Talkdesk and Zoom Contact Center. On the ticket and CRM side it names Freshdesk, HubSpot, Intercom, Kustomer, Salesforce, Zendesk and Zoho.

About evaluagent

evaluagent is a British contact-centre quality assurance company founded in 2012 by Jaime Scott, now chief executive, Michelle Dinsmore and Alex Richards, all three former contact-centre operators. G2 lists its headquarters as Middlesbrough. It launched AI scoring in 2018, took a $20 million growth investment from PeakSpan Capital in 2022, and in 2026 extended the platform to evaluate AI agents alongside human ones.

Read the full background

The product starts from the QA team's own scorecard. AutoQM scores every conversation against it; SmartScore lets the AI mark qualitative line items such as tone and gives its reasoning; auto-fail rules route a breach straight to a supervisor; and calibration sessions keep human reviewers in line with each other and with the AI. For bots, fabrication detection flags answers that invent information, and the AI Agent Observability side tracks missed intents and customer frustration. The upper plan adds conversation intelligence: reason-for-contact detection, topics, sentiment, summaries, and predicted scores for satisfaction, resolution and customer vulnerability.

It fits in-house contact centres and support teams that want a QA programme agents actually engage with. G2's Summer 2026 summary of reviews singles out how agent-friendly it is and how quickly it is set up. It plugs into Genesys Cloud, NICE CXone, Five9, Talkdesk, Amazon Connect, RingCentral and Aircall for calls, and Salesforce, HubSpot, Zendesk, Freshdesk and Intercom for tickets, with EU or US hosting.

It is not a speech analytics engine for research teams: a buyer who needs categories across millions of calls should look at CallMiner. Reviewers also say AI transcription occasionally needs a human check. A QA team that wants a built-in training library can compare ScorebuddyCX, and a contact centre that wants AI agents and coaching plans from one vendor can compare Observe.AI. Hokai lists it with other customer support tools.

Our verdict: evaluagent is the QA-first choice for a team that wants visible seat prices and bots graded on the same scorecard as people.

Sources and guides

More AI Tools on Hokai

  • ConvinIndian contact-centre AI suite with automated QA on calls, chats and email, coaching and an LMS, plus AI phone agents; built on its own 7B-parameter LLM.
  • Level AIContact-centre QA and customer-insight platform that scores calls, chats, emails and bot conversations on its own models; Five9 agreed to buy it in Nov 2026.
  • Verint QM IntelligenceAutomated quality management from Verint, now merged with Calabrio: grades up to 100% of calls and chats, from people or bots, and explains every score.
  • Agentforce for ServiceSalesforce's AI layer for Service Cloud: guided plans, drafted replies and summaries for reps, plus AI agents. Included only in the $550 Max edition.
  • Amazon ConnectAWS's cloud contact centre, now sold as Amazon Connect Customer, with AI agents included in a $0.038-a-minute voice rate and no seat licences.
  • AssembledSupport operations platform: an AI copilot that drafts replies and wrap-ups inside your helpdesk, plus AI agents and workforce scheduling. Launched 2020.