Arena for Amazon Connect
Built for Amazon Connect

A fair, live leaderboard for your contact center agents

Arena turns the activity your Connect instance already produces into points, streaks, challenges and rewards. Quality outweighs speed by design, and everything runs inside your own AWS account.

Available now on AWS Marketplace. Deploys in about an hour with one CloudFormation launch. Latest version 0.38.0, November 15, 2026: The league on wallboards, and admin scripts you can download.

The Arena supervisor console: team leaderboard, scoring mix, flagged agents and challenges

Up and running in an afternoon

No agents to retrain, no spreadsheets to maintain. Arena reads what Connect already records.

Watch the 95-second tour. Captions, no sound.
1

Subscribe and launch

Subscribe on AWS Marketplace. The registration page hands you a one-click CloudFormation launch into your own account.

2

Point Connect at Arena

Two settings in the Connect console: send agent events to Arena's stream, and add the panel as a third-party application.

3

Sync your people

One command creates agent and supervisor sign-ins from your Connect directory, each tied to the right agent and team.

4

Watch the first points land

The first contact an agent handles shows up in their panel within seconds. Evaluations and sentiment follow as Contact Lens produces them.

Everything a floor needs to stay in the game

Four screens, one scoring engine, no self-reporting anywhere.

Agent panel, inside the Connect workspace

Points for today and this week, level, quality streak, badges, and the team leaderboard, all in a tab beside the contact. Agents send kudos, redeem points and see their coaching plan without leaving the workspace.

  • Updated within seconds of each contact
  • Customer sentiment and survey score beside the evaluation average
  • Where you stand in a race and what you would win
The Arena agent panel showing points, rank, streak and the team leaderboard

Supervisor console

The live leaderboard with quality measured on every contact, flags that prompt a conversation rather than a penalty, a scoring mix you can tune, and challenges with real rules: weighted measures, minimum qualifiers, prizes by place, head-to-head and team against team.

  • Flags for agents gone quiet, volume without quality, low customer sentiment, auto-fails
  • Reward approvals in two clicks, with optional automatic gift cards
  • Slack, Teams and email keep the team posted
Supervisor console with leaderboard, scoring mix and challenges

Results report

Is it working? The report compares any period with the one before: evaluation average, customer sentiment, survey score, auto-fails, escalation rate and handle time, for the team and for every agent, with a daily chart and CSV export.

  • Headline sentence a manager can forward
  • Every coaching plan with the numbers before and after
  • Prints cleanly
The results report with period-over-period tiles and a daily chart

Coaching that closes the loop

Any flag opens a plan: the reason, the action agreed with the agent, a follow-up date and a private note. Arena records a 14-day baseline when the plan opens and shows the numbers since, so you know whether it worked before you close it. With the optional coaching assistant, Amazon Bedrock in your own AWS account drafts the action and talking points from the agent's numbers, for you to edit.

  • Agents confirm the plan in their panel
  • Private notes stay with supervisors
  • Before and after in the report
  • Optional AI drafts, inside your AWS account
Flags with Coach buttons, an open coaching plan with before and since numbers, and notification settings

Wallboard for the floor

A display sized for a TV: leaderboard, team totals, the active challenge, kudos as they land, and a moment card when something worth celebrating happens. A supervisor creates a sign-in-free link in one click, names can be shortened for shared spaces, and the screen speaks the floor's language.

The Arena wallboard on a floor display

Quality outweighs speed, by design

A fast contact earns a little. A good evaluation earns a lot. An auto-fail removes points that no setting can soften. Customer sentiment and survey answers count as quality, and never deduct.

Runs in your account

One CloudFormation stack. Agent data never leaves your AWS account. We see one number a day: how many agents were active, for billing.

Nothing self-reported

Points, challenge progress and coaching results come from Connect agent events, Contact Lens and surveys. Agents cannot type their own scores.

Tunable, with guardrails

Supervisors weight quality, productivity and adherence. The console warns when a mix would let volume beat quality.

Scoped by default

Agents see their own team and their own feed. Supervisors see everything. Wallboards can show initials only. Kudos are capped so they cannot be traded.

Five languages

English, Spanish, French, German and Brazilian Portuguese on every page and in every sentence, with numbers and dates written the local way. Each agent picks once; each team picks the language of its Slack, Teams and email messages. Languages

Simple pricing

Billed through AWS Marketplace on your existing AWS bill. No seats to count, no minimum, and the first 10 agents a day are free.

$0.75 per active agent-day

About $16 per agent per month for a full-time agent, a fraction of what seat-licensed gamification suites charge. An agent with no contacts on a day costs nothing that day.

  • Free for good: your first 10 active agents a day are never billed
  • 30-day free trial: up to 25 active agents a day at no charge
  • Every feature, every team, every screen, in five languages
  • Unlimited supervisors and wallboards
  • Updates included; see the release notes
  • Email support with a one-business-day response
What is it worth?
Put your own agent count, attrition and handle time into the ROI calculator. The defaults are conservative and every assumption is editable.
What else does it cost?
The stack's own AWS resources are billed by AWS at normal rates, around $15 a month for a 500-agent center.
How is an active agent-day counted?
One unit for each agent who had scored activity that day, reported once a night from your stack. Nothing else leaves your account.
Can I pilot it first?
Yes. The first 10 active agents a day are always free, and every subscription starts with a 30-day trial in which up to 25 a day are free. The stack can load the last 90 days of history from Connect so the first week is not empty. Larger pilots run outside Marketplace with the billing settings left empty; get in touch. We are also looking for three founding customers: six months free, installed with you, in return for honest feedback.
Where does our data go?
Nowhere. Arena runs in your AWS account and EKPK has no access to it. One number a day goes to AWS Marketplace for billing. The security write-up has the details for your reviewers.
Does it work with platforms other than Amazon Connect?
Everything above the data feed is platform-neutral. Any platform or middleware can send Arena its events through the ingest endpoint, and people keep their history if you switch. Genesys Cloud has a built-in connector: its finished interactions arrive through Amazon EventBridge and are scored as handled contacts. See connecting other platforms.
Is there a mobile app?
The console and report install from the browser on iPhone and Android as an app, with no store. See Arena on your phone.
Contact Lens and surveys?
Optional. Evaluations and customer sentiment need Contact Lens on your instance, billed by AWS as usual. Arena never re-runs analysis.

Questions people ask

Does it work with our existing agent workspace?

Arena is a third-party application in the Amazon Connect agent workspace, which means it opens as a tab beside the contact. It needs the standard workspace; custom contact control panels can open the panel in a browser tab instead.

Can agents game it?

Scores come only from Connect data. Kudos are capped per day, cannot be sent to yourself, and unusual kudos volume is flagged. Races have minimum qualifiers so a one-call wonder cannot win on a single evaluation.

What if an evaluation changes, or an agent is on leave?

If an evaluation is edited, appealed or reviewed after Arena scored it, Arena posts one correcting line for the difference and fixes that day’s figures, so averages and standings follow Connect. Approved time off from Connect scheduling holds streaks, replaces the quiet flag with a time-off note, and shrinks race qualifiers for the days away. Points-based and percentage evaluation forms score the same way.

What do agents see of each other?

Their own team's leaderboard and feed. Coaching plans are visible only to the agent concerned and their supervisors, and the supervisor's private note stays private.

Does Arena have quizzes?

Yes. A supervisor writes a short knowledge check, up to 15 multiple-choice questions, and each agent who passes earns its points once. Agents take it in their panel, get three attempts, and see the explanations afterwards. The supervisor sees who passed and which questions were missed most. With the optional assistant, Amazon Bedrock drafts the questions from a policy text you paste. Arena is not a learning management system: there are no courses or simulations.

Can teams or sites compete with each other?

Yes. The league ranks every team for the calendar month on one measure: points per active agent, evaluation average, survey score, customer sentiment or handle time. Points are counted per active agent, so a bigger team does not win by being bigger. Agents see their team's place in their panel, and only team totals are shown. For a head-to-head between two teams with a prize, use a team challenge. See challenges, rewards and kudos.

Which survey tools work with Arena?

Any that can write the customer's answer onto the Amazon Connect contact, or post it to Arena's ingest endpoint. The quickest is SurveyCX, our own post-contact survey product: press Connect SurveyCX in the console, paste two values into SurveyCX, and every survey answer about an agent earns that agent points. Answers on 1 to 5, 0 to 10 and 0 to 100 scales are all understood, and a contact never scores twice.

Can Arena send gift cards?

Yes, if you turn it on. Give a reward in the catalog a gift card value, and approving it sends the agent a claim link through your own Tremendous account; they pick a card from the ones you allow. Your account pays for the cards, Arena never holds money and adds no fee, and the order carries no agent name or email. There is a sandbox mode for testing without real money. Rewards without a value, like a half-day Friday, are handed over by a person as before.

Does Arena use AI?

Only if you turn it on. The optional coaching assistant asks Amazon Bedrock, in your own AWS account, to draft a coaching plan from an agent's numbers: a suggested action, talking points and a follow-up date. The supervisor edits it and decides. The request carries no name and no contact, points and scores are never set by AI, and Bedrock usage is billed to your AWS account at well under a cent per draft.

Which languages does it support?

English, Spanish, French, German and Brazilian Portuguese on every page, including the sentences built around numbers and names, with numbers and dates written the local way. Each team also picks the language of its Slack, Teams and email messages. The languages article covers what is translated and how to ask for another language.

Can we use our own identity provider?

Yes. Federate it to the user pool the stack creates, or launch with your provider's issuer and audience. The sign-in guide explains both.

How do we get our data out?

It is in a DynamoDB table in your own account. Export it to S3 for your warehouse, or delete the stack and it is gone. Per-agent deletion is built in.

See it with your own team's data

A pilot takes an afternoon to set up and runs inside your AWS account. We will walk you through it.