Three markets each solved part of the job, and none served a facilitator inside a company.
I run workshops and user tests as part of my work, and the results lived on sticky notes, whiteboard photos and spreadsheets.
In the room, people wrote on sticky notes. Afterward I photographed the whiteboard and retyped the results into a spreadsheet. Before committing to a direction, I audited the tools that claimed to replace that process.
| Market | Who sells there | Entry price | What was missing |
|---|---|---|---|
| Live engagement | Mentimeter, Slido, AhaSlides, Kahoot, Poll Everywhere | About $12 to $19 a month | The tools sell the same polls, quizzes, Q&A and word clouds, and compete on brand and customer logos. Kahoot reads as classroom software to a corporate buyer. |
| UX research | Maze, UserTesting | Free tier up to enterprise contracts | Strong research tools with no live presentation product, priced and run separately from workshops. |
| Jury selection | Trial consulting firms | $10,000 to $75,000 per case | Manual consulting work. I found no modern self-serve software. |
That gap set the strategy. Mentimeter claims more than 500 million users, so competing as the best live poll tool would mean competing on brand recognition, which a solo product cannot win. None of the seven competitors I audited offered live engagement, UX research and jury selection in one account. That became the opening.
”How might we let a facilitator collect group input live, without turning the meeting into a quiz show?”
One session engine with four products built on it.
Participants join in the browser with a 6-digit code or QR code, with no app and no account. The facilitator runs the session from a desktop app, and a second screen shows the room the question and the live results.
Every prompt is a typed block. The library grew beyond quizzes and polls to cover the exercises I kept running by hand, such as dot voting, Kano and SWOT.
Multiple choice
True/False
Yes/No
Poll
Dot voting
Rank voting
Planning poker
Likert scale
Rating
Semantic differential
Matrix
2x2 quadrant
Kano model
SWOT analysis
Journey curve
Card sort
Word cloud
Open-ended
Short answer
Q&A board
Figma prototype
Image
Text
Section
Try it. Three questions, three question types, written for someone hiring a product designer. Your answers stay in your browser.
Single choice · Hiring signal
What most separates exceptional product designers from good ones?
The four products share one account, one session lifecycle and one set of question types. Each targets a different buyer.
Workshops
For independent facilitators and training leads. Structured exercises such as dot voting and retrospectives, with results a facilitator can hand to a client. From $8.99 a month billed annually.
CrowdWorks
For hosts of town halls and conferences. Games, polls and Q&A on a big screen, joined from any phone.
UserWorks
For designers without a dedicated researcher. Preference and prototype tests with Figma embeds, A/B variants and click heatmaps. A $12.99 a month add-on.
JuryWorks
For litigation teams. Juror questionnaires with Likert scoring, case prep and exportable reports. A $179.99 a month add-on, sold through demos.
Choosing a position, then building the product to match it.
I started with sketches of the participant flow, because that screen had to work for a stranger holding a phone.
The May 2026 audit tested the positioning against seven competitors. It concluded that Answerama should not try to be the best live poll tool. It should sell one account that covers live engagement, research and jury work, and compete on price where that holds. Single User starts at $8.99 a month billed annually, and the three tools whose prices I recorded charge between $12 and $19.
The audit also found that I could not match the competitors’ customer logos. It recommended publishing pricing openly and leading with product demos instead of borrowed statistics, and the storefront follows that.
| What the market assumed | What I decided |
|---|---|
| Engagement means games, with points, countdown music and avatars. | People engage when they can contribute. Games stay as one mode among several. |
| Each job gets its own tool, with polls in one place, quizzes in another and workshops somewhere else. | One session model serves every job. Every session moves from Draft to Active to Ended, whatever the type. |
The second decision shaped the code as well. My August 2026 audit of the first version counted 18,000 lines in the desktop app and 15 separate participant pages in the mobile app, one for each question type. I replaced the per-type pages with a registry of question-type modules and a single participant renderer. Adding a type no longer requires a new page in every app.
Decisions along the way.
Decision 1: Joining without an account
Every login step costs a facilitator minutes in a live room. Participants join with a 6-digit code or QR code and no app. Facilitators and admins still sign in with accounts, and Team accounts can use SSO. The free tier allows 50 participants per session, so a facilitator can run a real client session before paying.
Decision 2: Keeping the games without the game show
Quiz mechanics work, but they look out of place in a work meeting. Quizzes remain one mode, and the product uses a muted green palette, plain type and starter templates for retrospectives, Q&A and polls.
Decision 3: Three ways to write the same session
Presenters building ten or more questions with forms stalled on repeated clicking and field focus, and raw text gave no preview of what participants would see. The composer now switches between a document editor with slash commands, a card deck for reordering and attaching media, and plain markdown (.answer.md) for bulk import and AI-generated sets. In testing, people preferred the document view.
Decision 4: Keeping the phone out of the way
In testing, people looked down at their phones instead of at the room, because the phone repeated the prompt and added decoration. I cut the phone to the controls needed to answer and left the question and results on the shared screen.
Decision 5: Staying connected on bad Wi-Fi
Corporate Wi-Fi handoffs and sleeping phones dropped connections and sent people back to the join screen, which lost their answers. Each participant now has a token stored on their device. After a drop, the page reconnects in the background, restores their state and shows the current question.
Decision 6: A navigation that can hold more than one product
As the user testing, event and jury tools arrived, the permanent left sidebar filled up and took width from the editing canvas. I replaced it with a dropdown in the top left. The apps it lists depend on the modules an account has, so the same control handles plan gating, and each add-on can have its own navigation without touching the core app.
Compose with the document, card or markdown editor.
Decision 7: Pricing for an individual and for an IT department
The product had to work for a facilitator paying with a personal card and for a company that requires single sign-on. The tiers differ mainly in participant cap and in the administration features a larger account needs.
| Tier | Price | Participants per session | What it adds |
|---|---|---|---|
| Free | $0 | 50 | Polls, surveys and Q&A, plus API access and the MCP server. |
| Single User | $11.99 a month, or $8.99 a month billed annually | 250 | Quiz, brainstorming and retrospective modes, custom branding, CSV and PDF export, and team management. |
| Team | $39.99 a month, or $29.99 a month billed annually | 500 | SSO and SAML through WorkOS, SCIM provisioning, a custom domain, and a dedicated account manager with an SLA. |
A session never stops accepting people at its cap. Everyone can join and every answer is stored. Results, reports and exports include the first 50, 250 or 500 participants by join order, depending on the plan, and the rest are stored and shown as locked until the account upgrades. Locked answers from Free sessions are deleted 30 days after the session ends. I chose this over a hard cutoff because locking people out mid-session would hurt the facilitator, who is the paying customer. There is no overage fee, and accounts that need more than 500 participants talk to sales.
JuryWorks costs $179.99 a month and the User Testing Suite costs $12.99 a month. The plan is to land with an individual who runs a free session, then expand to a team that needs shared administration, a custom domain or SSO.
Decision 8: Hosting that stays cheap as sessions grow
I moved the data layer from Supabase to Cloudflare D1. Every API request was paying for a network hop to a second provider, the row-level security was unused because the Worker handles authorization, and Supabase Pro would have added $25 a month on top of the $5 Workers plan. Each live session runs in its own Durable Object with WebSocket hibernation, so idle sessions cost almost nothing, and the plan projected about $5 a month in fixed infrastructure cost.
Decision 9: Auditing production before inviting anyone
In July 2026 I audited the live deployment before sending the first invitations. Login and registration silently failed because six client call sites read the token from the wrong field of the API response. The deployed participant app still pointed at localhost, so nobody could join a session. A content security policy also blocked the host’s WebSocket connection. I fixed all three and ran the full loop from sign-up to exported results before inviting anyone.
What exists now.
- A desktop host app, a phone-based participant app, a presentation display and an admin portal, running on Cloudflare Workers, Pages and D1.
- Billing through Stripe, single sign-on through WorkOS and transactional email through Resend.
- Rankings and summaries are calculated when a question closes, and results export to CSV, JSON or PDF.
- A REST API with API keys and an MCP server, so a user’s own AI editor can create and edit question sets.
- Add-ins for Teams, Zoom and PowerPoint are scaffolded and can be sideloaded. They are not yet published in the vendor marketplaces.
I don’t have adoption numbers to report. The audit recommends publishing measured figures, such as participants per session and response rate, once enough sessions have run.
What I learned.
Deciding where each question type ends. Every interaction I could build had to be weighed against keeping the phone usable on a 375px screen and against the cost of maintaining one more type.
The business structure and the product structure needed to match. Keeping the session engine separate from the add-ons made the pricing tiers, the navigation and the codebase easier to extend.
Publish the Teams, Zoom and PowerPoint add-ins in their marketplaces, and add Jira and Linear integrations so ideas from a brainstorm become tickets without retyping.