ListenSuper-admin

Release log

Every version shipped to production. Running now: v0.14.0.

v0.14.0

Current2026-11-09
Changed
  • Every score bar is now centred on zero: negative scores grow to the left, positive to the right. This applies to Explore and to "Your products" on the home page.
  • Score lists are sorted with the most positive first. A "Negative first" switch flips the order.
  • Home, Explore, Interactions and Themes now say clearly what the numbers cover: the time frame (5 Oct – 1 Nov 2026) and how many interactions they are built from.
  • The website now presents the Listen Score as "NPS 2.0: measured, not surveyed", with a side-by-side comparison against survey NPS. The scoring weights are no longer shown anywhere.
  • The interactions list shows whether each conversation lifted or lowered the score, without the underlying point values.

v0.13.0

2026-11-09
Changed
  • An interaction's effect on the Listen Score is now shown in words ("Lowers your score a little", "Lowers your score a lot", "Lifts your score") with the reasons behind it, instead of point values. How the score is weighted stays inside Listen.

v0.12.0

2026-11-09
Added
  • Automation: the queries and requests a bot could answer, ranked by how many contacts it would take, with the agent hours that frees and a typical customer question for each. Bot QA shows how many conversations your bot resolves and where it hands over to a person.
  • Actions: a board of fixes from Planned to In progress to Shipped, each with an owner and the topic it acts on. Shipped fixes show the points they won; open ones show the points still to win.
  • Reports: this week's CX digest (Listen Score and its change, calls QA-scored, average QA score and the three biggest wins) and the reports that run on a schedule, with who gets them.
  • Settings for each project: connected and available sources, the customer journey, privacy defaults (redaction on, data in Mumbai, retention), usage against the plan and API keys.
Changed
  • Reports, Automation, Actions and Settings are no longer marked "Soon". Inbox (phase 2) is the one section still to come.

v0.11.0

2026-11-09
Added
  • Projects inside a customer account: each brand, region or business line gets its own project with its own sources, interactions and Listen Score, so one customer can run several separate data spaces. Acme Living has three: Acme Living India (live), Acme Living Gulf and Acme Outlet.
  • Projects page (under More on phones): every project with its sources, volume and status, the one you're in marked "You're here", and a form to start a new project with a name check.
  • The app sidebar shows the project you're in; tap it to switch.
  • Super-admin customer pages list each customer's projects with their sources and volume, and the customer list counts projects.
Changed
  • Sources and interaction counts now belong to projects and add up per customer.

v0.10.0

2026-11-09
Added
  • Page speed from real visitors: Vercel Speed Insights now measures how fast Listen loads and responds for the people using it (Core Web Vitals such as load time and responsiveness), on desktop and mobile, in the Vercel dashboard. No personal data is collected.

v0.9.0

2026-11-09
Added
  • Customer accounts: each customer is an organisation with a plan, a stage (live, onboarding or trial), seats, connected sources and a team.
  • Team page (under More on phones): see everyone in the workspace with their role, invites still waiting, and seats left. Owners and admins can invite a teammate as Admin, Analyst or Viewer and get a join link to share. Listen checks the email, the role, duplicates and free seats first.
  • Join page: an invite link shows who invited you, to which workspace and with which role, and opens the workspace.
  • Roles explained in plain words: Owner, Admin, Analyst and Viewer.
  • Customers in the super-admin console: every customer with totals for customers, live accounts, people and interactions, and a page per customer with their team, sources and a button to open their workspace.
  • Switch personas in one tap: Pallav is the Owner of the Acme Living workspace, the app shows "Switch to super-admin", the console shows "Switch to customer", and the login page opens either one.
Changed
  • Accounts, teams and invites run on sample data for now. Saving invites and emailing them arrive with real sign-in.

v0.8.0

2026-11-09
Changed
  • Feedback now starts from one "Feedback" button in the corner of every screen, instead of thumbs on the page. It opens a short form: how is this page (thumbs up or down), what it's about, which part of the page, and optional words. Nothing has to be typed to send a report.
  • New "Too slow" choice for reports, and clearer labels with an icon for each kind of feedback.
Added
  • "Which part?" lets customers pick a section of the page, such as "Your products", or tap "Point at it on the page" and touch the exact number, chart or button they mean.
  • Every report now carries the full picture for whoever reviews it: page and title, the part or exact element picked, device and screen size, browser and system, light or dark mode, language and time zone, app version, and any script errors on the page. Customers can see what is sent under "Sent with your feedback".
  • The admin feedback inbox shows the part of the page, the exact element, device, browser and any errors for each ticket.

v0.7.0

2026-11-09
Added
  • Question packs page: your own call QA scorecard answered on every call, not a 2% sample. It shows how many calls were scored, the average QA score, how often a reviewer agreed with Listen's answers, and how many answers a person checked.
  • Each scorecard question shows its pass rate and weight, and the question worth coaching first is marked "Biggest coaching win".
  • Agents are ranked by QA score, each with the one thing to coach next. The best is marked "Top performer".
  • Every call now shows its QA answers: who handled it, its score, each question answered "Yes" or "Not yet", and which answers a reviewer corrected.
  • The pack library lists the packs running on the workspace (customer care QA, core, compliance, logistics NDR, bot QA).
Changed
  • Packs is no longer marked "Soon" in the menu.

v0.6.0

2026-11-09
Added
  • Every interaction shows its own Listen Score, line by line: the points for its type, any repeat-contact penalty and the total, plus how many points fixing the cause would win back. Outbound and excluded contacts say why they are not scored.
  • Themes now show the customer's own words: each theme carries the quote that best captures it, with a link that opens the call at that exact moment, and a link to every interaction that mentions it.
  • The interactions list can be filtered by theme.
  • Sensitive numbers are hidden automatically: payment card numbers (only the last four digits stay), Aadhaar numbers and Indian mobile numbers are masked in transcripts and quotes, and each transcript says how many details were hidden.
  • A new sample call (a delayed bed delivery in Pune) shows the Pune and Nagpur delivery theme and the redaction at work.
Changed
  • Themes lead with the points there are to win rather than a negative score.

v0.5.0

2026-11-09
Added
  • Feedback on every screen: a thumbs up / thumbs down pill sits in the corner of the app (just above the tab bar on phones). One tap opens a short sheet to say what it's about ("Something broke", "Numbers look wrong", "Hard to understand", "Idea", "Love it") and add a few words. The screen, product, version and screen size are attached automatically, so nobody has to explain where they were.
  • Every piece of feedback becomes a ticket that moves from Received to Accepted to Fixed, or is marked Not planned (with a reason) or Noted (for praise).
  • "My feedback" (under More on phones): each customer sees the tickets they raised, how far each has got, and the team's latest reply.
  • Admin feedback inbox at /admin/feedback: how many tickets wait for review, the mix by status, the share of thumbs up, and a list you can filter by status, each linking to its ticket.
  • Tickets are worked by Claude: it reviews new feedback, accepts or declines it with a short reply, builds the accepted ones, and the ticket is marked Fixed automatically when the change ships.
Fixed
  • The admin console's header no longer pushes the page sideways on phones; its links wrap onto a second line.

v0.4.0

2026-11-09
Changed
  • New, modern look: Geist typeface throughout (no more serif), cool neutral greys, an electric violet accent with a cyan "voice" highlight, and numbers set in Geist Mono where they sit in tables.
  • Home page redesigned: grid-and-glow hero, gradient headline, product preview framed like an app window, headline stats, and cleaner sections for how it works, the Listen Score and each team.
  • Buttons are now high-contrast pills; the header is translucent and stays on screen while scrolling.
  • Dashboard tiles, bars and weekly chart restyled to match; dark mode updated to a true near-black.
  • App home rebuilt to be read at a glance and to encourage rather than alarm. A dial shows today's Listen Score and, in green, where your three biggest wins would take it. Next to it: an encouraging status (Healthy, Room to grow or Big opportunity), the change since last week, a four-week trend line, and how many questions a bot could answer.
  • "What's working" opens the page with the topics customers praise.
  • "Where to gain, across the journey": the nine journey stages as a heatmap of points to win. The biggest win is marked.
  • "Your biggest wins": the three topics with the most points to win, each with a ring showing its share of the upside.
  • "How customers feel": one bar headed by the share of happy customers and the share to win back. Ideas customers raised show how often they came up, in large numbers.
  • Mobile first: on phones the app now has a bottom tab bar (Home, Explore, Interactions, Themes, and More for upcoming sections), a sticky top bar, and headline numbers in a two-by-two grid.
  • On phones, the interactions list shows each conversation as a readable block with its type, topic, emotion and points. Explore shows contacts, negative share and top emotion under each row. Nothing is cut off or needs sideways scrolling.
  • Filter chips sit on one line you can swipe on phones, instead of stacking into several rows.
Added
  • Product lines. Every conversation is tied to a customer and a product line (Sofas & recliners, Beds, Wardrobes & storage, Dining, Study & office in the demo), and each line has its own orders and Listen Score.
  • The app follows one hierarchy: company, then product, then the product's customer journey, then function and topic. Home shows "Your products" ranked by points to win; tapping a product (or its chip) shows the same home view for that line alone.
  • Explore now starts at product lines and drills into journey stage, function and topic. Interactions can be filtered by product and show the product on every row.

v0.3.1

2026-11-09
Added
  • Performance tracing in Sentry: every page load and server request on preview deploys, and 20% of them in production, so slow pages and requests show up under Explore → Traces.

v0.3.0

2026-11-09
Added
  • Error monitoring with Sentry: crashes in the browser, on the server and at the edge are reported to the "listen" project, tagged with the version and whether it was production or a preview.
  • A friendly "Something went wrong" screen if the whole app fails, which also reports the error.
  • No customer data leaves with errors: IP addresses and cookies are never sent, and local runs report nothing.

v0.2.3

2026-11-09
Added
  • Full JSDoc documentation across the codebase: a file overview for every source file, and documented functions, components, props, types and fields.
  • Lint rule (`eslint-plugin-jsdoc`) that fails CI when code is undocumented.
Fixed
  • The docs date check no longer fails when a Markdown file is edited twice on the same day. It now checks that each file's "Last updated" date is not older than its last change.

v0.2.2

2026-11-09
Added
  • README badges: CI, release, coverage (100%, enforced by the gate) and production link.
  • `AGENTS.md` with the working rules for AI agents and contributors, and `CLAUDE.md` pointing to it.
  • Docs gate in CI: every Markdown file carries a "Last updated" date, a changed file must bump it, and every PR must add release notes to this changelog.

v0.2.1

2026-11-09
Added
  • Hosted on Vercel: every merge to main deploys to production, and every pull request gets a preview link.
  • `vercel.json` pins the build settings and runs the app in the Mumbai region (bom1), close to Indian customers.

v0.2.0

2026-11-08
Changed
  • New visual identity: Fraunces display type with Inter for the interface, a deep teal brand colour with an amber "voice" highlight, and a waveform logo.
  • Home page hero now shows the product at work: a real call, its tags, and the opportunity Listen found.
  • Listen Score points shown as a visual scale on the home page.
  • Dashboard: the Listen Score tile is highlighted, and weekly scores hang below a zero line so negative reads as negative.
  • Scores use a true minus sign.
  • Navigation marks sections that are not built yet as "Soon".
Fixed
  • Red text for negative scores now meets WCAG AA contrast.
  • Filter and function chips are at least 32px tall for easier tapping.

v0.1.0

2026-11-08
Added
  • Marketing site: home, ROI calculator, security, book a demo.
  • Login page with access to the demo workspace.
  • App home with the Listen Score, journey-stage breakdown, weekly trend, costliest topics and emerging themes, with a view per function.
  • Explore drill-down from journey stage to function to topic.
  • Interactions list with channel and type filters, and an interaction detail page with transcript, tags and evidence-linked pack answers.
  • Themes page.
  • Super-admin release log and a version number on every screen.
  • Test suite with a 100% coverage gate, CI, and automatic GitHub Releases.