Skip to content

What’s new

Product updates.

New features and fixes, listed by release date. Send feedback to .

October 16, 2026

A minimum group size for anonymous surveys, and eNPS over time

  • A survey can hide results for small groups. Turn on the minimum group size under Settings, choose 3, 5, 7 or 10, and any summary of fewer people is left out: the whole survey until enough have answered, a filter down to one team, a cross-tab row or column (small ones are merged into one group when together they are enough, and the table's totals count only what is shown, so a hidden group cannot be worked out from them), a segment comparison, the research measures, AI summaries of written answers, study trends and every shared report link, which never receives the numbers at all. It is free on every plan and applies as soon as it is saved
  • It hides summaries, not rows. The people in your workspace who can open responses and exports still can, and the setting says so
  • An NPS question can be an Employee NPS, asked about your workplace and labelled eNPS everywhere its results appear. The Employee NPS entry in the question picker sets it up, and the engagement, pulse and exit interview templates now start with one, and with small groups hidden
  • NPS and eNPS questions show their promoters, passives and detractors as one bar, and, when a survey has answers across two weeks or more, the score week by week (or month by month, past ten weeks). A period with too few answers for the minimum group size shows a dash. Study trends follow an NPS question by its score now, not its average
  • Fixed: the NPS on the Questions tab counted only 10s as promoters, so it read lower than it should whenever anyone answered 9. Shared reports, the Report tab and weighted results were already right

October 14, 2026

Conjoint and reaction time: choices between whole offers, and the IAT

  • Conjoint asks people to choose between whole offers, each a level of every attribute you set (a price, a plan size, a kind of support), several times, with “None of these” if you want it. Everyone gets their own random choices, spread so each attribute's levels come up evenly. The results fit a logit model: how much each attribute matters, and each level's pull, on one scale
  • Reaction time asks people to sort words left or right as fast as they can, by key (E and I) or by tap, and keeps each word's side and time. The results give every word's sides, its median time, and how often it went where you say it belongs
  • The same question runs an Implicit Association Test: seven short parts, the pairings in a random order for each person, a red X and a wait for the right key after a mistake, scored with the standard improved D algorithm. The results show the mean D score and how many people fall in each conventional band either way, with those excluded for answering too fast
  • Both are free to build and preview, and need Pro to publish, like the other research question types

October 14, 2026

Card sorting, click maps, highlighting, voice answers and five-second tests

  • Card sorting asks people to put cards into your groups, or, if you allow it, into groups they name themselves. The results are a placement matrix: for every card, the share of people who put it in each group, with the groups people named merged however they capitalised them
  • Click maps ask people to click on a picture you upload or link. The results draw the clicks as a heat map over it, count them in areas you mark, and time the first click. With one click allowed it is a first-click test, ready-made in the question picker. The picture is one control for the keyboard too: the arrow keys move a marker and Enter places it
  • Highlighting asks people to mark the words of a passage they like and the words they do not, or simply the ones that stand out, by tapping words or dragging across a phrase. The results shade the passage word by word and list the phrases marked most
  • A file-upload question can record a voice answer instead. The browser asks for the microphone only when Record is pressed, the recording plays back before it is saved, and it plays in Responses. Anyone who cannot record can upload a recording. It works on the published survey, in embeds and on preview links
  • A page's reading material can be shown once, for a set number of seconds, before the page's questions: a five-second test. It waits for Show it, counts down, and never comes back, not even on Back. It is free
  • Card sorting, click maps, highlighting and voice answers are free to build and preview and need Pro to publish, like the other research question types

October 14, 2026

Kano, price sensitivity, pairwise comparison and Q-sort, as questions

  • A Kano pair asks about each feature twice, with it and without it, on the five standard answers. The results place every feature as a must-have, a performance feature, an attractive one or one nobody minds, with its satisfaction and dissatisfaction coefficients. It is free
  • Price sensitivity asks Van Westendorp's four questions about one offer, in your currency and with its name in the wording, and holds each price above the one before as people answer. Prices are read however they are typed: 9.99, 9,99 and 1,000 all mean what they say. The results draw the four curves and the acceptable price range
  • Pairwise comparison asks which of two someone prefers, over every pair or a random few per person, with an optional “about the same”. The results give each option a share of preference from every pair answered, so options compared against strong rivals are not marked down for it
  • Q-sort asks people to place statements on a fixed grid from most agree to most disagree, a few at each end and most in the middle, by picking a statement and then its row. The results show where each statement landed and how much people disagreed about it
  • Price sensitivity, pairwise comparison and Q-sort are free to build and preview, and need Pro to publish. All four are translated into the twelve languages surveys run in, read back on the review screen and in Thread, and come out in exports, Sheets and emails as plain sentences

October 14, 2026

Sliders that run continuous, faces, addresses, times and date ranges

  • The slider (it used to be called the opinion scale) can move in steps or run continuous. A continuous slider is a visual analogue scale: the answer is kept to one decimal, and you can hide the number on the thumb so people place it by feel. A slider can carry a unit and a label under its middle, and the results chart a continuous one in ten ranges with its average
  • Ratings can be faces, running from unhappy to happy, for children, patients and anyone answering quickly on a phone
  • Date questions can ask for a time, a date and a time, or a range with a start and an end, which checks that the end does not come before the start
  • A new address question asks for the street and the city, and, if you keep them, a second line, a region, a postal code and a country. Browsers fill it in from a saved address, the country list is named in the respondent's language, and the results count the answers by country
  • Short text can ask for a website and check that one was typed, with the web keyboard on a phone. The signature pad can take a sketch instead, on a larger pad with undo
  • The add-question picker is grouped and searchable, and has ready-made Yes / No, Website, Time, Date range, Country, US state, Faces and Sketch questions. Each is an ordinary question type already set up, so its results and exports work as they always have

October 5, 2026

A quota-full exit for panel studies

  • A full quota can now send people to its own link with its own code: the over-quota link that sample vendors give you alongside complete and terminate, so the panel counts them as over-quota rather than screened out. Set it under Settings → Screening & quotas. Surveys without one keep sending people from a full quota by the screen-out exit

September 30, 2026

A new Editorial, new themes, and lighter answers

  • New themes, each a four-color palette you can see in the theme picker: Plume, now drawn as a palette like the rest, Honey, Cornflower, Iris, Lagoon, Admiral, Azure and Slate, and the dark Charcoal. A theme can now color the buttons, question card, headings and text from its palette, not only the page and accent. Surveys that used a previous theme keep their colors
  • Editorial is a magazine page now: the page’s title in huge capitals with its number beside it, each question under a hairline with a bracketed label, answers as ruled rows that switch on when picked, and your pictures framed in an arch beside the questions (each page can have its own)
  • Answers, scales and fields no longer carry dark outlines on any template: they are soft surfaces, and a picked answer keeps its colour
  • Split stage on a phone keeps a band of its panel above the questions and one below them, with the progress in the lower one
  • A new progress placement: a thin bar straight across the very top of the screen
  • Split’s panel can stand on the left or the right, its drawing changes on every page and can drift slowly, and the Fluted template’s glass can drift too

September 29, 2026

A blog

  • Ten articles on the questions that come up most: writing unbiased questions, choosing question types, how long a survey should be, raising response rates, open-ended questions, NPS, CSAT or CES, employee pulse surveys, planning skip logic, conversational or classic, and testing a survey before you send it
  • More to come, with an RSS feed for anyone who wants them as they land

September 29, 2026

A clearer Design tab

  • The Design tab has three parts, in the order you decide them: Layout for the shape of the page (template, display mode, header, progress, reading width), Style for the paint (a look to start from, colours, the page background, type, question cards) and Brand for who is asking (logo and name, the words on every page, links, the badge). Every setting has one place; the page background, the brand colour and progress each used to be set in two or three
  • The display mode moved from Settings to Design → Layout, next to the template that can decide it
  • Colour roles fold away under the brand colour, and applying a look says what it replaced and offers an undo
  • Pro is marked on each setting that needs it rather than on whole panes. The display mode stays free; the logo, the name and the brand colour need Pro to publish
  • The Glass question card is now called Frosted, so it is not confused with the Fluted template's glass

September 29, 2026

Sturdier surveys and a faster site

  • A reload, the Back button or a closed tab no longer sends a respondent back to question 1: their answers and place come back in the same tab
  • A submit that is retried or clicked twice is saved once, and a respondent whose quota filled while they answered leaves by the screen-out exit instead of the completion one
  • Completion codes and completion links are given only once a response is complete, and each experimental condition's code only to its own participants
  • Answers to questions a respondent's path skipped are no longer stored, scored or counted in quotas, and logic that compares a number question with a number now matches
  • Two browser tabs, or two teammates, can no longer overwrite each other's questionnaire: the builder notices and asks which version to keep
  • Respondents can upload files again on Pro surveys with a file question
  • The home page no longer animates what is off screen, public pages load less code and show sooner, and exports of large surveys no longer slow down as they grow

September 29, 2026

Smaller type, and backgrounds made plain

  • Survey text is a little smaller on every template: questions, answers and descriptions are 2px under their old sizes. Typed answer fields stay at 16px on phones and tablets, so the page never zooms in on them
  • A Small text size, a step below each template's own (Pro)
  • The page background is now a choice of Color, Gradient or Image. Gradients start from five ready-made pairs and either end can be changed; an image uploads with one button or comes from a link (Pro)
  • The Form template is retired. It was a quieter take on Editorial, and a survey that used it now shows Editorial

September 28, 2026

Surveys that keep working mid-fieldwork

  • People who decline consent can go straight back to the panel: set a decline redirect and code on the consent screen, such as a Prolific No consent path, which asks for the study's return and frees the place. Without one, a panel participant who declines is told how to return the study, in the survey's language
  • A survey's webhook signing secret is readable by owners and editors only. It used to be stored where viewers could read it too
  • The Analysis tab is worked out in the database: cross-tabs, segment comparisons, weighting, Kano, Van Westendorp and scale scores, the panel payouts card and the methods paragraph read counts instead of every answer, so a large study no longer risks timing out there. Weighted results on the Questions tab work the same way. The panel payouts card now counts each participant's first response, as the bonus and reject files always have
  • Grid, ranking, constant-sum and semantic-differential questions are counted cell by cell, so the Questions tab, the Report tab and shared results pages stay the same size however many people answer
  • Google Sheets sync keeps every row when responses arrive faster than Google accepts them. Rows go to the sheet in batches, within seconds as before, and a batch Google turns away is sent again later instead of being lost
  • The Results API lists a response 30 seconds after it is completed, so a sync that checkpoints with since can no longer step past a response that was still saving
  • Prolific bonus lists pay submissions. Prolific's bulk bonus box takes submission IDs, which arrive on the study link as SESSION_ID, and refuses participant IDs, which is what the list used to contain. A participant whose link had no SESSION_ID is counted on the Panel payouts card, to pay from Prolific directly
  • Panel IDs copied from the Quality tab are separated by commas, the way Prolific's Bulk approve box takes them
  • Step-by-step guides for running a study on Prolific and on CloudResearch Connect: the survey settings, what to enter on the panel's side, testing every exit, and approving, rejecting and paying bonuses, with screenshots of each setting
  • CSV and XLSX exports of any size. The file is prepared in the background, with progress on the button, and downloads when it is ready; an export left running finishes anyway and is offered on the results page for seven days
  • The weekly CSV email sends a download link when the file is too large to attach, and one large survey no longer holds up the others
  • Submissions to surveys with quotas or one response per panel participant no longer wait on each other, so a panel launch with many people finishing at once goes through without queueing
  • Publishing changes to a live survey no longer turns away people who are part-way through it. Answers from anyone who opened an earlier published version are accepted, and each response keeps the version it answered
  • A participant who reloads the page stays in the same experiment arm and sees the questions in the same order. Both used to be drawn again on every load. Test mode (?test=1) still starts fresh
  • The Results API answers 503 when the database does not answer. It used to return an empty page, which a sync client reads as being caught up. Answers are now fetched in chunks, so a long survey's page is never cut short
  • Emails from the template list carry an unsubscribe link. It asks you to confirm, then takes you off the list
  • Slack and HubSpot deliveries leave the queue many times faster. The queue used to send five a minute across every workspace
  • Google Sheets sync tries an append once more when Google is rate-limiting or briefly unavailable, and a row that still fails is now recorded as a failure on our side instead of as sent
  • API uptime is recorded: /api/v1/status is checked every four minutes, and the results appear on the Results API page. The endpoint had been answering 503 to every monitor since it shipped

September 26, 2026

Slack, HubSpot and a report you can share

  • Connected delivery on Pro: post each completed response to a Slack channel (the survey title, a dashboard link and the response ID; answers stay in Plumeform), or create and update HubSpot contacts matched by email, with name, company and phone mapped from your questions. Workspace owners set it up under Settings → Delivery, can send only the completions that match an answer condition, and see recent deliveries, with Retry on any that failed
  • A Report tab in results: choose the charts, write findings and a note on each chart, set a date range and an answer filter, and compare against a baseline period. Saving updates the survey's shared results link, and Print / save PDF gives you a copy for a meeting
  • Delete your account yourself from Account & team. Type the workspace name to confirm, and your account, the workspaces you own and their surveys, responses and uploaded files are removed, and any subscription is cancelled at once. Other owners' workspaces stay as they are

September 23, 2026

Seven ways for a survey to look

  • Templates, on Pro: the shape of the respondent’s page, picked before the colours. Classic stays free and unchanged. Editorial sets the questions like a printed page, serif, with the numerals in the margin. Fluted glass floats a frosted card on deep reeded glass. Split stage puts the question on one side and the answers beside it. Form is an application rather than a survey — labels left, fields right, sections ruled, and it prints. Gallery gives half the page to a full-height picture beside your questions, and the halves swap sides as you go. Thread reads as a conversation, and always asks one question at a time, because that is the layout rather than a setting
  • Six template names have been retired into the templates above, because each was a version of one of them rather than its own idea: Manuscript, Spotlight, Tactile and Kiosk became Editorial; Ledger and Instrument became Form. A survey already published on any of them keeps working and takes its new page automatically — nothing to change and nothing to republish
  • Each layout has its own transitions: a slower entrance for Editorial, sideways movement for Split and Thread, and minimal motion for Form. Animations avoid bounce and respect reduced-motion preferences
  • Thread is a conversation you can watch happen. The next question is typed out and arrives as a message, its answers follow a beat behind it, and the last few exchanges stay on screen while the page follows them down. Answering sends your reply; tap any answer you have already given to go back and change it. Nothing reloads between questions, or between pages
  • Six more themes: Sage, Clay, Cobalt and Rose on light pages, Slate and Ember on dark ones. Fourteen in all — the first eight leaned warm and light, with no blue or green at all
  • Your name beside your logo. Either half brands the page on its own, so a workspace with no image can still sign its surveys. Choose whether the logo is landscape or square, how its corners are rounded, and how big it sits
  • Upload a logo instead of hosting one: PNG, JPEG or WebP up to 512 KB. Background images can be uploaded too, up to 1 MB. The URL field still takes an address on your own site
  • Every template is checked against every theme before it ships — contrast is tested for all seven against all fourteen one-click themes, and question text, answers, small print, progress and footer all clear the AA standard. Each template is also checked at phone and desktop widths for anything that overflows sideways
  • Sticky headers are opaque now, so questions pass under them instead of showing through, and they take the colour of the top of your page rather than a fixed pale bar
  • Dragging to reorder a ranking question is smooth. The row follows your pointer exactly and the rows it passes move once, instead of the whole list shuffling while you hold it
  • The Share panel shows the embed on a pretend website — browser chrome, a page around it, and your survey where the snippet would put it. Inline and plain iframe sit in the page; popup and launcher open for real. The survey in the preview cannot be answered, so nothing lands in your results
  • Adding a link per channel works. It could report a failure while writing the link up to four times, and it now says why when it cannot
  • The favicon is the plume. It had been the framework placeholder since launch, which is what Google was showing beside our search results

September 22, 2026

Pictures and video with a question

  • The surveys list shows each survey's week: seven bars of completed responses, the count and the finish rate, beside what it is doing (Live, closes in 3 days, opens 1 Oct, Draft, Closed). Sort by Busiest this week
  • Folders: move a survey to a folder from its menu, then filter the list by folder. A folder exists while a survey is in it
  • Press Ctrl+K (⌘K on a Mac) anywhere in the dashboard to jump to a survey's results, builder or share panel, or to any page, by typing
  • Methods appendix on every results page: generate a summary of fielding dates, study design, response counts, completion rate, median time, quality checks, exclusions, and weighting. Review and edit it for your report
  • Response review: on the Individual view of results, each response has tags, a note for your team, and Exclude with a reason. Excluded responses leave results, cross-tabs, segments, weighting, exports and the shared results page like any quality flag, come back under Show everything, and are never deleted. Tags and notes travel into CSV and XLSX exports. Review from the keyboard: J/K move, X excludes or restores, T tags, N notes. A time strip under each response shows where the seconds went, with answers under three seconds in red
  • Pre-flight in the builder: under the publish bar, a running check of the survey as you edit. It stops a publish for anything that would break the survey (a logic rule pointing at a deleted question, a scale that ends before it starts), flags what is worth fixing (duplicate options, an empty page, a piped answer that no longer exists, text that will not read on the chosen background, a loop in the logic) and points at wording worth a look (leading questions, two things asked at once, absolutes), with the estimated time, question count and reading grade. Every finding jumps to its question. On Pro, Ask AI to review the wording gives a methodologist's read of every question with a rewrite to paste
  • Links per channel: give each place you share a survey its own link (a newsletter, a poster, a partner's site) from the Share panel, with a QR code to download. Every link counts its own views, starts and completes, and the source travels with each response as the via field, in results and exports alike
  • Any question can carry a picture, a GIF or a video: paste an image link or a YouTube, Vimeo or Loom link (or a video file) under the description, add the alt text, and choose whether it sits above the question, beside it from tablet width up, or behind it as a full-bleed backdrop. Works in both display modes
  • Ranking questions are reordered by dragging the grip on each row; the arrows stay for keyboards and screen readers
  • The slider question follows your finger smoothly and shows its value above the thumb without moving anything below it
  • Anything that changes on hover eases into it, across the site and the surveys
  • The same header on every page, the dashboard included: the section links sit in a centred capsule with the page you are on filled in

September 21, 2026

A new front door

  • Free surveys now close by sending respondents on to plumeform.com: after the thank-you screen a thin bar in the survey's accent fills for seven seconds, with a Stay on this page link, and the page moves on. Never for embedded surveys, panel surveys with a completion code, or surveys with their own redirect; Pro surveys are unaffected
  • The live demo closes on the survey's own results: responses, NPS, completion, time to finish, and one card per question with your answers marked
  • The Results API keeps microseconds in cursors and the since filter. The second page of a paginated pull could repeat the last row of the first, and polling with since returned the newest response again on every poll
  • New workspaces start on the free plan: 100 responses a month, every question type except file upload, both display modes, jump logic and CSV export, with nothing that expires. Pro is $15 a month, or $12 a month paid yearly, when a study needs the research pack or the volume. The week of Pro that new workspaces used to get is retired
  • Pay yearly now checks out the yearly plan. The button used to start the monthly checkout
  • Creating an account says why a password was refused. Passwords that appear in known data breaches are turned away, and the form used to answer with a generic error
  • 1% of every Pro payment goes to removing carbon dioxide from the atmosphere, through Stripe Climate
  • The homepage opens on one full screen: the statement, then three example surveys you can answer in place. Each ends in sample results you can filter and export, with a link to the template it came from
  • Try an example now brings the example fully into view and puts you on its first answer. It used to nudge the page a few pixels on desktop, where the example already sat beside the headline
  • Account & team: cards fill their row, and two cards side by side always share a height
  • Survey colours always stay readable. Pick a dark background and the questions turn light while the answer boxes keep dark text; pick a pale accent and button text turns dark. Every combination clears the accessibility contrast standard, and the builder warns you when an accent is too close to the background to see
  • Live preview in the builder: a small, working copy of your survey floats in the corner and updates as you change colours, fonts or wording. Nothing in it is stored
  • Embed a survey in your own site, on every plan: inline (it grows and shrinks with the survey, so there is no inner scrollbar and no blank band), as a popup from any button, or as a floating launcher that opens a side panel. One small script on your page; the survey itself still loads no third-party scripts. Send your own values with each response, and hear when it is submitted so you can close the popup or record a conversion. The code is in each live survey's Share panel
  • Workspace theme on Pro: save a survey's look (colours, font, shape, logo, brand kit) as your workspace's own, and every new survey starts with it, whether it is blank, from a template, imported or drafted by AI. It sits first in the theme row as Workspace, so an existing survey takes it in one click
  • Background images on Pro: paste an https image address and it fills the screen behind the survey, staying put while the page scrolls. The questions always sit on a card over it, because text on a photo cannot be trusted to read
  • Clean dataset on Pro: decide once which quality flags make a response unusable, and it is left out of results, cross-tabs, segments, weighting, CSV and XLSX exports, the weekly export email and the shared results page. Nothing is deleted, the rule can change at any time, and Show everything is one click away
  • Exports now match what is on screen: with Hide flagged or the clean dataset on, the CSV and XLSX you download leave the same responses out (the file name ends in -clean). They used to include everyone regardless
  • Shared results pages chart completed responses only. They were also counting answers from test runs, screen-outs and abandoned partial responses under a total of completed ones
  • The Live preview button in the builder is easier to find
  • Themes on Pro: eight one-click looks (Sand, Mono, Lilac, Sunrise, Forest, Midnight, Dusk and the default) that set colours, font and shape together and keep your logo, brand kit and wording. Each one is checked for readable text
  • More control over the look on Pro: gradient backgrounds, soft, round or square corners, a logo size, and the option to put the questions on a card. If a gradient runs from dark to light, where no single text colour could be read on both ends, the card switches itself on
  • Pale accents read properly: a selected answer's label darkens until it is legible, the pinned progress stays readable on dark backgrounds, and links on the page follow the text colour instead of the accent
  • Bot protection on Pro, with no CAPTCHA and nothing for respondents to do. Switch it on under Settings and submissions that never loaded your survey, arrive within two seconds of loading it, or replay an earlier one are refused in the database itself, so scripts that skip the page are stopped too
  • New quality flags in Results and every export. With bot protection on: automated (the browser says software is driving it, a script did the clicking, or answers arrived with no typing or pointer activity) and pasted text. On every survey: duplicate text, for word-for-word identical open answers. As always, flags never reject anything; you decide
  • A double-clicked or retried submit on a protected survey can no longer create a second response
  • Answering and submitting within a fraction of a second, as happens on one-question polls, no longer leaves a phantom in-progress response behind in the drop-off funnel

September 17, 2026

Make a survey look like yours

  • Pro surveys can take one of six curated fonts and a background, alongside the accent color and logo
  • A brand kit for Pro: hide the progress bar, number the questions, reword the Next, Back and Submit buttons, and sign every screen with a footer line such as your lab or company name
  • The closing screen can carry a button that links anywhere, like back to your site or on to a follow-up (Pro)
  • Custom survey links on Pro: rename plumeform.com/s/x7k2… to something readable. Links you already shared keep working and forward respondents with every parameter, panel IDs included
  • Google Sheets sync is now part of Pro, and connecting asks Google only for access to the sheet Plumeform creates rather than to all of your spreadsheets
  • Response alerts go to one address on Free and up to five on Pro
  • Shared results pages carry Plumeform branding, and drop it when a Pro survey has its badge switched off
  • Workspaces on Pro without a subscription that switch off the Plumeform badge now actually lose it on the live survey
  • Deeper results on Pro: cross-tabs now say whether two questions are really related, with a chi-square test, effect size and markers on the cells that drive it
  • Compare a segment: pick one answer, like an age band, and see how that group answered every other question, with a significance test on each difference (Pro)
  • Weighting: rebalance your sample to population targets on one question. Percentages and averages on the results page, and a Weight column in CSV and XLSX exports, use it (Pro)
  • Panel payouts on Pro: download a bonus list ready for Prolific's bulk bonus box and a reject-review list of participants who failed attention checks, and give each experimental condition its own completion code
  • Team seats on Pro: invite up to four people by email as editors, who build and publish, or viewers, who see results and export. Viewers get a read-only builder, and if a workspace drops to Free its teammates wait until it is back on Pro
  • Belong to more than one workspace? Switch between them from your account page
  • Dashboard pages always open at the top. The account, results and builder pages used to keep the previous page's scroll position
  • The contact form opens above the page footer everywhere; on the terms, privacy, security and What’s new pages it used to sit behind it
  • Surveys with a logo show their progress inside the logo header. It used to float over the header and cover the logo once respondents scrolled
  • Each new survey page opens at the top straight away instead of scrolling up, and pressing Next with a required question left blank brings that question into view

September 16, 2026

A new look, a fixed publish gate, and email that looks like us

  • The site has a new design: an animated gradient hero, a liquid-glass header and dialogs, rounded surfaces and pill buttons everywhere, and a gradient footer. The dashboard, builder and results pages follow the same language
  • Workspaces given Pro without a subscription can now publish surveys that use Pro features. The database gate was checking the billing plan alone, and the app hid the reason behind a generic error; publish errors now say exactly what needs Pro
  • Live events: an audience answering from one venue Wi-Fi is no longer rate-limited. The per-connection submit brake is ten times more generous
  • Every email Plumeform sends now looks like Plumeform, including sign-up confirmations, magic links, password resets and the security notices sent on our behalf
  • Fewer speculative fetches on page load: nav and footer links no longer prefetch a dozen routes you may never open
  • Moving between pages no longer fades the screen out and back in while the next page loads. The page you are on stays put, a thin line at the top shows when the next one is taking a moment, and every page opens at the top
  • The app stays warm: a quiet ping every few minutes keeps the server ready, so the first click after a lull no longer waits several seconds
  • Pages you have already opened come back instantly instead of reloading behind a skeleton, and quietly refresh in the background when they are more than a few seconds old. Anything you change still shows up right away
  • Asking for the template pack now really sends it: all 31 templates attached as a readable file and as Plumeform documents, plus a one-click link that opens each one in the builder

September 12, 2026

A smaller free tier, and two quiet emails

  • The free tier is now 100 responses a month; Pro is unchanged at $15/mo for 10,000
  • Two new email notifications: a reminder about an unpublished draft and a notification when its first response arrives
  • Surveys whose logic loops back on itself now always terminate, and the Logic tab warns you about the loop
  • Billing state is written only by Stripe, and repeated webhook deliveries are de-duplicated

July 18, 2026

Results API, multi-language, longitudinal studies

  • Results API with bearer keys and keyset pagination (Pro)
  • Multi-language surveys with per-locale translations and localized survey chrome (Pro)
  • Longitudinal studies: link waves by participant, retention funnel, wide linked CSV export (Pro)
  • Classic-mode Back button and soft-required questions with a review screen

July 14, 2026

Panel research, end to end

  • Prolific / Connect / MTurk IDs captured automatically, with completion codes and redirects
  • One response per participant, enforced server-side
  • A/B experiment arms with balanced automatic assignment
  • Test mode (?test=1) with one-click purge, MTurk preview handling, per-question response timing
  • Panel launch kit in the share screen with a pre-flight checklist

July 11, 2026

Quality-of-data release

  • Scoring and quizzes with outcome screens
  • New question types: image choice, opinion scale, signature — then maxdiff, constant sum, semantic differential
  • Cross-tabs and AI summaries of open-text answers
  • Email notifications for new responses and respondent confirmations
  • Google Sheets sync

July 6, 2026

The research pack

  • Informed consent blocks, attention checks, screening with quotas, question and option randomization, minimum-time locks on questions
  • AI survey drafting: describe the study, edit the draft
  • Survey import: paste an old survey (or a whole multi-wave study) and get an editable draft

June 2026

Foundations

  • Two display modes per survey: one-question-at-a-time and classic form
  • Branching logic, answer piping, live charts, CSV/XLSX export, webhooks
  • 250 free responses a month; Pro at $15 for 10,000