# Glossary of Sequential terms

> Plain definitions of the terms used in Sequential and in these docs, such as session, verdict and finding. Each entry links to the page that explains it.

> **In short**
>
> - The terms are in alphabetical order, in groups of letters.
> - Each entry gives a short definition and a link to the page that explains it.
> - Each entry and table row starts with its term in bold. Inside a definition, words in **bold** are exact labels from the app.
> - The table **Terms that are easy to confuse** shows pairs that people often mix up.

## Terms that are easy to confuse

| Term | Not the same as |
|---|---|
| **Member**: a person on your team who signs in to Sequential | **User**: a person who uses your product |
| **Session**: one visit to your product | A sign-in session, which Settings → **Security** lists |
| **Masking**: hides text, and the layout stays | **Block**: leaves the element out, so it plays back as an empty box |
| **Retention (data)**: how long Sequential keeps what it collects | **Retention (report)**: the **Retention** page |
| **Role (organisation)**: a member's permission level | **Role (rubric)**: a tag on a rubric question |
| **Rubric version**: the count of rubric saves | **Rubric hash**: a fingerprint of what the judge reads |
| **Event property**: describes one event | **User property**: describes the person. The filter field **Event Property** matches event properties. |

## A–B

- **Access token.** A project credential that an agent or a CI job sends to Sequential. Owners and admins create it on **Agents** or under Settings → Projects → your project → **Access tokens**. Sequential shows it once. Treat it like a password. [Limits](/docs/limits#programmatic-access)
- **Account.** A group of users, such as a customer company. Today, accounts come from Pendo and from the Sequential SDK's `sequential.account()`. [Users and accounts](/docs/users-and-accounts)
- **Agents.** The page in the left rail where owners and admins see which agents are connected: claude.ai, ChatGPT and coding agents. There, they connect claude.ai, create access tokens, end connections and tokens, and see the calls that the agents made. [Your first hour](/docs/quickstart#connect-your-agent)
- **Anonymous id.** The user id of a visitor before your app identifies them. It shows on **Users** like any other user, until the Sequential SDK links it to one identified value. [Users and identity](/docs/users-and-identity)
- **Archive.** The reversible way to stop a project. Sequential refuses requests that use its keys. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Ask.** The **Ask** tab on **Sessions**: you describe a behaviour in plain words. A short or empty result does not mean that nothing like it happened. [Ask: find sessions by describing them](/docs/ask)
- **Attribution parameters.** The marketing query parameters whose values Sequential keeps in stored page URLs. [What Sequential keeps from page URLs](/docs/url-policy)
- **Audit log.** A record of security and configuration changes, with who made each change and when. [Audit log](/docs/audit-log)
- **Autocapture.** Events that an SDK sends without code from you. [Events](/docs/events)
- **Belief.** The name the API uses for a stored judgment (`/api/beliefs/…`). The app says judgment. [Judgments](/docs/judgments)
- **Block.** To leave an element fully out of a recording. It plays back as an empty box. [Masking and redaction](/docs/masking-and-redaction)
- **Breakdown.** A grouping of the events on the **Events** page by one or two fields. [Explore events](/docs/events-explorer)

### More about A–B terms

- **Account:** Sequential identifies an account by an account id. To see accounts, open **Accounts**. An event without an account id belongs to no account. **Accounts** does not list it as a row. A line under the list counts the sessions that have no account. Group calls from PostHog, Mixpanel and Amplitude are not used for accounts yet.
- **Anonymous id:** The SDK creates it. The Sequential SDK keeps its own in the browser's storage. Nothing marks it as anonymous.
- **Archive:** The project no longer shows in the project switcher. Sequential keeps its data, keys and settings. **Restore project** brings it back with the same keys.
- **Ask:** The judge rates how likely each candidate session is to match. Ask does not list sessions that it rates unlikely.
- **Attribution parameters:** Examples: `utm_source`, `utm_campaign` and ad click ids like `gclid`. Every other query value becomes `[redacted]`.
- **Audit log:** It covers changes in your organisation. Examples: saved recording settings, rubric saves and deleted projects. Owners and admins can read it.
- **Autocapture:** The events are page views, clicks and form submits. The Sequential SDK autocaptures by default. It never records what anyone types.
- **Belief:** Background judging makes them. It is on by default for each project. Owners and admins turn it off or on under **Settings** → **Projects** → your project → **Insights**. See also *Judgment*.
- **Block:** Sequential blocks elements with the `data-rr-block` attribute under both **Strict** and **Balanced**. Compare *Masking*.
- **Breakdown:** For example, by country and then browser. On **Events**, it is the group table: set **grouped by** and **then by** in the query card. It shows the event count and the share of all events for each value. Grouped by user, account or session, the events without that id are one row: "(no user)", "(no account)" or "(no session)". That row has no **Filter to this** and no **Watch sessions**.

## C–D

- **Calibration.** A page that uses your team's 👍/👎 on **Watch** to check if the judge is right as often as it says. [Feedback and calibration](/docs/feedback-and-calibration)
- **Class (event class).** One of nine fixed categories that every event gets on arrival. To filter on it, use **Event Type**. [Events](/docs/events)
- **Connection (claude.ai).** The link from claude.ai to one project. An owner or admin makes it with a sign-in and a consent page, not with a token. It shows on the **claude.ai** card under **Agents**, and on the **Access tokens** tab with **via OAuth**. [Connect claude.ai](/docs/connect-claude-ai)
- **Console text.** The words that your app writes to the browser console. Sequential stores them only when **Record message text** is on. It is off by default. [Masking and redaction](/docs/masking-and-redaction)
- **Custom event.** An event whose name Sequential does not recognise as another class, such as `checkout_started`. [Send custom events](/docs/custom-events)
- **Date range.** The start and end dates at the top of each report. A range that you set on one page does not apply to the next page. [Filters, date ranges and saved views](/docs/filters-and-saved-views)
- **Dead click.** See *Rage click and dead click*.
- **Delete (project).** Removes a project with its keys, recording settings and rubric. You cannot undo it. Sequential permanently deletes the project's events within 14 days with the default schedule. [Retention and deletion](/docs/retention-and-deletion)
- **DevTools.** The JavaScript errors, console output, network requests and page performance that Sequential records with a web replay. [DevTools in replays](/docs/devtools)
- **DevTools states.** *Counted*, *not captured* or *not measured*. A missing badge in the **Problems** column does not prove that the session was clean. [DevTools in replays](/docs/devtools)
- **Digest.** The short text version of a session that the judge reads, with at most 255 events. [Judgments](/docs/judgments)
- **Distinct id.** The id that an SDK sends for the current person. Sequential stores it as the *user id*. [Identify your users](/docs/identify-users)

### More about C–D terms

- **Calibration:** The page is under Settings → Projects → your project → **Calibration**. For example: when a question says 90%, was it right about 90% of the time?
- **Class (event class):** Sequential decides the class from the event name. The classes are Page view, Click, Form submit, Identify, Error, Rage click, Dead click, Custom and System. A class means the same for every SDK.
- **Console text:** An example is "payment failed: card_declined". **Record message text** is a project setting. Whether it is on or off, Sequential keeps the console levels, counts and code locations.
- **Custom event:** Usually your own code sends it. Custom events are how your product's milestones reach funnels, flows and the judge.
- **Date range:** The dates are in the date bar. They take effect when you press **Apply**. They are whole calendar days in UTC. The first time you open a page, the range is the last 10 days, today included. After that, each page opens on the range that you last applied there for the project, in this browser.
- **Delete (project):** Sequential can no longer open the project's data. It permanently deletes the events, and the user, session and account records, within 14 days with the default schedule. Judgments and identity links are deleted too. Recordings become unreadable at once, because Sequential deletes their encryption keys. The storage deletes the recording files later.
- **DevTools:** DevTools is the technical context of a replay. Sequential never records the headers or bodies of network requests. You turn each capture on or off per project. All four are on by default.
- **DevTools states:** They say how much Sequential knows about a session's DevTools counts. Only problems get a badge, and console errors never do.
- **Digest:** The judge never watches the recording. Sequential does not keep the digest.
- **Distinct id:** It is `distinct_id` in the Sequential SDK, PostHog and Mixpanel. Before your app identifies the visitor, it is an anonymous id. After that, it is the value that you pass.

### The three DevTools states

| State | What it means |
|---|---|
| *Counted* | Sequential checked the session. Badges exist for two kinds of problem: JS errors and failed requests. Examples: "2 JS errors", "1 failed request". "12+" means at least 12. |
| *Not captured* | That capture was off, or your mobile SDK does not report it. |
| *Not measured* | Sequential never checked the session. For example, a session recorded before counting began. |

The player's DevTools counts console errors, but console errors never get a badge.

When you filter **Sessions** by **With JS errors** or **With failed requests**, the page shows two counts:

- how many sessions "couldn't be checked for this filter" (not captured)
- how many sessions "weren't checked for DevTools signals" (not measured)

### What the digest holds

- One numbered line per event, except System events.
- Pauses.
- A few facts, such as duration, device and, when measured, error counts.

The digest holds at most 255 events. In a longer session, Sequential leaves out the least important events first: clicks, and then custom events.

## E–H

- **Entities.** The old name of the **Accounts** page. An old `/entities` link opens **Accounts**, or the account's page when the link names an account. [Users and accounts](/docs/users-and-accounts#the-accounts-list)
- **Environment.** A project label: production, staging or development. It does not separate data inside a project. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Event.** One thing that happened in a session, such as a page view or a click. [Events](/docs/events)
- **Event name.** The exact name that the sending SDK used, such as `$autocapture`. Sequential never rewrites it. [Events](/docs/events)
- **Event property.** A property of one event, such as `plan` on `checkout_started`. The **Event Property** filter matches one with a type and an operator. [Send custom events](/docs/custom-events)
- **Filter.** Exact-match conditions on an event's fields, in groups that read **All of**, **Any of**, **None of** or **Not all of**. [Filters, date ranges and saved views](/docs/filters-and-saved-views)
- **Finding.** A problem that Sequential counts across sessions: an exception, a failed request, a dead click, a rage click or a dead end. A JavaScript error, a server error (5xx) or a network failure opens one from its first session. Other problems need at least 3 sessions of at least 2 people in the last 7 days. A dead end where people came back with the Back button needs at least 6 sessions of 4 people, and 20 % of the page's sessions. Errors from browser extensions and requests to ad or analytics hosts never open one. With the Sequential SDK, PostHog or Mixpanel, one person counts once, even if they signed in between. Its id reads like F-12. [Findings and fixes](/docs/findings)
- **Fix brief.** The text about one finding that you copy for a developer or an agent. Its quoted blocks are data from your users' sessions. [Findings and fixes](/docs/findings)
- **Fix link.** A link from a finding to the pull request, commit or issue for its fix. [Findings and fixes](/docs/findings)
- **Flow.** The **Flows** report: the order in which users move between a few events that you pick. [Flows](/docs/flows)
- **Forward-only.** Sequential judges each session once, and its stored answers never change. [Customise the rubric](/docs/rubric)
- **Funnel.** The **Funnels** report: how many users complete a series of steps in order. [Funnels](/docs/funnels)
- **Handed off.** A finding stage: someone copied the fix brief or opened a GitHub issue from it. Or a token read the finding or its brief: GET `/api/findings/:n` counts too. [Findings and fixes](/docs/findings)
- **Heat lane.** Marks on the replay timeline where the question with the **Struggle** role placed its probability. [Session insights](/docs/session-insights)
- **Heatmap data.** Where sampled visitors click and how far they scroll on each page of a website, kept as counts with no text. On by default. Owners and admins turn it on or off under **Session replay** → **Heatmaps**. The **Heatmaps** page, under **Explore** in the left rail, shows its click counts and scroll depth, drawn on a page picture. [Privacy at a glance](/docs/privacy-overview#heatmap-data)
- **Heatmap page picture.** A picture of a page, taken from one visitor's masked recording, with the clicks or the scroll depth drawn on it. The **Heatmaps** page shows it below the controls, to all four roles (owner, admin, member, viewer). Anything that the recording shows unmasked appears in it. A separate builder service makes it in the background. Where no builder runs, the page says "Page pictures aren't available right now. The ranked list is complete." and shows the ranked list without a picture. On the scroll map, the message ends "The Scroll depth card is complete." [The page picture](/docs/privacy-overview#the-page-picture)

### More about E–H terms

- **Entities:** Since 13 October 2026, the page is **Accounts**. It lists only accounts. To see people, open **Users**.
- **Environment:** It has one effect: a project created as staging or development starts with replay sampling at 10% instead of 100%.
- **Event:** Another example is your app's `checkout_started`. Sequential stores each event with its class, original name, source, time, page, device, user, session and properties.
- **Event name:** Other examples: `[Amplitude] Page Viewed` and `checkout_started`.
- **Event property:** You can see event properties in an event's raw data. The filter field **Event Property** matches them with a type and an operator.
- **Filter:** An example condition is **Event Type** = `rage_click`. Each condition tests one event at a time. The same builder defines funnel and flow steps.
- **Flow:** It draws the paths as a Sankey diagram. Each user has one path for the full date range. An event counts toward every step that it matches. If the same step comes two times in a row, Sequential keeps it once.
- **Forward-only:** If you change what the judge is asked, the change applies only to sessions that Sequential judges after the change.
- **Funnel:** It also shows the time between steps, and which users left at which step. Sequential measures conversion against step 1. The date range is the only time limit.
- **Heat lane:** By default, that question is "Where they struggled". The lane appears after the **Insights** tab loads.
- **Heatmap data:** Not the same as the heat lane, which marks one replay. It follows the replay period. An owner or admin turns it off with `replay.heatmaps.enabled` in the settings API. That also stops the page picture.

## I–L

- **Identify.** The call that tells Sequential who a visitor is, such as `sequential.identify()`. Two different values for one person make two users. [Identify your users](/docs/identify-users)
- **Identity claim.** The link that Sequential records when the Sequential SDK identifies a visitor. The **Users** list, each user's page, **Sessions**, funnels, flows, retention, **Events** and the **User ID** filter use claims. The **Dashboard** tiles do not: they count user ids. The **Traffic** tile labels them **User IDs**, with the hover text "Distinct user ids. A visitor who signed in during a visit counts once per id they used." [Users and identity](/docs/users-and-identity)
- **Ingest key.** A public key for one source in a project, such as `sq_posthog_…`. It cannot read your analytics. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Insights tab.** The replay player tab that judges the open recording on demand. It does not show a signal that the judge rates unlikely. [Session insights](/docs/session-insights)
- **Judge.** The AI model that answers your rubric's questions. See [AI and third parties](/docs/ai-and-third-parties). [AI and third parties](/docs/ai-and-third-parties)
- **Judged with an earlier rubric.** A note on a judgment made with different question wording, or a newer version of the judge. [Watch: sessions worth your time](/docs/watch)
- **Judgment.** The judge's answers about one session, stored as probabilities. When background judging is on for your project, it makes one judgment per session that was active in the last 7 days. Sequential marks a session with fewer than 3 meaningful events as too short. It never judges that session. [Judgments](/docs/judgments)
- **Judgment budget.** Each project's daily judging allowance. It resets at midnight UTC. When an allowance runs out, the judging it covers stops until the reset. [Judgments](/docs/judgments)
- **Latest.** The default tab on **Sessions**: the sessions in your date range, most recently active first. [Find and replay sessions](/docs/sessions-and-replay)
- **Loosening.** A change to the recording settings that records more. To save one, type the project's slug. [Recording settings](/docs/recording-settings)

### More about I–L terms

- **Identify:** Events sent before it keep the anonymous id. Sequential uses the value that you pass as it is. Any stable identifier works. Always pass the same value for a person, everywhere.
- **Identity claim:** A claim says: this anonymous id became this identity in this session. The pages that use claims then show one user.
- **Ingest key:** It can send data to the project and read its recording settings. To rotate or revoke it, go to Settings → Projects → your project → **Sources & keys**. Pendo uses your own Pendo API key instead.
- **Insights tab:** It reuses the result for up to 7 days. It has its own daily allowance. **Re-run** forces a fresh judgment, at most once every 10 minutes per session. The tab never changes Watch.
- **Judge:** It reads the digest and your questions. It returns probabilities, never prose.
- **Judged with an earlier rubric:** The note means that the judge used different question wording, or a newer version of the judge, from what the project uses now. The answers stay as they were. Sequential does not judge the session again.
- **Judgment:** Background judging judges a session when about 30 minutes have passed since the session's last event.
- **Judgment budget:** Background judging, **Ask** in the app and **Ask** from an agent each have their own allowance. The **Insights** tab and the rubric editor's **Try it** each have their own too.
- **Latest:** A ▶ marks a session that has a recording that you can play now. From the moment the list appears, a dim "…" ("Checking recording…") shows until Sequential has the state of its recording. If Sequential cannot read the state, also on a second try, the row reads "Couldn't check recording", or shows ▶ when Sequential has an index of a recording. The **Problems** column shows a badge when the session had JS errors or failed requests.
- **Loosening:** Examples: a move from **Strict** to **Balanced** masking, or turning on **Record message text**. The audit log flags it.

### What a judgment holds

A judgment stores one probability for each of these:

- each option of a list question
- each level of a scale
- yes, on a signal
- which event a moment points to

### Identity claims

- Claims come only from the Sequential SDK.
- The session where identify ran always follows its claim.
- The other sessions of the anonymous id follow its first claim. If two values claim the same anonymous id, those sessions go to the first value that Sequential received for that anonymous id.
- A claim links a single step, not a chain of identify calls.
- A new claim can take up to 5 minutes to show.

## M–P

- **Masking.** Hiding what people type and read in a web recording, in their browser. Under **Strict**, it also hides attribute values and most of each address. Masked text plays back as solid blocks (█), and the layout stays. [Masking and redaction](/docs/masking-and-redaction)
- **Member.** A person on your team who signs in to Sequential. [Roles and permissions](/docs/roles-and-permissions)
- **Moment.** A "where" question in the rubric: did something happen, and at which event? [Customise the rubric](/docs/rubric)
- **Near miss.** A session that **Ask** judged and rated close to a match, but below the cut-off. Ask does not list it. The app shows the count under the results, for example "3 came close but fell below the match threshold." An agent gets it as `near_misses`. [Ask: find sessions by describing them](/docs/ask)
- **Not judgeable.** A session that is too short for **Ask** to judge. The judge does not read it. The app shows the count under the results, for example "2 sessions had fewer than 3 events, too few to judge." An agent gets it as `not_judgeable`. [Ask: find sessions by describing them](/docs/ask)
- **Not measured.** A finding result: Sequential could not compare the sessions before and after the release. The card says why. It is never a count of 0. [Findings and fixes](/docs/findings)
- **Organisation.** The company or team that owns your projects and members. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Policy version.** The version of the recording settings that a session was recorded under, such as "policy v3". [Recording settings](/docs/recording-settings)
- **Project.** One app or site in your organisation. [Organisations, projects and keys](/docs/organisations-and-projects)

### More about M–P terms

- **Masking:** The person's browser does it before it sends anything. Sequential's servers mask attribute values, addresses, input values and DevTools addresses again when a recording arrives. The preset is **Strict** (the default), **Balanced** or **Custom**. Mobile apps set masking in their own code.
- **Member:** A member has one role in your organisation. Do not confuse a member with a *user* of your product.
- **Moment:** Examples are a struggle or a drop-off. **Watch from m:ss** on Watch and the key moments on the **Insights** tab go to it.
- **Near miss:** Its match is at least 0.2 and less than 0.4. If an answer has no results but has near misses, the judge read sessions and none matched. This is true in the app and for an agent.
- **Not judgeable:** Such a session has fewer than 3 events that the judge can read.
- **Organisation:** It holds your members' roles, invitations, the two-factor requirement and the audit log.
- **Policy version:** The player shows it as a badge. It goes up by one each time someone saves the recording settings.
- **Project:** It has its own ingest keys, data, recording settings and rubric. Every report shows one project at a time: the one picked in the project switcher at the top of the left rail.

## R

- **Rage click and dead click.** Repeated clicks on one spot, and a click that gets no response. Sequential stores them only when the sending SDK detects them. The Sequential SDK sends neither yet. [Events](/docs/events)
- **Recorder.** Sequential's web script that records the screen, masks it in the browser and uploads it. [Session replay (web)](/docs/session-replay)
- **Recording.** The captured screen of one session. [Find and replay sessions](/docs/sessions-and-replay)
- **Recording settings.** A project's rules for what Sequential records and how. [Recording settings](/docs/recording-settings)
- **Redaction.** Sequential replaces values that have a recognisable shape with `[redacted]`. It does not redact a name or an address in a console message. [Masking and redaction](/docs/masking-and-redaction)
- **Release basis.** A way to measure a finding after a release: only the sessions that carry the fix release, and no other release, count. Sequential uses it when your events report the release. [Findings and fixes](/docs/findings)
- **Release marker.** A record that you shipped a version, and which findings it is for. You send it with an access token, for example from CI. [Findings and fixes](/docs/findings)
- **Replay key.** The project's **Session replay** ingest key. Keep at least one replay key active. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Replay period.** How long a recording stays viewable: 30 days by default. The storage deletes the recording files about 2 days after it ends. [Retention and deletion](/docs/retention-and-deletion)
- **Retention (data).** How long Sequential keeps what it collects. Events stay for the organisation's events retention period, 365 days by default. The storage deletes recording files about 2 days after the replay period. Sequential sets both periods for each organisation. A server can set a shorter limit for every project. [Retention and deletion](/docs/retention-and-deletion)
- **Retention (report).** The **Retention** page: the share of people who did a return event exactly N days after their first start event. [Retention](/docs/retention)
- **Role (organisation).** A member's permission level: owner, admin, member or viewer. [Roles and permissions](/docs/roles-and-permissions)
- **Role (rubric).** A tag that connects a rubric question to a place in the product. [Customise the rubric](/docs/rubric)
- **Rubric.** Your project's questions, up to 24, that the judge answers for every judged session. [Customise the rubric](/docs/rubric)
- **Rubric hash.** A fingerprint of everything that the judge reads from your rubric. [Customise the rubric](/docs/rubric)
- **Rubric version.** The count of saves of your rubric, such as "Customised · v3". [Customise the rubric](/docs/rubric)

### More about R terms

- **Rage click and dead click:** A rage click is repeated clicks on one spot. A dead click is a click that gets no response. Sequential stores them as events. The replay player also marks rage clicks and dead clicks that it detects itself. It shows these only on screen.
- **Recorder:** It uploads under your analytics SDK's session id. The Sequential SDK and the Smartlook compatibility script start it for you.
- **Recording:** The replay player plays it back with the session's events on its timeline. On the web, the recorder makes it. On iOS and Android, PostHog's mobile SDK makes it.
- **Recording settings:** They cover on or off, platforms, sampling, which pages, masking and DevTools captures. They are under Settings → Projects → your project → **Session replay**. Owners and admins can change them.
- **Redaction:** It applies in DevTools data and stored URLs. Examples: email addresses, card numbers, JWTs and `Bearer` credentials. It is always on. It matches shapes, not meaning.
- **Replay key:** The web recorder uses it. The Sequential SDK and the Smartlook compatibility script fetch it themselves.
- **Replay period:** Sequential counts it from when the first part of the recording arrived. The organisation has one replay period for all its projects. The summary line on your project's **Session replay** settings tab shows it, for example "30 d". When the period ends, you cannot play the recording. The player shows "This recording has expired."
- **Retention (data):** Events, and the user, session and account records, stay for the organisation's events retention period: 7, 30, 90, 180 or 365 days, counted from when each event arrived, and up to about a week longer. Recordings stop being viewable after the replay period, and the storage deletes their files about 2 days later. Some derived data expires on fixed periods that you cannot change, unless the server sets a shorter [limit](/docs/retention-and-deletion#a-server-wide-limit):
  - judgments: about 400 days after the session ends
  - 👍/👎 feedback: about 400 days after someone gave it
  - Ask searches: after about 180 days
- **Retention (report):** It puts each person in a cohort by the UTC day of their first start event in the date range. Then it shows the share of people who did a return event exactly N days later. By default, both events are any event, and N is 1, 3, 7, 14 and 30.
- **Role (organisation):** A member has a role per organisation, and it applies to every project in it. There are no project-level roles.
- **Role (rubric):** The roles are **Intent**, **Outcome**, **Friction**, **Worth watching**, **Struggle**, **Goal** and **Drop-off**. Only one question can hold each role.
- **Rubric:** Examples: Intent, Outcome, Friction, Hit an error and Worth watching. Owners and admins edit it under Settings → Projects → your project → **Insights**.
- **Rubric hash:** A new hash marks earlier judgments "judged with an earlier rubric". It also restarts the **Calibration** counts.
- **Rubric version:** The editor header shows it. It is separate from the hash. A save that changes only question labels or weights gets a new version. The hash stays the same.

### Rage clicks and dead clicks per SDK

| SDK | Rage clicks | Dead clicks |
|---|---|---|
| PostHog | Sends them by default | Sends them only if you set `capture_dead_clicks: true` when you initialise it |
| Mixpanel | Its autocapture can send them | Its autocapture can send them |
| Sequential SDK | Does not send them yet | Does not send them yet |

### What changes the rubric hash

The hash covers these parts of the rubric, in order:

- each enabled question's text and criteria
- each option's key with its criteria, or its label when the option has no criteria
- each level's label and criteria

| Change | Changes the hash? |
|---|---|
| Rename a level | Yes |
| Rename an option that has no criteria | Yes |
| Reorder options or levels | Yes |
| Change question labels | No |
| Change roles | No |
| Change weights | No |
| Change the summary wording | No |
| Change the order of questions | No |

## S

- **Sampling.** The share of sessions that a project records: by default 100% for production and 10% for staging and development. [Recording settings](/docs/recording-settings)
  - A sampled recording can still end early. **Max minutes** and **Max MB** under **Session caps** on the project's **Session replay** tab cut a long recording short. The defaults are 60 and 25.
- **Saved view.** A funnel, flow or filter that you saved by name. It belongs to the project, and everyone in the project sees it. A saved filter can also be a funnel or flow step. The step is a copy: a change to the step or to the saved filter does not change the other. [Filters, date ranges and saved views](/docs/filters-and-saved-views)
- **Scale.** A rubric question with ordered levels, lowest first, such as Friction. [Customise the rubric](/docs/rubric)
- **Scroll map.** The **Scroll** view of the **Heatmaps** page. For each 5 % of a page, it shows the share of the sampled views that scrolled which saw all of that part. Bands on the page picture and the **Scroll depth** card show it. [Heatmaps](/docs/sessions-and-replay#the-scroll-map)
- **Session.** One visit, as your analytics SDK defines it. It is not a sign-in session. [Sessions](/docs/sessions)
- **Session id.** The value on each event that groups it into a session and joins it to the recording. An event without one still counts as an event. It is never a session: **Sessions** does not list it, and Sequential never judges it or links it to a replay. [Sessions](/docs/sessions)
- **Signal.** A yes-or-no rubric question, such as Hit an error. [Customise the rubric](/docs/rubric)
- **Source.** The channel that data arrives through, such as PostHog. Each source has its own ingest key. [Organisations, projects and keys](/docs/organisations-and-projects)
- **Super property.** A property that you register with `sequential.register()`. Call `register()` on every page load. [Send custom events](/docs/custom-events)
- **System event.** SDK housekeeping, such as `$pageleave`. Event lists hide system events until you choose to show them. [Events](/docs/events)

### More about S terms

- **Sampling:** Sequential decides once per session, so sampling never records part of a session. Sequential never samples analytics events. Your value is under Settings → Projects → your project → **Session replay** → **Sampling**.
- **Saved view:** The pages and **Dashboard** list the saved views of the selected project. Owners, admins and members can rename and delete any of them. Viewers can only apply them. Each one shows who changed it last. Items that a browser kept from earlier versions can be moved into the project. **Not now** asks again in 7 days. **Past filters** is not a saved view: it is your own list of the filters that you applied, and nobody else can read it. In **Funnels** and **Flows**, **+ Step from saved view** copies a saved filter into a new step. The step shows **From saved view:** and the name of the filter. **Update from view** copies the current conditions again.
- **Scale:** The judge gives a probability for each level.
- **Session:** A session is all the events, and at most one recording, that share a *session id* in a project. Settings → **Security** lists sign-in sessions.
- **Session id:** Your SDK supplies it, and Sequential stores it exactly as sent.
- **Signal:** Another example is Worth watching. The judge gives the probability of yes.
- **Source:** The sources are Sequential, PostHog, Mixpanel, Amplitude, Pendo, Smartlook and Session replay. Every event records the SDK that sent it.
- **Super property:** You register it once. The Sequential SDK then adds it, as an event property, to every later event from the same page. It lasts only as long as the page.
- **System event:** Other examples: `$web_vitals` and `$set`. A system event is not something the person did. Sequential stores system events and counts them in charts. The digest leaves them out.

## T–W

- **Time basis.** A way to measure a finding after a release when your sessions do not report the release: the sessions that started 2 hours or more after the release count. The card labels it "Time-based". [Findings and fixes](/docs/findings)
- **Too close to call.** A collapsed lane below Watch's ranked list, for sessions where the judge could not decide the goal or the outcome. [Watch: sessions worth your time](/docs/watch)
- **Try it.** A rubric editor panel that judges your unsaved draft on one recent session. Every run is a real judgment. [Customise the rubric](/docs/rubric)
- **Two-factor authentication.** A code from an authenticator app, asked for at sign-in with a password or with Google. [Two-factor authentication and account security](/docs/two-factor-and-security)
- **Unclear.** An answer that means the judge could not tell from the session's events. It is not the same as no. Sequential is reworking how the app shows unclear answers. [Reading a verdict](/docs/reading-verdicts)
- **User.** A person who uses your product, known by one *user id*. [Users and identity](/docs/users-and-identity)
- **User property.** A property of the person, such as their plan. Only some SDKs store them. [Users and identity](/docs/users-and-identity)
- **Verdict.** The judge's best answer to a rubric question, with its likelihood. It is a forecast, not a finding. [Reading a verdict](/docs/reading-verdicts)
- **Watch.** The **Watch** tab on **Sessions**: judged sessions in three groups, each ranked by Watch score. [Watch: sessions worth your time](/docs/watch)
- **Watch score.** The number that Watch ranks by in each group, from each question's **Watch score %**. [Judgments](/docs/judgments)
- **Worth watching.** The default rubric's question: should a product manager watch this recording? [Watch: sessions worth your time](/docs/watch)

### More about T–W terms

- **Too close to call:** The lane holds judged sessions where the judge could not decide the person's goal or the outcome. Nothing else stood out in these sessions. This is a result, not a failure.
- **Try it:** To use it, pick the session under **Session** and press **Run**. You see the answers before you save. It uses the daily allowance for **Try it**.
- **Two-factor authentication:** Sequential asks for the code when you sign in, with a password or with Google. An organisation can require it of every member.
- **Unclear:** An unclear answer is a result, not a failure. Sequential keeps "I don't know" apart from "I don't think so" on purpose.
- **User:** Before your app identifies them, the user id is the anonymous id. After that, it is the value that your app passes. Each user id is one row on **Users**. With the Sequential SDK, PostHog or Mixpanel, an anonymous id that is linked to one identified value is counted in that value's row.
- **User property:** It describes the person, not one event. User properties show as **Visitor Property** in filters and under **User properties** in the replay's **Info** tab.
- **Verdict:** Example: "Buy or subscribe" at 81%. In the rubric editor, **Verdicts** are also the questions that pick one answer from a list, such as Intent.
- **Watch:** Each row has its reason and a way in. On a wide screen, the way in is in a preview beside the list. If you can play the recording now and the judge placed a moment confidently, **Watch from m:ss** opens the replay at that moment. If you can play it but there is no such moment, the row shows **Open replay**. Otherwise it shows **Open events**.
- **Watch score:** The weights are in each question's **Watch score %**. Each row has a badge: **Watch now**, **Watch** or **Maybe**. A session whose recording you cannot play now shows **Review now**, **Review** or **Maybe**. So does a row while Sequential checks its recording. If you hover over the badge, it shows the score. For example: "Attention 0.80".
- **Worth watching:** It is a yes-or-no question. It has the largest default Watch score weight. The 👍/👎 on Watch rows record whether it was right.

### User properties per SDK

| SDK | Stores user properties? |
|---|---|
| Pendo | Yes |
| Amplitude, with `Identify` | Yes |
| Smartlook compatibility script, with `identify` | Yes |
| Sequential SDK | No |
| PostHog | No |
| Mixpanel, with the browser `people.set` | No |

### How the Watch score is calculated

- Each weighted yes-or-no question adds its weight times the probability of yes.
- Each weighted scale adds its weight times how far up the scale the session sits.

The default weights:

| Question | Watch score % |
|---|---|
| Worth watching | 45% |
| Friction | 25% |
| Hit an error | 15% |
| Product bug | 15% |

## Related

- [Troubleshooting](/docs/troubleshooting)
- [Limits](/docs/limits)
- [Support matrix](/docs/support-matrix)
- [What is Sequential](/docs/what-is-sequential)