Pyze Documentation – Capabilities

Docs: Autocapture & Event Tracking

Mon, 01 Jan 0001 00:00:00 +0000

Pyze automatically captures user interactions — page and screen views, clicks, field interactions, and custom events — across the enterprise applications your teams use, with no manual instrumentation required. Engineering can also send explicit events via the SDK/API when richer metadata is needed.

Capabilities

Tagless autocapture of front-end interactions — events are available retroactively, with no per-event tagging
Explicit event-level tracking via SDK/API for rich, typed metadata
Server-side and browser-side collection — no screen recording, no desktop agent
Works inside web apps, embedded enterprise apps (Pega, Salesforce, SAP UIs), and portals

How it works

The Pyze JavaScript agent is added once (server- or browser-side).
Interactions are captured automatically; define custom events where you need typed metadata.
Events stream to Pyze (and optionally your warehouse) for analysis.

Availability

Available across all JavaScript-based agents (Web, Pega, and embedded enterprise apps). See SDK Integrations.

Docs: Cross-Platform Tracking

Mon, 01 Jan 0001 00:00:00 +0000

Pyze unifies behavioral data across web, mobile, and desktop — including the in-branch/teller and back-office desktop apps that consumer analytics tools cannot reach — into a single, coherent view of the work.

Capabilities

Web, mobile, and desktop coverage in one model
Reaches internal enterprise apps and virtualized (Citrix/VDI) environments
Consistent event schema across surfaces for apples-to-apples analysis

How it works

Deploy the appropriate Pyze agent per surface.
Events share a common schema and are stitched to a single user/customer identity.

Availability

Available across the Pyze SDK family (Web, mobile, desktop).

Docs: Identity Stitching

Mon, 01 Jan 0001 00:00:00 +0000

Pyze links anonymous sessions to the known user and, where relevant, the customer and the servicing staff member — producing an accurate end-to-end journey across channels and devices.

Capabilities

Anonymous → known identity resolution
Stitch sessions across devices and channels
Tie a journey to both the customer and the servicing employee

How it works

Call the identify API when a user becomes known (login or case association).
Pyze merges prior anonymous activity into the persistent identity.

Availability

Available via the SDK/API identify call.

Docs: Data Privacy, PII Masking & Consent

Mon, 01 Jan 0001 00:00:00 +0000

Sensitive fields are masked at the point of capture — analytics see the interaction, never the raw PII. Pyze supports field-level masking, hashing, and consent management for regulated, enterprise environments.

Capabilities

PII masked/hashed at source — raw values never leave the browser
Field- and zone-level exclusion for sensitive content
Consent capture and enforcement
GDPR/enterprise-compliant; no screen recording

How it works

Configure which fields/zones are masked or hashed (see Data Hashing).
Consent state gates collection where required.

Availability

Docs: Warehouse Integration & Reverse ETL

Mon, 01 Jan 0001 00:00:00 +0000

Pyze integrates with your cloud data warehouse — landing governed behavioral events and syncing processed metrics and SoP/conformance signals back into operational systems (reverse ETL).

Capabilities

Warehouse-native event landing (no data duplication where supported)
Reverse ETL of processed metrics and signals to operational systems
Full SQL transparency over the underlying model

How it works

Connect your warehouse; map the Pyze event model.
Schedule or stream events in; sync metrics back out.

Availability

Available with warehouse connectors; see Integrations.

Docs: Funnel & Conversion Analysis

Mon, 01 Jan 0001 00:00:00 +0000

Define a sequence of steps and Pyze tracks progression, conversion, and drop-off — broken down by any segment, trended over time, and drillable into the cases that dropped and why.

Capabilities

Step conversion and drop-off, overall and step-to-step
Breakdown / compare by segment (region, team, channel, case type)
Conversion-over-time trend and time-to-convert distribution
Click a drop to drill into the cohort and the root-cause contributors

How it works

Pick the ordered steps (events or stages).
Apply global filters; break down by a dimension.
Click any drop to drill into why.

Availability

Available in the analytics workspace.

Docs: Cohort Retention

Mon, 01 Jan 0001 00:00:00 +0000

Build behavioral cohorts (e.g. users who adopted an AI assist or a new flow) and measure whether the behavior persists week over week — alongside the outcome lift (handle time, conformance) versus non-adopters.

Capabilities

Behavioral cohort builder (users who did X in a window)
Retention triangle/heatmap + curves (n-day, unbounded, bracket)
Compare cohorts and segment retention
Outcome lift: adopters vs non-adopters

How it works

Define the cohort’s start and return behavior.
Pyze computes the retention triangle and outcome comparison.

Availability

Available in the analytics workspace.

Docs: Feature Adoption & Engagement

Mon, 01 Jan 0001 00:00:00 +0000

Measure how widely and how deeply each feature or workflow step is used, so you know what drives value and what to retire, automate, or augment.

Capabilities

Adoption breadth and depth per feature/step
Time-to-adopt and adopted-vs-dormant views
Adoption funnels that prove a rollout changed behavior

How it works

Pyze derives adoption from captured events automatically.
Filter and break down by segment.

Availability

Available in the analytics workspace.

Docs: Customizable Dashboards

Mon, 01 Jan 0001 00:00:00 +0000

Build dashboards without engineering — KPI cards with trends and period-over-period deltas, mixed charts, dashboard-level filters, and drill-from-tile. Pin charts (including Ask-the-Data results) to share across the organization.

Capabilities

KPI cards with sparkline + delta; mixed chart grid
Dashboard-level global filters and time comparison
Drill from any tile; pin charts to a dashboard
Role-tailored layouts

How it works

Add charts via the chart builder (measure → dimension → type) or pin from Ask the Data.
Apply dashboard filters that cascade to every tile.

Availability

Available in the analytics workspace.

Docs: Anomaly Detection & Alerts

Mon, 01 Jan 0001 00:00:00 +0000

Pyze surfaces metrics, branches, or segments drifting from the network baseline — with no hand-set thresholds — and ranks the dimensions that drove the change. Alerts route to the right owners.

Capabilities

Statistical detection (no manual thresholds)
Contribution / root-cause: what segment drove the change
Severity scoring and an alert feed
Configurable alert routing

How it works

Pyze learns the baseline and flags deviations beyond it.
Open an alert to see contributing segments and drill in.

Availability

Available in the analytics workspace.

Docs: Session Replay

Mon, 01 Jan 0001 00:00:00 +0000

Session Replay reconstructs a case as a step-by-step timeline — every screen, with handling time, page load, app switches (swivel), and clicks — so teams can see the friction behind a metric and jump straight to the moments that matter.

Capabilities

Step/screen timeline with handle, load, swivel, and click signals
Jump-to-friction: seek to rage/dead clicks and re-entry
Session search and filtering on any dimension
Link a session from any chart, funnel step, or anomaly
Privacy-first: PII masked at capture, no raw screen content

How it works

Sessions are reconstructed from autocaptured events — no separate recording.
Open a session from a chart/funnel, or search the session list; use jump-to-friction to seek.

Once it's enabled, there's nothing to set up

Session Replay runs automatically through the Pyze agent already on your site — once Pyze turns it on for your app, sessions are captured with no extra code or configuration. The capture and sanitization options below only come into play if you have specific needs, such as masking particular areas or changing what’s recorded.

What gets captured

Session Replay reconstructs each session from page changes and captured signals. You choose what’s recorded — and what’s kept private — per app.

Network requests — captures fetch and XHR calls with their URL, timing, and status. Request and response bodies are off by default; turn on body capture only when you need it. Sensitive headers (cookie, set-cookie, authorization) are never recorded, and you can add more headers to ignore. Optionally record only failed (4xx/5xx) requests.
Console logs — records console output (log, info, warn, error, debug, assert). Choose which levels to keep, or turn console capture off.
JavaScript errors — captures unhandled errors with their stack traces, so you can see what broke during a session.
Page performance — captures page-load and performance metrics (on by default; can be turned off).
Respect Do Not Track — when enabled, Session Replay honors the browser’s Do Not Track signal and skips recording for those users.

Sanitization

Session Replay never records your screen as video. Instead, it captures the page’s structure and the changes to it, then rebuilds the session as a faithful re-render of what the user saw. Because the replay is reconstructed rather than filmed, sensitive values can be stripped out before anything ever leaves the page.

When sanitization is turned on, masking is the default — anything typed into a form appears as asterisks instead of the real text, email addresses shown in page text are hidden, and passwords are always fully hidden. You can see how someone moved through and interacted with your app, but never what they typed, and because masking happens at the point of capture, masked content can’t be recovered later.

Choosing a protection level

You decide how form-field values are recorded, per app:

Level	What the replay shows
Show values	The text exactly as typed
Mask values (default)	Dots matching the number of characters typed — you see how much was entered, not what
Hide values	Nothing at all, not even the length

At the default level every field is masked, so nothing a user types is visible. If you choose to show values in clear text instead, Session Replay can still mask the most sensitive types on its own — email addresses, long numbers, and dates — so those stay protected even when other fields are visible. Protection can only be added this way, never removed: passwords stay hidden no matter what. (Some fields, such as dates and dropdowns, appear empty when masked rather than showing dots.)

Masking specific areas

When you need more control than the per-app level, you can protect individual parts of a page:

Mask an element — its text becomes asterisks, and any fields inside it record only how much was typed. This applies to the element and everything within it.
Hide an element — the element and its contents are left out of the recording entirely.

The simplest way to do this across a whole app is to give Pyze a list of CSS classes: any element using one of those classes is masked automatically, so you can protect sensitive areas everywhere they appear without marking each one by hand.

Availability

Enabled with the Pyze agent; PII masking applies (see Data Privacy & Consent).

Docs: Heatmaps & Element-Level Impact

Mon, 01 Jan 0001 00:00:00 +0000

See where attention and effort concentrate at the field and screen level. Element-level analysis ties each field to downstream effort and rework — a high re-edit rate is a precise automation candidate.

Capabilities

Click / interaction intensity by field and screen
Re-edit (friction) rate per field — element-level impact
Zone-based views that handle responsive layouts

How it works

Pyze aggregates field/screen interactions from captured events.
Sort by intensity or re-edit rate to find friction.

Availability

Available in the experience workspace.

Docs: Frustration & Struggle Detection

Mon, 01 Jan 0001 00:00:00 +0000

Pyze detects behavioral signals of friction — rage clicks, dead clicks, excessive app switching (swivel-chair), hesitation, and field re-entry — and scores sessions so the worst experiences surface automatically.

Capabilities

Rage/dead-click, swivel, hesitation, and re-entry signals
Per-session struggle scoring
Auto-surfacing of the highest-friction sessions

How it works

Signals are derived from captured interaction data.
Filter or sort sessions by friction to triage.

Availability

Available in the experience workspace.

Docs: Performance Monitoring

Mon, 01 Jan 0001 00:00:00 +0000

Pyze captures performance — page load, server, and wait time — per screen and correlates it with behavior, so you can tell slow software apart from hard work and prioritize the screens that cost the most time.

Capabilities

Load / server / wait time captured per screen
Correlation of performance with handling time and drop-off
Slowest-screen ranking

How it works

Performance metrics are captured alongside interaction events.
Review by screen; correlate with funnels and friction.

Availability

Available in the experience workspace.

Docs: A/B Testing & Feature Flags

Mon, 01 Jan 0001 00:00:00 +0000

Toggle features for segments, ramp rollouts, and read out experiments with significance, confidence, guardrail metrics, and a clear ship/hold verdict — so a feature only scales when it actually helps.

Capabilities

Feature flags with segment targeting and gradual rollout
A/B readouts: lift, significance, confidence intervals
Guardrail metrics and a ship/hold decision

How it works

Define the flag and target segment; ramp the rollout.
Pyze reads out the experiment against your metrics.

Availability

Available in the adoption workspace.

Docs: In-App Guidance & Onboarding

Mon, 01 Jan 0001 00:00:00 +0000

Guide users in the flow of work — contextual tooltips and hotspots, onboarding checklists, step-by-step walkthroughs, in-app announcements, and an embedded help center — with adoption analytics on every guide.

Capabilities

Tooltips, hotspots, and beacons
Onboarding checklists and guided walkthroughs
In-app announcements and an embedded help center
Adoption analytics on each guide

How it works

Author guidance against the app’s screens/elements.
Target by segment; measure views → completion → downstream adoption.

Availability

Available in the adoption workspace.

Docs: Survey Setup — In-Product Surveys, NPS & Feedback

Mon, 01 Jan 0001 00:00:00 +0000

Collect feedback in the flow of work — event-triggered microsurveys, NPS/eNPS, and always-on feedback widgets — with frequency capping and segment targeting. Build a survey once, choose when and to whom it shows, and every response ties back to the respondent’s behavior.

Capabilities

Question & survey types

NPS and eNPS (0–10) with automatic promoter / passive / detractor scoring
Rating questions — star, smiley, or number scales (3 to 10 points)
Single- and multiple-choice, ranking, and picture selection
Open text with email, URL, number, and phone input variants
Matrix, consent, date, address, and contact-info questions
A call-to-action card for in-product announcements and prompts
Welcome screens and custom thank-you screens — end on a message with a call-to-action button, or redirect to a URL
Multi-step flows with branching logic, computed variables, recall (pipe an earlier answer into a later question), hidden fields, and per-survey languages

Triggering & display

Code-triggered surveys fired from a custom event in your app
No-code triggers — page view, element click, exit intent, and 50% scroll — with URL matching rules
Frequency capping: show once, show again after a recontact wait, or limit total displays
Sampling by percentage, an initial delay, auto-close, and a start/close scheduling window
Close the survey automatically once it reaches a target number of total responses

Targeting

Segment on attributes, person identifiers, and device
Route a survey to specific apps by app key (app targeting)
User targeting filters for advanced, attribute-based audiences

Formats

In-product surveys shown inline or as a modal, positioned in any corner or centered, with an optional dark overlay
Shareable link surveys for surveys that run outside the app
Branding to match your app — colors, logo, progress bar, background, dark mode, and card layout
Email follow-ups triggered on response or on a specific ending

Getting Started

Step 1: Enable Surveys for your app

Surveys are turned on per app by Pyze. Contact Pyze to enable Surveys for your app, and the Surveys item appears in the left-hand menu.

Don't see Surveys in the menu?

The Surveys menu item only appears once Pyze has enabled the capability for your app. If it’s missing, reach out to Pyze to have it turned on.

Step 2: Create and publish a survey

Open Surveys from the menu.
Start from a template or build from scratch — add your questions (NPS/eNPS, rating, multiple choice, open text, and more), a welcome and thank-you screen, and set branding, branching, and languages as needed. A live preview shows the survey as you build it.

×
Choose when the survey appears with a no-code trigger — no changes to your app are needed; the Pyze agent already on your site displays it:
- Page view — when someone lands on a page
- Element click — when a specific button or element is clicked
- Exit intent — as the pointer moves to leave the page
- Scroll — once someone scrolls part-way down the page
Narrow any trigger to specific pages with URL rules (exact match, contains, starts with, ends with). You can also add a short delay before the survey appears and auto-close it after a set time.

×
Set who sees it and how often:
- Audience — target by segment or by user attributes, or route the survey to specific apps.
- Recontact — show it once, allow repeat displays, or wait a set number of days before showing it to the same person again.
- Display cap — limit the total number of times it’s shown, or show it to only a percentage of eligible users (sampling).
- Schedule — set start and end dates so the survey runs only within a window.
Publish the survey to make it live.

In your app

Nothing beyond the Pyze agent you already use for analytics is required to display surveys. If you target on user attributes, set those attributes from your app with the Pyze SDK so the agent can match them.

Responses flow into analytics tied to the user, so you can compare what people say with what they do. Tag and filter responses to organize them.

Availability

Available in the voice-of-customer workspace. PII is masked at the point of capture (see Data Privacy & Consent).

Docs: AI Feedback & Sentiment Analysis

Mon, 01 Jan 0001 00:00:00 +0000

Pyze uses NLP to cluster verbatim feedback into themes, score sentiment per theme, and surface emerging issues — with representative quotes and the ability to tie a theme back to the behavior behind it.

Capabilities

Automatic theme/topic clustering of verbatims
Sentiment per theme and trend over time
Emerging-theme detection and representative quotes

How it works

Open-text responses are analyzed automatically.
Review themes, sentiment, and drill to the underlying responses/behavior.

Availability

Available in the voice-of-customer workspace.

Docs: Role-Based Access (RBAC)

Mon, 01 Jan 0001 00:00:00 +0000

Pyze provides role-based access control with granular permissions over data and platform features, plus SSO and provisioning — so each role sees the right scope and enterprise governance standards are met.

Capabilities

Granular role/permission model
Row- and column-level scoping
SSO (SAML) and provisioning (SCIM)

How it works

Map roles to scopes and features in admin settings.
Integrate SSO/SCIM for provisioning.

Availability

Configured in admin settings; SSO via your IdP.

Docs: Integrations (CDP, CRM, Warehouse, BI)

Mon, 01 Jan 0001 00:00:00 +0000

Pyze connects to CDPs, CRMs, cloud data warehouses, BI tools, and APIs — combining behavioral data with customer and business data for enriched analysis, and exposing the model over open APIs.

Capabilities

CDP / CRM / warehouse / BI connectors
Open Enterprise APIs
Warehouse-native SQL transparency (no data copy)

How it works

Configure the connector and field mapping.
Query via the APIs or directly in your warehouse.

Availability

See Enterprise APIs and warehouse connectors.

Docs: MCP Server for Agentic Interoperability

Mon, 01 Jan 0001 00:00:00 +0000

Pyze exposes a Model Context Protocol (MCP) server so AI agents can query execution data, baselines, and machine-checkable SoPs in real time — grounding decisions and writing outcomes back for continuous measurement.

Capabilities

Standardized MCP endpoint for agent access
Tools for metrics, conformance, opportunities, and SoP retrieval
Real-time, governed access for agentic orchestration

How it works

Point your agent framework at the Pyze MCP endpoint.
Agents call the exposed tools with governed scope.

Availability

Available as a managed MCP server endpoint.

Docs: AI Agent Effectiveness Analytics

Mon, 01 Jan 0001 00:00:00 +0000

Pyze measures the performance and impact of deployed AI agents — adoption, task success, accuracy, exception/escalation rates, and cost/cycle-time impact — against the pre-deployment human baseline and the governing SoP.

Capabilities

Adoption, success, accuracy, and exception rates per agent
Human vs agent vs hybrid comparison against the baseline
Agentic ROI ledger and a ship/hold/expand verdict

How it works

Instrument the agent’s actions as events.
Pyze compares against the baseline and the SoP.

Availability

Available in the AI analytics workspace.

Docs: Analyst AI Agents

Mon, 01 Jan 0001 00:00:00 +0000

Ask questions in natural language and get grounded answers with the right chart — generated against a governed semantic layer, with the query shown for trust, and pinnable to a dashboard.

Capabilities

Natural-language → grounded SQL → answer + chart
Semantic-layer grounding with the generated query shown
Pin results to a dashboard; suggested follow-ups

How it works

Type a question in the Ask-the-Data panel.
Pyze grounds it to the semantic layer, runs validated SQL, and returns an answer + chart.

Availability

Available as the Ask-the-Data copilot in the analytics workspace.

Docs: GenAI User-Behavior Insights

Mon, 01 Jan 0001 00:00:00 +0000

Pyze uses LLMs to synthesize behavioral data into plain-language narratives — surfacing the biggest leak, the costliest friction, and the segments worth attention — so cross-functional teams get the story, not just the numbers.

Capabilities

Auto-generated narrative summaries from live data
Highlights leaks, friction, and outlier segments
Grounded in the same governed metrics as the dashboards

How it works

Insights are generated from the current (filtered) metrics.
Review on the AI analytics page or via the copilot.

Availability

Available in the AI analytics workspace.