Overview
Insights is the platform's built-in reporting workspace. It is accessed from within the platform — no separate installation, account, or login is required. Saved questions, dashboards, and other artefacts persist within Insights and are accessible to colleagues with the appropriate collection permissions.
Three activities account for the majority of usage: viewing dashboards prepared by colleagues, building custom questions from models, and assembling questions into dashboards.
Core concepts
The following five terms recur throughout Insights and throughout this guide. Familiarity with these concepts is sufficient to navigate the remainder of the documentation.
Question
A question consists of a single query and a single visualisation. Questions are constructed by selecting a data source, applying filters, defining aggregations, grouping by dimensions, and selecting a chart type. Saved questions may be reused across dashboards.
Three question types are supported:
Simple question — built by selecting a model, applying filters and a summarisation, and selecting a chart. Suitable for ad-hoc exploration.
Custom question (also referred to as a Notebook question) — built in the Notebook editor with multi-step filters, joins, custom columns, and custom expressions. This is the primary question type used in this guide.
Native question — written in SQL. Outside the scope of this guide; the published models cover the reporting use cases this guide addresses.
Model
A model is a curated, ready-to-use dataset prepared by the platform team. Each model presents the underlying data with friendly column names, descriptions, appropriate data types, and pre-joined dimensions.
Models are listed in the left sidebar under Models and surface as the default data source in the data picker when a new question is initiated. Each model is purpose-designed around a single topic — shifts, financial lines, workers, audit events, scheduling stages, agency performance, and so on.
New questions should always be initiated from a model, never from a raw database table. Raw tables expose internal column names and require manual joins across several tables. Models encapsulate that work and present a single, clean source.
Dashboard
A dashboard is a single screen composed of one or more cards — typically questions, with optional text, link, and iframe cards — arranged in a coherent layout. A dashboard answers one high-level question for one audience: site-level operational performance for a site manager, monthly spend for finance, SLA compliance for procurement.
Dashboards support tabs (multiple subject areas on one dashboard), filters (controls at the top of the dashboard that re-slice every wired card), click behavior (drill-through, cross-filtering, navigation to other dashboards), and subscriptions (scheduled delivery to email or Slack).
Collection
A collection is a folder. Saved questions, dashboards, models, sub-collections, and Metrics all reside in collections. Each user has a Personal collection, visible only to that user; in addition, shared collections organised by department, audience, or topic are provided by the administrator.
Items are moved between collections via the Move action in the "..." menu. An item resides in exactly one collection at a time, but a single dashboard may be pinned in multiple locations and a single question may be referenced from multiple dashboards.
Metric
A Metric (capital M) is a saved, named aggregation — for example, "Active workers", "Fill Rate %", or "Total Charge". Once a Metric has been defined by an administrator or curator, any user building a question may select that Metric from the data source list. The definition is centralised, ensuring consistent application of the calculation across every question that references it.
Metrics are exposed as a top-level entry in the left sidebar alongside Models. The pattern is recommended whenever the same expression would otherwise be re-implemented across multiple questions.
Workspace tour
The initial landing surface of Insights displays:
The home page, listing pinned items, recently viewed items, and a search box.
The left sidebar, with the following entries:
Home — returns the user to the home page.
Bookmarks — shortcuts to frequently accessed items, populated by clicking the bookmark icon on any item.
Browse — three sub-entries: Databases, Models, Metrics. Models is the recommended starting point for most new questions.
Search (magnifying glass, top bar) — searches titles, descriptions, and content across every dashboard, question, and model accessible to the user.
+ New (top-right blue button) — the entry point for creating Questions, SQL queries, Dashboards, Collections, and Models.
Personal collection — the user's private workspace, listed near the foot of the sidebar.
Brief familiarisation with these surfaces reduces subsequent navigation time.
Summary
A typical workflow consists of: building a question from a model (or referencing a saved Metric), saving it to a collection, and placing it on a dashboard. Subsequent chapters address each of these building blocks in turn.