Platform Overview

The workspace has a top bar that defines the current selection and a sidebar that switches between analysis sections. You can try it with the illustrative datasets at /demo.

Top bar

Dataset selector

Chooses the dataset that all sections read from.

Date range

Presets Last 90 days, Last 6 months, Last 12 months, Year to date and All data, plus custom from/to dates. Presets are resolved relative to the latest observation in the dataset. Details in Data Explorer.

Search

Keeps only observations whose ID, date, category or value contains the search text.

Export

Downloads CSV files of the current selection. See Reports.

Help

Gives access to this documentation.

Sidebar sections

The category filter is part of the selection as well. The dataset, date range, category filter and search together determine which observations every section uses. Statistics are computed in the browser from the loaded data, so changing the date range, category filter or search recalculates all panels from the dataset already loaded; only switching datasets loads new data.

A dataset is a list of observations plus metadata. Each observation is one value recorded on one calendar day for one category.

Dataset metadata contains the name, description, unit, value label, category label, list of categories, value kind (continuous or discrete), recording precision (number of decimals), source kind (illustrative or api), coverage start and end, observation count and last update. The value kind controls how histograms are binned (see Frequency Analysis); the precision controls how values and class boundaries are displayed.

When no customer data source is configured, the workspace serves three synthetic datasets. Each is generated for every day from 2023-01-01 to 2025-12-31 (1,096 days) and every category, using a deterministic pseudo-random generator (mulberry32) with a fixed seed, so the data is identical on every load. Values are clipped at 0 and rounded to the dataset's precision. In the formulas, t is the number of years since 2023-01-01 (days ÷ 365.25).

Regional Electricity Demand (energy-demand)

Daily demand in MWh for the regions North and South: an annual cycle peaking in winter, a weekend reduction, a trend of +1.2% per year and Gaussian noise with σ = 14 MWh. Continuous, 1 decimal, 2,192 observations.

Support Response Times (support-response)

Daily median first-response time in minutes for the channels Email, Chat and Phone. The values are log-normal with σ of log = 0.42, a channel-specific scale and an improvement of 6% per year. Continuous, 1 decimal, 3,288 observations.

Items per Order (order-items)

Number of items in a sampled order for the channels Web, Store and Mobile. Discrete integers, 3,288 observations.

Customer data is read by a server-side adapter from an HTTPS API operated by the customer. The adapter calls two endpoints and validates every response before it reaches the workspace.

GET {ANALYTICS_API_URL}/datasets
Accept: application/json
Authorization: Bearer {ANALYTICS_API_TOKEN}

200 OK
[
  {
    "id": "daily-demand",              // lowercase letters, digits and hyphens
    "name": "Daily demand",
    "description": "Daily demand per region.",
    "unit": "MWh",
    "valueLabel": "Demand",
    "categoryLabel": "Region",
    "categories": ["North", "South"],   // at least one
    "valueKind": "continuous",          // "continuous" | "discrete"
    "precision": 1,                     // integer 0-6
    "source": {
      "kind": "api",
      "label": "Operations warehouse",
      "notice": "Optional disclosure text"
    },
    "coverage": { "start": "2023-01-01", "end": "2025-12-31" },
    "observationCount": 2192,           // non-negative integer
    "updatedAt": "2026-01-02T06:00:00Z"
  }
]

GET {ANALYTICS_API_URL}/datasets/{id}

200 OK
{
  "meta": { ...same object as in the list above... },
  "observations": [
    {
      "id": "obs-000001",
      "date": "2023-01-01",             // YYYY-MM-DD, UTC calendar day
      "value": 421.7,                   // finite number
      "category": "North",
      "source": "Operations warehouse"
    }
  ]
}

The comments and the example values are for illustration; JSON responses must not contain comments. notice is optional; all other fields are required.

Validation rules

• id (dataset) matches ^[a-z0-9-]+$. The workspace accepts dataset IDs of up to 64 characters.
• name, valueLabel, categoryLabel and source.label are non-empty strings; description and unit are strings and may be empty.
• valueKind is continuous or discrete; source.kind is illustrative or api (use api for customer data).
• precision is an integer from 0 to 6; observationCount is a non-negative integer.
• coverage.start, coverage.end and every observation date use the format YYYY-MM-DD.
• Every observation has a non-empty id, category and source and a finite numeric value.
• Fields not listed in the contract are ignored.

The validator does not check that observation categories appear in categories, that observations are sorted, or that observationCount and coverage match the observations. Keep them consistent, because the category filter is built from categories and the workspace displays the metadata as provided.

Configuration

These variables are set on the server by the Insider Magazin team during onboarding. The token is never sent to the browser. Responses from the customer API are cached on the server for up to five minutes, so changes at the source can take that long to appear.

Error handling