Simple credit rules

Each new generated topic analysis (including a new seasonality demo), demo comparison or demo market analysis costs one credit. Comparing imported CSV topics is free. Reopening an existing snapshot, importing data, filtering, smoothing charts and calculating seasonality from an existing series cost nothing. Rules are evaluated manually and use stored snapshots. Demo balances renew in UTC and are stored locally; production balances will need server-side enforcement. Credits don’t roll over.

Data provenance

Radar uses the Google Trending Now RSS export. Official Google Trends alpha and direct Bing integration are not connected. Explorer also includes an inactive optional DataForSEO adapter, a separate third-party service. If explicitly enabled, its Google data uses 0–100 relative interest indices; its Bing data uses monthly search counts with an explicit country and language. Official Google alpha data uses a different consistent scale and is not provided by this adapter. Explorer demo queries and market analyses use synthetic daily observations. Imported CSV series preserve your supplied values and units. Explorer supports generated demos and local CSV data. API requests run only when you submit a live analysis. A missing connection, failed request or unavailable data never generates a replacement demo. No search-volume estimates are derived from interest indices.

Recent growth & acceleration

Recent change compares the means of two adjacent windows, each containing up to 30 observations. A third window supplies the previous growth rate. Acceleration is their difference in percentage points. A zero baseline returns “Not available”, never infinite growth. A point means one supplied observation, not necessarily one day.

Volatility & anomaly flags

Volatility is standard deviation divided by the full-series mean. Exploratory spike and dip flags use the median of up to 30 previous observations and a threshold of four scaled median absolute deviations, with a 25% baseline floor and one-unit minimum. These flags are descriptive, not statistically calibrated alerts or forecasts.

Seasonality

Monthly averages are calculated for calendar years with observations in all 12 months. At least two such years are required. Each year is normalized to an average of 100 before the corresponding months are averaged. Pattern consistency is the ratio of variation in the monthly profile to total normalized variation; it is not a confidence score or prediction accuracy.

Comparison & missing values

Comparison uses only dates shared by all selected series and rebases each to 100 from its first seven common observations. It compares relative evolution, not absolute popularity. Missing dates are not interpolated. CSV imports reject invalid dates, duplicate dates, missing values, negative values, and censored values such as <1.

Costs & credits

Radar requires no API key and consumes no demo credits. Production quotas and permitted reuse must be confirmed before paid monitoring is sold. Free and Pro limits and the proposed price are provisional; a credit is not yet a measured provider cost.

Watchlists & production status

Watchlists store local series snapshots and saved RSS topics. Successful Radar loads record summaries of up to 30 listed topics, with 12 checks per market and 60 total. Feed changes compare checks in the same market. Absence from a feed is not evidence that a trend ended. Rule evaluation checks series snapshots; Radar topics have observation history rather than numeric growth rules. No background refresh or notifications run. Demo credits are local. Direct Google integration requires alpha approval, implementation using Google’s supplied documentation and an end-to-end check with real data. Direct Bing integration remains open separately. Accounts, background monitoring, billing and advertising services remain outside this beta.