Log in to see these improvements for yourself.

Forecast accuracy tiers, recalibrated for every metric

The Elite / Top / Competitive / Developing tier shown next to each accuracy metric is now calibrated to that metric’s own scale. Temperature tiers are unchanged. For precipitation, probability of precipitation, and cloud-cover metrics, the tier badge previously placed nearly every provider in the “Elite” band, because the thresholds were set for temperature error in degrees Fahrenheit. Each of those metrics now has thresholds matched to its real accuracy distribution, so the tier genuinely separates a leading provider from a developing one, and wind tiers gain a fuller spread across the middle bands. Provider accuracy scores themselves were always correct; only the derived tier label was miscalibrated. Tier labels for these metrics are accurate from this release forward.

Report PDFs, available to every account as a watermarked sample

Every account can now generate any of the three platform reports, Monthly Insights, Market Overview, and Government Comparison, as a PDF. Accounts without the Sales & Marketing License get a watermarked sample: every page carries a diagonal “SAMPLE” mark and a footer band naming the license, and the file downloads with a _SAMPLE suffix so it is never mistaken for a finished deliverable. Accounts that hold the license generate clean PDFs. Every report is now generated from one place: the PDF button on Market Overview, Government Market, and Monthly Insights opens the Data Export screen with the report type and the current filters (month, location, days-out) already selected. That screen’s Reports source is always visible, offers Government Comparison as a third report type alongside Monthly Insights and Market Overview, and carries the banner that explains the license, with a “Request access” button that records the request and notifies the ForecastWatch team. On the Order Form, the Sales & Marketing License scope now covers “PDFs generated by the platform” rather than one named screen, so the permission tracks the platform as more screens gain PDF export.

API integration prompt in daily alert emails

When a Forecast Risks or Daily Insights email shows more stations in a category than fit on screen, the “and N more” overflow line, a callout now appears below that section: “Have too many stations to manually review? Integrate our API for scaled, automated verification,” linking to the API portal. On lighter-volume days where every flagged station already fits, the callout stays hidden so it adds no noise. The plain-text version of each email carries the same line.

WeatherStack accuracy, scored through day 13

WeatherStack’s /forecast endpoint now returns 14 daily entries instead of the original 7, and we collect and score the full range. Days 0 through 6 are unchanged. WeatherStack also appears in the days-out aggregations and bucketed views from day 7 through day 13, where it can be compared against the other longer-range API providers (WeatherAPI, Open-Meteo, Visual Crossing, Weatherbit). Existing scores at days 0 through 6 are unaffected, and the new days-out coverage begins from the deploy date forward.

/v1/insights/daily/ accepts a quadrant filter, and the catalog matches the API

There is a new ?quadrant=<label>[,<label>...] query parameter on /v1/insights/daily/. It validates against the known set (bold_wrong, bold_right, dragged_down, conventional_correct, unremarkable) and returns 400 on typos so a client integration fails loud rather than silently fetching nothing. The summary block stays unfiltered so the “landscape today” totals stay complete, while the stations array reflects the filtered slice. This halves the bandwidth on common queries like “find the chronic bold-wrong stations over the last month.”

The self-documenting catalog at /v1/api-docs/ also got a full sweep. Example responses now match the live API across more than 20 endpoints, including /v1/insights/daily/dates/ (the catalog described a bare list; the view actually returns {dates, count}) and the full shapes for /v1/insights/monthly/, /v1/insights/station-trend/, /v1/reference/stations/, and /v1/risks/overview/. Real provider names in the catalog examples have been replaced with a generic placeholder (“Your Forecast, Inc.”); government providers stay real since their identities are public.

Wettermanufaktur and Meteum are now in the correct market segments

Wettermanufaktur, a B2B energy-sector forecasting firm, moves to Enterprise Weather Intelligence, and Meteum, the consumer weather app descended from Yandex Weather, moves to Consumer Weather. Accuracy scores are unaffected; only the market grouping changes, and it is correct from this release forward.