Demand Outlook

What you can build

• Two weeks of contacts for the scheduling team

  Demand Outlook projects contacts by channel per day from Amazon Connect history and the warehouse, marks the two public-holiday days from the Nager feed, and proposes the baseline to the scheduling sheet.

  With Amazon Connect, BigQuery, Holidays & FX Reference, Google Sheets

• Snow week for a regional business

  Open-Meteo shows snowfall for three of the next five days in the main region. The operator applies the historical snow-day effect to orders and proposes a revised baseline to the planning channel.

  With Open-Meteo, Snowflake, Slack

• Warn inventory before a peak day

  A projected order day exceeds the pick capacity recorded for the warehouse. The operator proposes a Jira task to the inventory owner with the day, the projection, and the history behind it.

  With BigQuery, Jira

Requirements

• History tables with at least the prior year of daily orders and contacts, readable by the warehouse connector
• An Amazon Connect instance the connector can query with GetMetricDataV2, for contact history and channel mix
• Your countries and regions, so the holiday and forecast signals know what to fetch
• A planning sheet and channel where the baseline is proposed, and a named owner who approves it

Authentication

Demand Outlook has no keys of its own. Warehouse reads go through the BigQuery or Snowflake connector, contact metrics through the Amazon Connect connector, holidays and weather through the signals, and writes through Google Sheets, Slack, and Jira.

Limits

• GetMetricDataV2 returns the last three months and at most thirty-five days per request, so longer contact history comes from your warehouse.
• Open-Meteo forecasts run sixteen days; the National Weather Service forecast runs seven. Days past the horizon carry only history and holidays.
• It projects volume. It does not schedule staff, place purchase orders, or change a forecast someone else owns.
• A projection is only as good as the history behind it. A channel launched last month has no seasonal pattern yet.

Questions and answers

What does a person approve?

The baseline: fourteen rows of projected orders and contacts, each with the history it drew on, the holiday flag, and the weather adjustment. You approve it as proposed or edit any day first. A mid-period revision is a separate proposal showing what moved.

What record does it leave?

Each baseline is kept with the history window it used, the holiday and forecast data as read, the method applied per day, who approved it, and the actuals recorded afterward. The next baseline shows the previous one's error by day.

Does it write the forecast into planning tools on its own?

No. It proposes the baseline to a sheet and a channel and waits. It never creates schedules, purchase orders, or forecasts in another system, and it never posts before approval.