Anomer
Sign in

Why did orders drop last week?

Ask why a number changed. Anomer checks your ads, Instagram, orders and weather, and shows how much each cause contributed.

Why were orders lower Oct 12–18?

Oct 5–11
1,105
Ads paused
-125
Instagram
+3.4
Trend
+3.1
Weather
+2.3
Unexplained
-10.1
Oct 12–18
979
Simulated bakery with a planted ad-pause effect of about −129
-129.4
Planted ad-pause effect (orders)
-124.8± 29
Anomer's estimate (95% range)

The demo shop has a planted effect from pausing ads for a week. Without knowing the answer, Anomer recovered this value from the change in orders.

From question to answer

  1. 1

    Muse Spark reads the question

    It picks the number and the two periods to compare, and leaves out numbers that are results rather than causes (such as orders when asking about revenue).

  2. 2

    Lasso tests every candidate

    Every connected number, shifted 0–7 days, goes into one regression with the day of the week, the day of the month and the long-term trend. Candidates that don't matter drop to zero.

  3. 3

    Muse Spark explains the result

    It gets only the computed breakdown and explains the main causes in plain words.

Data you connect

Every data source is a plugin with the same shape. Adding one adds candidate causes.

  • Orders

    CSV: order_id, created_at, customer_id, total

    Orders, Revenue, Average order value, New customers, Returning customers

  • Meta ads

    Marketing API Insights

    Ad spend, Ad impressions, Ad reach, Ad clicks, Ad CTR, Ad CPC, Ad CPM, Ad purchases

  • Instagram

    Instagram Graph API

    Accounts reached, Views, Likes, Comments, Saves, Shares, Profile link taps, Feed posts, Reels

  • Weather

    NOAA GHCN-Daily

    High temperature, Low temperature, Precipitation

Code computes the numbers

The AI makes up no numbers

Every number in the breakdown comes from the regression. The AI only writes the explanation.

Ranges, not just points

Each cause has a 95% range. When the range crosses zero, the answer says it may be noise.

The rest stays visible

Whatever the causes don't explain is shown as "Unexplained".

Questions

What data do I need?
Upload orders as a CSV with order_id, created_at, customer_id and total. Ads and Instagram are read in the shape of Meta's Graph API, and weather only needs a city.
How much history do I need?
The demo shop uses 400 days. About a year lets Anomer see monthly patterns such as paydays.
Is the demo data real?
Weather is real NOAA data for Portland. Ads, Instagram and orders are simulated, with causes planted so the answer can be checked.
What happens to uploaded orders?
They are stored as daily counts and totals. Individual orders are not kept.

When a number moves, ask why.