Exporting sales data from Toast

The mechanics of getting data out of Toast Web aren't complicated, but there are specific pitfalls that break downstream inventory work. This article covers the format, date, location, and file-type details so you pull exports that actually work.

Which exports to pull

Two, for inventory:

1. Item Selection Details — one row per menu item per order
2. Item Modifier Selection Details — one row per modifier per order

Both live under Toast Web → Reports → Sales. See the dedicated article for what's in each.

Other Toast reports (Product Mix, Sales Summary, Menu Engineering, labor reports, kitchen display reports) are either aggregated — losing the granularity you need — or for other jobs entirely. Skip them for inventory.

Match the date range across both

Both files must cover the same date range. Downstream, we link rows by Order # — if Item Modifier Selection Details covers Feb 1–Mar 31 and Item Selection Details covers Mar 1–Apr 30, the February modifier rows have no parent items and the April item rows have no modifier context.

Practical cadence:

• Pull 60 days at a time. That matches the trailing window for PAR math.
• Pull monthly going forward. Once you're live, pull a rolling trailing window — add new data as it arrives.
• Weekly is overkill. Toast's data doesn't change after the business day closes. You don't need to re-pull last week's data.

CSV vs Excel

Toast offers both:

Par Inventory's ingest handles both, but CSV round-trips faster and is less likely to trip up on sheet-level weirdness.

Date format

The Sent Date column comes out as MM/DD/YYYY HH:MM in most Toast accounts. A few notes:

• The date portion is the business date, which aligns to the operator's defined business day (many restaurants define the business day as running 4am–4am). A late-night sale at 2am on a Tuesday calendar day often posts to the Monday business day.
• When rolling up to daily counts, use business date, not calendar date. Par Inventory's ingest parses the date-only portion of Sent Date and applies it as the business date.
• Two-digit years are possible on older exports. Our parser handles MM/DD/YY by assuming 21st-century dates.

Multi-location exports

If you run more than one Toast location under the same account, Toast's exports can include rows from every store by default. The Location column tells you which.

Our recommendation: re-export filtered to one location at a time. Toast Web has a location filter in the report settings — set it before running the report. This keeps files smaller and avoids a class of ingest errors.

Par Inventory's upload surface detects multi-location files and flags them with a "Multiple locations in this file — re-export per-location" warning. The data will still ingest, but it will all land under the store you uploaded under. For multi-location operators this is wrong and should be fixed before running PAR math.

Voided rows

The Void? column is either blank / No or Yes. Voided rows are not sales — a customer changed their mind, a barista miskeyed, something. If you include them in usage calculations, you'll overestimate.

Par Inventory's ingest filters voided rows out of the daily rollups. If you're doing ad-hoc analysis in a spreadsheet, filter Void? = No (or the blank equivalent) before counting anything.

Re-uploading data

Pulled a 60-day export last month and now want to refresh with the latest 30 days? Safe to upload — we dedupe by (store, business date, menu item / modifier). Overlapping date ranges replace old rows, don't double-count.

The API alternative

Toast has an API that exposes the same underlying data. If you have partner credentials, you can skip the manual-export dance entirely. The downside: Toast API access requires a partner agreement, which most independent operators don't have. Manual CSV upload is the more common path for single-store operators.