The Integrations Hub

Integration Core · checked 2026-10-15

The Integrations Hub, in the Desk sidebar, answers one question: is anything waiting for a person? It has six count tiles across the top and five cards of links below them.

Your daily check

Open the Hub once a day:

1. Connection Health first (on Your connections): every connection in one line. A connection that is not Live explains many stuck-looking records.
2. Failed — fix the cause, then retry.
3. Cannot tell if it arrived — leave alone.
4. Needs a decision — decide, then close each with a note.
5. Reversals not confirmed (on Day to Day) — the other system has not confirmed them; call support if one is old.
6. Money records waiting — how many money records are in flight.
7. Set aside (on Waiting) — usually expired sign-in details.

Once a week: Waiting on a setup change and Differences.

The bell tells you when something is genuinely stuck, but one alert can stand for many records: count the rows, not the alerts.

The tiles

Each tile counts something that needs a person. Click it to open the list of exactly the rows it counts; the list keeps the tile's filters, shown as pills under its title. Test records are never counted.

Needs a decision

Records waiting for a person to decide what happens to them. See Needs a decision.

Failed

Records that did not go through and that nobody has acknowledged. Records that need a decision and records nobody can tell about have their own tiles. See Failed.

Cannot tell if it arrived

Records whose outcome cannot be confirmed (Unverified). Never press anything to clear them. See Unverified.

Money records waiting

Money records not finished and not decided, coming in and going out. A count of records, not a sum of money: a sale and its payment both count. Most of them are also on another tile; this is the one number that says how much money is in flight.

Differences

Open places where the two systems disagree about a record. See Differences.

Connections with a problem

Switched-on connections that are refused the sign-in, cannot reach their system, stopped part-way, are late looking for changes (past their Check Every (minutes) plus an hour), or have not reported yet. It notices a connection that has gone quiet at its next hourly check, so it can be up to an hour behind Connection Health. See What the word on a connection means.

Getting started

The five set-up steps, in order: 1. Add an Integration Manager · 2. Connect a system · 3. Test it (on its setup page) · 4. Switch it on · 5. Watch the first run, and Help. Each is explained on Set it up.

Your connections

Connections (the list of every connection) and Connection Health, then one report per thing that can be wrong with a connection: Cannot reach it · Sign-in refused · Stopped part-way · Switched off · Not checked for over a day · Not reported yet. See Connection reports.

Each connector also gets a card of its own on the Hub, named for its system, with a link to its setup page.

Waiting

One report per reason a record can wait: Kept while switched off · Held behind an unknown attempt · Waiting on a setup change · Waiting, by cause · Set aside · In progress. See Waiting records.

Day to Day

Integration Activity (every change that crosses), Incoming Records (what arrived, as it arrived), Differences, and the report Reversals not confirmed (Waiting records).

Advanced

Record Links (which record here is which record there) and Sync Positions (how far each sync has got). The system keeps both; see Record Links and Sync Positions.

The bell

The bell at the top of Desk carries an alert when something is genuinely stuck: a reversal not confirmed for four hours, a connection that stopped looking for changes, refused sign-in details, a system that cannot be reached. Each alert says its cause in its first words; the short code in brackets at the end is for support.

• Alerts are grouped per connection, per kind, per hour, so one alert can stand for many records.
• A connection's alert goes to the Integration Managers who may see that connection; if none may, to the System Managers.
• Every alert comes from a scheduled job: if the site's scheduled jobs stop, nothing rings.
• A connection that has never run raises no alert at all. It shows Not reported yet, and somebody has to look.