Transfer only what changed
"Artikel Cambira Austria" runs at night at 2:00 AM and each time reads all 1,186 items that its data filter covers, each as a job of its own — of which around 70 have changed. Only these 70 are transferred; the rest end as Skipped (no change). This guide sets up the route so that it reads only the changed ones. Why that is the rule and the full run the exception is described in the core concept Triggering.
Three places are responsible for this: the Change filter decides which records are read at all, the Run type switches the schedule over to it, and Change detection sorts out what was read but is unchanged.
What you need
- The permission Routes at the level Full — see Permissions.
- A route with a schedule and a fetch endpoint whose field structure is resolved.
- Starting point: Configuration → Routes, then select the route.
Xchange saves every entry automatically. It takes effect in operation only with the activated configuration snapshot — that is step 4.
1. Check the change filter
It is on the Source data tab in the collapsible section Change filter, next to the hint "for incremental runs".
- Open the Source data tab and expand the Change filter section.
- Usually something is already there: Xchange enters the date fields of the source once by
itself as soon as their field structure is available — for SAP items
UpdateDateandCreateDate. Just check that the rows are there. - If the section is empty: Add field and choose the date field of the source. There is nothing more to set — there is no operator and no value.
- With several fields, a record comes along as soon as one of them has changed since the last
run. A number field such as
DocEntrydetects only newly created records — detecting changes needs a date field.

If Source returns a change token is chosen under Detect changes on the connection, the section shows the sentence "The fetch endpoint returns only the changes since the last run. A change filter is not required here." instead of the fields — then go straight on to step 2.
The fields of the section are listed in Routes — Source data.
2. Set the run type to "Incremental"
- Open the Automation tab and select the route's scheduled trigger.
- Under Run type, choose Incremental.
If the change filter from step 1 is missing, the Source data tab now reports it as a required field — hence this order.
If the route reads through a REST connection that pages by Page number or Offset, the connection needs a sort — unless it is set to Source returns a change token. Without the sort, the check reports "Connection is not set up to read in sorted order" at the Run type. What to enter on the connection is described under Connections — Batch fetch.
If a Cleanup is attached to the route, keep a full run next to it, e.g. weekly: a route may carry several triggers, and the cleanup runs only along with the full run. The fields of the trigger are listed in Routes — Automation.
3. Take fields out of change detection
Before every transfer, Xchange compares the result of the mapping with what it last sent for the same record. If both are the same and the record has not changed in the target system since, the transfer is dropped.
This comparison comes to nothing if the mapping produces a new value on every run — a row with the transformation Current timestamp or New GUID, for example. Then the result always differs, and every record read counts as changed. Take such rows out of the comparison:
- Open the Mapping tab and find the row in question.
- Click the gear at the end of the row.
- In the dialog, switch to the Options tab.
- Under Change detection, check the Exclude field checkbox.
- Close the dialog.

Leave at least one row in the comparison. Xchange points this out: "All mappings are excluded from change detection — changes will never be detected". The remaining fields of the dialog are listed in Routes — Mapping.
4. Validate and activate
Your changes are in the draft. Validate it and activate it — only then does the next nightly run run incrementally. The way there is described in Validate and activate.
Only an incremental run remembers the position it reads from: per field of the change filter, the largest value it has read. The first incremental run after activation does not have this position yet and therefore processes the full scope once — the UI announces it with "processes the full scope — no position has been stored yet.". For "Artikel Cambira Austria", the first night therefore still reads all 1,186 items, and only from the second on just the changed ones. How you keep this run small is described in Limit the first run.
5. See whether it works
You can start the first incremental run right away, without waiting for the night:
- Open Execution and select the route in the list.
- Start run, choose Incremental as the run type, Start.
If the dialog offers no run type and only asks "Start a full run for this route now?", the route cannot do an incremental run. The sentence below it names the reason:
| Sentence | What is missing |
|---|---|
| "Incremental runs become possible once the route's change filter contains at least one field." | step 1 |
| "The connection does not read in sorted order. Set up a sort on the connection to make incremental runs possible." | the sort on the connection from step 2 |
| "The connection lacks details for the chosen change detection. Complete them on the connection to make incremental runs possible." | the details of the option chosen under Detect changes on the connection |
In that case, cancel instead of starting a full run.
Two places then show that it works:
- In Execution, the route shows Next incremental run and below it, per field of the change
filter, a line such as "reads UpdateDate from … onwards", with the largest value the run has
read. Because SAP Business One keeps
UpdateDateonly to the day, the next run reads that day's records again — they are sorted out by the change detection from step 3. For a field with a time of day, the run starts five minutes earlier, so changes saved at the same moment are not lost. - In History, the unchanged records carry the status Skipped (no change). If a later run for "Artikel Cambira Austria" reads around 143 items and about half of them have this status, the setting is right.
See also
- Narrow down with a data filter (guide) — transfer less by having the route take only a subset into its scope from the start
- Limit the first run (guide) — so that the first incremental run stays small instead of taking along the whole existing dataset
- History (reference) — the status values of a job in detail
- Triggering (core concept) — when a full run is still the right one