Use Compare Datasets to Detect New, Updated, and Deleted Records

The Compare Datasets node compares two snapshots of data and categorizes records into three buckets: new, updated, and unchanged (or deleted). This is the foundation of incremental sync workflows that avoid reprocessing the entire dataset on every run.

Real-world example: Every hour, you pull the subscriber list from Mailchimp. You need to detect which subscribers are new (to send a welcome Slack message), which changed their email preferences (to update your CRM), and which unsubscribed (to log the churn).

Workflow structure:

[Get Current Subscribers] ──────► [Compare Datasets] ──► [New] ──► Slack Welcome
         (Mailchimp)                     ▲               [Updated] ──► CRM Update
                                         │               [Deleted] ──► Churn Log
[Get Previous Snapshot] ────────────────┘
         (Database)
```text
Configure the Compare Datasets node:

| Setting                    | Value                            |
|----------------------------|----------------------------------|
| Input A (current data)     | Connected to Mailchimp node      |
| Input B (previous data)    | Connected to Database node       |
| Fields to Match            | `email`                          |
| Fields to Compare          | `status`, `preferences`          |

The node produces three output branches:

```json
// Output 1 — New items (in A but not in B):
{ "email": "newuser@example.com", "status": "subscribed", "name": "New User" }

// Output 2 — Updated items (in both, but fields differ):
{
  "email": "existing@example.com",
  "status": "unsubscribed",  // changed from "subscribed"
  "name": "Existing User"
}

// Output 3 — Unchanged items (in both, fields identical):
{ "email": "stable@example.com", "status": "subscribed", "name": "Stable User" }
```text
> **Tip: Save the Snapshot**
>
> After comparison, upsert the current dataset into your snapshot table so the next run compares against fresh data. Use `ON CONFLICT (email) DO UPDATE` in Postgres or equivalent.

This pattern reduces API calls, prevents duplicate notifications, and gives you audit-ready change tracking.

**Related:** [Flatten Deeply Nested API Responses](../code-node-mastery/01-flatten-deeply-nested-api-responses.md) | [Use the HTTP Request Node as a Universal Connector](../integration-patterns/01-use-the-http-request-node-as-a-universal-connector.md)