Detect Price & Stock Changes | Dataset Diff

Spot price drops, stock changes, and catalog updates — compare two dataset snapshots and get exactly what changed as clean, field-level JSON. Point it at yesterday's scrape and today's scrape and it tells you which records were added, modified (with before / after values and numeric deltas), missing, and removed. Ideal for recurring e-commerce price monitoring and change detection — and it processes only user-supplied data, with no scraping.

Built for recurring monitoring: e-commerce price & availability tracking, catalog audits, lead-list change detection, and any workflow where you run a scraper on a schedule and only care about what moved.

• No scraping, no browser, no anti-bot games. It only processes data you supply — inline JSON, or two Apify datasets your own token can read. Zero third-party Terms-of-Service risk.
• Dependency-free & fast. Pure Python standard library. Typical runs finish in seconds.
• Truthful by default. It won't call a record "removed" from a half-finished scrape — see Missing vs. removed below.

---

What you get

Every run writes one summary row plus one change row per affected record:

// summary row
{ "recordType": "summary", "oldCount": 1200, "newCount": 1198,
  "added": 5, "modified": 41, "missing": 0, "removed": 7,
  "unchanged": 1152, "evaluatedKeys": 1205, "errors": [] }

// a modified record
{ "recordType": "change", "changeType": "modified", "key": "sku-1",
  "changes": [
    { "field": "price", "before": 19.99, "after": 17.99, "numericDelta": -2.0,
      "beforePresent": true, "afterPresent": true }
  ] }

// an added record (full record included)
{ "recordType": "change", "changeType": "added", "key": "sku-4",
  "record": { "id": "sku-4", "title": "Mechanical Keyboard", "price": 79.0 } }

---

Inputs

Field	Type	Default	What it does
oldDatasetId	string	—	Apify dataset ID of the previous snapshot. Your token must read it.
newDatasetId	string	—	Apify dataset ID of the current snapshot.
oldItems	array	—	Previous snapshot as inline JSON (used when oldDatasetId is empty).
newItems	array	—	Current snapshot as inline JSON.
keyFields	array	["id"]	Field(s) that identify a record across snapshots. Use several for a composite key (e.g. sku + variant).
compareFields	array	[]	Only check these fields. Empty = compare all fields except keys and ignored fields.
ignoreFields	array	["scrapedAt","fetchedAt","crawledAt","#debug"]	Volatile fields to skip so they don't create noise.
numericFields	array	["price"]	Fields compared as numbers; a numericDelta is computed and the tolerance applies. Currency symbols and thousands separators are tolerated ("$1,299.00" → 1299.0).
numericTolerance	integer	0	Ignore numeric changes at or below this absolute difference.
snapshotsComplete	boolean	false	See Missing vs. removed.
includeUnchanged	boolean	false	Also emit a row for records that didn't change.
maxRowsPerSnapshot	integer	50000	Safety cap; the run errors out instead of silently truncating.

Run it with no input and it compares a small built-in demo snapshot, so you can see the exact output shape before wiring up your own data.

Outputs

Field	Present on	Meaning
recordType	all	summary or change.
changeType	change rows	added / modified / missing / removed / unchanged.
key	change rows	The composite key of the record.
changes	modified rows	Array of {field, before, after, numericDelta, beforePresent, afterPresent}.
record	added / missing / removed rows	The full record.
oldCount,newCount,added,modified,missing,removed,unchanged,evaluatedKeys	summary	Counts.
errors	summary	Non-fatal issues (duplicate keys, missing key fields, skipped non-objects).

---

Missing vs. removed — why it matters

If a record is in the old snapshot but not the new one, that can mean two very different things:

• The product was genuinely removed, or
• Your new scrape was incomplete (a timeout, a blocked page, pagination that stopped early).

Treating an incomplete scrape as mass deletions is how monitoring pipelines send false "everything is gone" alerts. So by default this Actor labels those records missing (a soft signal). Set snapshotsComplete: true only when you are confident both snapshots are complete crawls — then they are labelled removed.

---

Example: daily price monitoring

1. Schedule your product scraper to run every morning; each run produces a dataset.
2. Schedule this Actor right after, with:
   • oldDatasetId = yesterday's dataset ID
   • newDatasetId = today's dataset ID
   • keyFields = ["sku"], numericFields = ["price"], numericTolerance = 0
3. Wire the output to a webhook / integration. Every modified row with a price change and a negative numericDelta is a price drop worth acting on.

Notes

• Matching is by exact key equality; pick keys that are stable across runs (an internal id or sku, not a position or a timestamp).
• Records with duplicate keys keep the last occurrence and are reported in errors.
• The Actor reads and writes datasets only through the documented Apify REST API and never logs your input values.