Amazon listing monitor: title, price, image and buy box change alerts

Give it a list of ASINs. Every time it runs, it tells you what Amazon changed on each one since the last time you looked, with the old value next to the new one.

Someone edits your title. A bullet point gets rewritten. Your main image is swapped for a competitor's. The buy box quietly moves to another seller. All four happen without a notification from Amazon, and all four cost money for as long as nobody notices.

This is the only Actor in this family built to run on a schedule rather than on demand. It keeps a snapshot of every product you watch and compares against it, so what you get back is not another catalogue dump but a short answer per product: changed, unchanged, or seen for the first time.

Read this before your first run

Three things decide whether this is what you want, and all three are easier to learn now than after a month of runs.

The first run cannot find changes. There is nothing to compare against yet, so every product comes back as first-seen and every one is charged. That first run is what makes the second one useful. Budget for it: watching 500 products costs $2.03 the first day and the same every day after, because the price is per product checked, not per change found.

It charges for the answer, not for the news. A product that did not change is a result you asked for and paid to be sure of. Charging only when something changed would sound generous and would make this Actor lose money on the quiet days, which are most days, and an Actor that loses money gets switched off. The honest version is the one you are reading.

It watches eleven fields, and six of them only when Amazon shows a buy box. Always compared: title, bullet points, main image, brand, and the number of other sellers. Compared only when the buy box arrived on both checks: price, list price, coupon, price source, stock status and availability text. Every row carries buyBoxSeen so you can tell which kind of check you got. Ratings and review counts are deliberately not watched: they drift every day on their own, so watching them would mean an alert every day about nothing.

What it answers

Every row carries one of three verdicts in changeStatus.

Verdict	What it means
first-seen	No snapshot existed. One has been saved. Nothing to compare yet.
unchanged	Every watched field is identical to last time.
changed	At least one watched field moved. changedFields names them, changes holds the old and new value.

A row with no verdict at all is a product whose page did not arrive complete. That row is delivered anyway, with whatever did arrive, and it is never charged for. This matters more than it sounds: a half-loaded page is missing bullet points, and comparing that against yesterday would report a bullet rewrite that never happened. A false alarm is the one failure this Actor cannot afford, so it refuses to guess.

Every field it returns

Price and offer

Field	What it holds
price	Current price from Amazon's own buy box data
listPrice	The struck-through price, when Amazon shows one
currency	Three-letter code, from the page itself
priceSource	Where the price came from, so you can audit it
coupon	The clip-coupon text, when there is one
otherSellersCount	How many other sellers offer it

Stock and delivery

Field	What it holds
inStock	True, false, or null when Amazon does not say
availability	Amazon's own availability sentence

Identity and catalogue

Field	What it holds
asin	The product id
title	Full product title, shortened only past 1000 characters
brand	Brand name
marketplace	Which shop this row came from
url	The page this was read from
scrapedAt	Timestamp of the read

Description, media and variants

Field	What it holds
features	The bullet points at the top
image	Main image at the largest size Amazon offers

What changed

Field	What it holds
changeStatus	changed, unchanged or first-seen, the verdict for this check (this is what you are charged for)
changedFields	The names of the fields that moved since last time
changes	The old and the new value for each of those fields
checkedAt	When this check ran
previousCheckedAt	When the snapshot it compared against was taken
buyBoxSeen	Whether Amazon served the buy box on this check, which decides whether price and stock could be compared at all

The shape of a change

changedFields is a plain list you can filter on. changes holds the detail:

"changeStatus": "changed",
"changedFields": ["title", "price"],
"changes": {
  "title": { "from": "Echo Dot (4th Gen)", "to": "Echo Dot (4th Gen, 2024)" },
  "price": { "from": 49.99, "to": 39.99 }
},
"checkedAt": "2026-08-30T06:12:44.108Z",
"previousCheckedAt": "2026-08-29T06:11:57.402Z"

Filtering for changeStatus = changed gives you your alert list. Filtering changedFields for priceSource gives you the buy box events on their own, which is usually the list people actually want first.

What each verdict is worth watching for

Title and bullet points. On Amazon anyone with the right permissions can edit a listing, and brand owners routinely find their copy rewritten by a reseller or by Amazon's own catalogue automation. The change is silent and permanent until someone contests it.

Main image. Swapped images are a common way for a competing seller to make a listing represent their variant instead of yours.

Price and list price. Not as a price tracker, but as a change alert: a from and a to on the same row is what you forward to somebody.

Coupon. A coupon appearing or disappearing on your own listing when you did not put it there is worth knowing the same day.

Price source. This is the buy box, and it comes with an honest limit that is worth reading twice. Price, list price, coupon, stock and the buy box itself only exist on the page when Amazon serves the buy box, and Amazon does not always serve it. Measured here on 30 August 2026: on one run, thirteen of twenty products arrived without a buy box at once. Nothing had changed on Amazon.

So these six fields are only compared when the buy box was there both times. When it was not, they are left out of the verdict entirely and buyBoxSeen is false on that row, so you can see why.

The cost of that rule is stated plainly: a buy box that genuinely moves to another seller will not be reported as a change, because from outside there is no way to tell that apart from Amazon not serving us the buy box. What stays reliable is title, bullet points, image, brand and seller count, and those do not live in the buy box.

Other seller count. Going from one seller to four overnight means somebody has started listing against you. This one is read outside the buy box, so it works on every check.

What to put in

Products, the same three ways as everywhere in this family:

• an ASIN on its own, B09B8V1LZ3
• a product link, https://www.amazon.com/dp/B09B8V1LZ3
• pasted text with an ASIN in it, ASIN: B09B8V1LZ3, which is what you get when you copy out of a spreadsheet

You can also give search terms or category links in search, and it will monitor whatever comes back. Each product gets its own snapshot, so a list that shifts between runs still works. For a fixed watch list, productUrls is the one you want.

Settings worth knowing about

Setting	What it does
productUrls	The products to watch. ASINs, links, or pasted text.
search	Search terms or category links, if you would rather watch whatever ranks.
marketplace	Which Amazon. Nineteen available. Snapshots are kept per marketplace, so the same ASIN on .com and .de never mix.
stateStoreName	Where yesterday's snapshots live. Leave it alone unless you run two separate watch lists.
maxItems	Hard ceiling on products for this run.
maxItemsPerList	How many products to take from each search term or category.
maxListPages	How deep to page through a search or category.
proxyConfiguration	Which proxy to use. Datacenter by default.
proxyCountry	Which country to exit from. Empty means follow the marketplace.
residentialFallback	Switch the later attempts to residential addresses. Off by default; it costs more and, measured on this engine, it did not deliver more.
maxAttempts	How many times to retry a product Amazon refuses.
lanes	How many products to work on at once.
minDelayMs maxDelayMs retryPauseMs	The pacing. The defaults are deliberately unhurried.

The snapshot store, in plain terms

stateStoreName is the name of a filing cabinet in your own Apify account. Each product gets one folder in it, holding what the listing looked like the last time this Actor saw it. Nothing else reads that cabinet and nothing else writes to it.

Two things follow from that. Changing the name starts a fresh cabinet, so the next run reports everything as first-seen again and charges for it. And two different watch lists that should not share a baseline need two different names.

The default, amazon-listing-watch, is right for almost everybody.

How it works, in plain terms

For each product it fetches the product page, reads the eleven watched fields, and looks in the cabinet for what it saw last time. If there is nothing there, it files today's copy and reports first-seen. If there is, it compares field by field, reports what moved, and replaces the copy with today's.

Two details make the comparison trustworthy rather than merely automatic.

Empty, blank and missing are treated as the same thing. A coupon that is absent comes back sometimes as nothing, sometimes as an empty text, and sometimes not at all, depending on how Amazon built the page that second. All three mean there is no coupon, so none of them counts as a change. Without that rule this Actor would alert on its own reading noise.

And the page has to be complete before any verdict is given. A page that arrived with a title but no bullet points is not evidence that the bullet points were deleted.

If Amazon refuses a page, the product is retried from a different exit address, up to maxAttempts times, and once more in a second pass at the end of the run. Only after all of that does it come back without a verdict, and free.

What you can build with it

A daily hijack alarm. Run it every morning on your own catalogue, filter for changedFields containing priceSource or otherSellersCount, and route those to email or Slack. That is the whole product several companies sell as a subscription.

A listing integrity log. Keep every run's dataset. Over a few months you have a dated record of every edit anyone made to your listings, which is exactly what Amazon asks for when you contest a change.

Competitor watch. Point it at twenty competitor ASINs instead of your own. Their price moves, coupon launches and stock changes arrive on the same schedule as your own.

Supplier and reseller compliance. If you set minimum advertised prices, a daily price change list against your distributor's ASINs is the enforcement report.

Catalogue drift detection. Amazon's own automation rewrites listings too. Running this across a large catalogue surfaces the ones that quietly stopped matching what you uploaded.

Working it into your stack

Set a schedule in Apify and let it run at a fixed hour. Add a webhook on run completion so your own system gets the dataset the moment it exists, or pull it from the API:

https://api.apify.com/v2/datasets/<datasetId>/items?clean=true&format=csv

Filtering for the alert list is one parameter away, since changeStatus is a plain string in every row.

For a first look, run it twice by hand on the same ten products a few minutes apart. The first run will say first-seen ten times, the second unchanged ten times, and that pair tells you more about what you are buying than any description can.

Marketplaces

Nineteen, from amazon.com to amazon.co.jp, and pages are read in the shop's own language. Snapshots are stored per marketplace, so the same product watched on two Amazons keeps two separate histories and never reports a change that is really just a different shop.

One caveat carries over from the rest of this family: on some marketplaces Amazon does not show a buy box to visitors it cannot place, and where there is no buy box there is no price and no availability. On those marketplaces the price and stock fields stay empty, and because empty and empty are equal, they simply never report a change. Title, bullet points, image and brand still work everywhere.

Pricing

Four dollars per thousand products checked, plus three cents to start a run. That is four tenths of a cent per product per check.

Watching 100 products daily costs 43 cents a day, or about $13 a month. Watching 500 costs $2.03 a day. You are not charged for a product whose page did not arrive complete, or for one that never arrived at all. Platform usage is included, so there is no separate compute bill and no proxy bill.

The comparison is charged the same whether it found a change or not, for the reason at the top of this page.

The rest of this family

This Actor is one of twelve on the same engine. If your question is not "what changed", one of these fits better:

• Amazon Price Tracker when you want the price itself on a schedule rather than an alert when it moves.
• Amazon Stock Checker for availability and delivery promises.
• Amazon Offer Checker when the question is who else is on the listing and on what terms.
• Amazon Product Scraper when you want all sixty-four fields per product.
• Amazon Bestsellers Scraper and Amazon Search Results Scraper to build the watch list in the first place.

They share the retry engine, the pricing model and the rule about never guessing, so what you learn on one carries to the others.

Questions people ask

How often should I run it. Once a day covers most listing edits and buy box events. Several times a day is reasonable for a small, high value list and still cheap.

Why did everything come back as first-seen. Either it was the first run, or stateStoreName changed. A new store name means a new baseline.

Can I get alerted only when something changed. Yes, but do that on your side: filter the dataset for changeStatus = changed. The Actor delivers all three verdicts because "nothing changed" is also an answer worth having, and because a run that only returns news gives you no way to tell news from a broken run.

Does it tell me who changed it. No. Amazon does not publish that, and seller names are personal data this project does not collect. What you get is what changed and when.

Why is changes empty on a first-seen row. Because there is nothing to compare against yet. The snapshot has been saved; tomorrow's run has something to work with.

What happens if a run is interrupted. Products already checked have their snapshots saved. The rest are simply not updated, and the next run picks them up against their older baseline.

What it does not collect

No review text, no reviewer names, no customer questions and answers, no seller names. Review text and reviewer names are personal data, and scraping personal data is almost always a problem under European privacy law regardless of what you intend to do with it. Aggregate numbers, the star rating and the review count, are available in the sister Actors that report them.

You can see how many sellers are on a listing and watch that number move. You cannot see who they are.

Legal note

This Actor reads pages that Amazon serves publicly, at a deliberately unhurried pace. It is not affiliated with, endorsed by, or connected to Amazon.com, Inc. What you may do with the data it returns depends on where you are and what you use it for, and that is your call to make.