Competitor Ad Monitoring: A Weekly System for Decisions

What public sources give you

The Meta Ad Library shows ads currently running across Meta apps, with the advertiser page, start date and creative. Its official API returns ads about social issues, elections or politics, and ads of any type only when they reached the EU. A commercial ad that never reached the EU, such as a US-only subscription app campaign, shows on the Ad Library website but is not returned by the API. Check what the API returns for your target countries before you plan a script around it.

The Google Ads Transparency Center lets you look up an advertiser and see the ads it has shown, with formats and regions. TikTok publishes a Commercial Content Library whose API covers only some regions, such as the EU, and a Creative Center with top ads and trend views. Treat the Creative Center as a source of ideas, since it shows a curated selection of ads.

App store listings add a third layer. Screenshots, descriptions, subscription pricing and update notes change when a team shifts positioning or tests a new offer. Save these pages on a schedule, because stores show only the current version.

Rules of the road: ToS and GDPR

Use official APIs and published interfaces. Do not log in with borrowed accounts, bypass rate limits, rotate proxies to dodge blocks, or scrape pages the terms forbid. A library's search page, plus its API where the API covers your markets, gives you enough for a weekly routine, and staying inside those channels keeps your access stable.

Collect public ad content and metadata: creative, copy, start date, format, landing page URL and advertiser name. Skip personal data about people shown in or reacting to ads. Faces, names in comments and user handles can be personal data under GDPR, and you need a lawful basis to process them. Competitor monitoring seldom needs them, so leave them out of your snapshots.

Terms change. Check each source's current terms and API policy before you build, and again on a schedule such as each quarter. This is practical guidance, not legal advice. If you store data at scale or work in a regulated vertical, ask a lawyer.

What to track

Start with a short list of signals per advertiser, so the weekly review fits in about an hour.

• Active ad count and run duration per advertiser, recorded with the snapshot date
• New creative angles, hooks and formats, such as a founder video replacing a UGC testimonial
• Offer and price changes on landing pages and store pages, including trial length and discounts
• Landing page destinations, so you can see which pages receive the new creatives

Take an illustrative diff. Say a rival had 12 active ads last Monday. This Monday the library shows 15, and 4 of them are new while 1 from last week has stopped. The check: 12 plus 4 minus 1 equals 15. Read the 4 new ads first, then look at the stopped one to see which angle the team dropped.

Reading longevity as a signal

Advertisers usually switch off ads that lose money, so an ad that runs for weeks is likely profitable for them. Treat that as a hypothesis. The team may run it for brand reasons, may have a budget that hides weak returns, or may have forgotten to turn it off. Libraries show how long an ad ran. They do not show revenue.

Take an illustrative case. An ad has been live 45 days, and 45 divided by 7 is about 6.4 weeks. Six variants of the same angle are in rotation, which means the original plus five more versions. A team that keeps one angle alive for over six weeks and keeps producing versions of it probably sees it hold up. A single ad with no variants at 45 days is weaker evidence, since it may be running on autopilot.

Combine this with your own test data. If your tests of a similar angle show a weak click-through rate or a high cost per install, the competitor's longevity matters less to you, because your audience, price and funnel differ. Use longevity to choose which angles earn a test slot, and let your own results decide what stays.

The weekly cadence

• Monday: pull the snapshot and diff it against last week's.
• Tuesday: tag new angles into a swipe sheet.
• Wednesday: decide to test, ignore or counter, and give each decision an owner.

On Monday, save advertiser, ad ID, start date, format, copy, landing URL and the store page text. The diff then lists new ads, stopped ads and changed pages. On Tuesday, tag each new ad with angle, hook, format, proof type, offer and first-seen date. Consistent tags let you filter the sheet later.

On Wednesday, each tagged angle gets one of three outcomes. Test means a brief gets written. Ignore means a one-line note explains why. Counter means you change your own creative, landing page or offer in response. Write a name and a due date next to each outcome, because a decision with no owner is only a note.

From observation to brief

Turn each tagged angle into a testable hypothesis in one sentence: if you show this proof with this hook to this audience, the chosen metric will beat your current control. If you cannot fill in all four parts, the angle is not ready for a brief.

Then write the creative brief. Keep it short and specific.

• The angle in one sentence
• The proof you will use, such as a demo, a testimonial or a number you can substantiate
• Format, length and placement
• The metric and threshold you will judge by, and the spend or sample size you will wait for before reading it

Borrow the idea and produce original creative. Copying a rival's assets creates copyright and platform policy risk. After the test ends, log the result next to the competitor note with a win, loss or inconclusive label and the metric. After a few months the sheet shows which competitor signals predicted wins in your account, and you can weight them accordingly.

Automating it safely

A scheduled script or workflow can call the official endpoints, store each snapshot with a timestamp, compute the diff and post a digest to email or a team chat. Automate only the sources whose official endpoint returns your markets. Where none does, for example US commercial ads on Meta or TikTok, keep that source manual or use a data provider whose terms allow the use, and do not script the website. Options include a cron job running a Python script, or a workflow tool such as n8n, Make or Zapier. Keep raw snapshots in a spreadsheet or database so you can compare any two dates.

An LLM can summarize the diff and propose tags such as angle and hook type. A person should read the output before it reaches the swipe sheet, since models misread creative, invent patterns and miss context. Feed the model only the public ad content you collected, with no personal data.

Keep a source log with the endpoint or page, the date pulled, the version of the terms you checked and the person who approved the source. If a platform changes its terms or access, the log shows what to review.

Set up your first run this week

Keep the first version manual. Two manual cycles show which fields you use, and automation can then cover only those.

• Pick three competitors and one public source each for ads and store pages.
• Pull the first snapshot by hand and save it with the date.
• Create the swipe sheet with columns for angle, hook, format, offer, first seen and decision.
• Next Monday, pull again, diff against the first snapshot and tag the new ads on Tuesday.
• Hold the first Wednesday decision session and assign an owner to each outcome.