Watch competitor Facebook ads on a 24h schedule (auto-refresh + New badge)

The boring loop ad spy actually is

Watching a competitor's paid creative is high-leverage research — the ad they kept running for 8 weeks converted; the one they killed in 3 days didn't. But manually visiting the Meta Ad Library every morning to spot what's new is exactly the kind of repetitive admin work marketing teams quietly drop after week 3.

KOLens' original Ad Spy already let you scrape any brand's full creative wall in one click. What was missing: a passive loop that does the daily check for you and tells you when something actually changes.

How the watchlist works

1. 1

   Open any advertiser detail page.

   Either scrape a fresh brand on /ad-spy, or click into one already in the workspace data store from /ad-spy/watchlist.
2. 2

   Click 'Watch · 24h refresh' next to the page name.

   The toggle adds a per-user row in the watchlist table. Re-clicking unwatches. Optimistic UI — feels instant; rollback on error.
3. 3

   Wait — KOLens re-runs the scrape every 24h.

   The scheduler picks any watch row whose last_refreshed_at is older than 24 hours, calls Apify against the advertiser's page URL, and stamps the timestamp. Capped at 5 refreshes per scheduler tick so a single tick stays bounded.
4. 4

   On your next visit, new ads light up.

   Each ad card whose first_seen_at is later than your per-user last_viewed_at gets a small New pill in the corner. After one second on the page the timestamp bumps forward, so badges only clear after you've had a chance to see them.

The aggregate view

/ad-spy/watchlist rolls every watched advertiser into one list with a per-row "N new" pill. The page header sums them across the whole watchlist:

• 3 advertisers · 14 new ads since you last looked — the headline number you actually want when triaging the morning.
• Per-card last-refresh timestamp so you can spot a watch that's stuck (e.g. an advertiser that deleted their Facebook page → no more ads to scrape).
• Pause / resume from the same row when a brand goes quiet for the season.

Design choices that actually mattered

Free for the user, on the platform's dime

Charging credits per auto-refresh would have made the Watch button feel adversarial — "do you really want this?" anchored against an increasing bill. We made it free instead. The Apify cost is real and bounded (24h cadence × small refresh batches) and the watchlist is a sticky retention feature, so it pays for itself.

Apify errors don't bump the timestamp

When Apify hiccups (~1% of runs, usually a captcha rotation), we log the failure but leave last_refreshed_at alone. That way the next scheduler tick picks the row up again instead of creating a 24-hour data hole when the actor was down for an hour.

"New" is per-user, not global

Two users on the same workspace shouldn't see each other's New badges. Each watch row carries its own last_viewed_at — opening the page bumps your timestamp, leaving everyone else's intact. Same advertiser, two private notification states.

vs alternatives

• AdSpy / BigSpy / Magic Ads — large indexed libraries with subscription billing. KOLens doesn't compete on breadth; we compete on freshness + watch-friction (one click to star, one click to disconnect, no annual contract).
• Manual Meta Ad Library — Meta's free tool. Fine for one-time research; doesn't loop, doesn't diff, doesn't notify.
• Apify alone — viable if you're an engineer who wants to roll your own. KOLens packages the scrape + deduplication + per-user New tracking + UI so a marketer can use it without writing scheduler code.

Coming next

On the roadmap (tracked as AS-5 follow-ups): push the New events out via the same email + webhook delivery we just shipped for KOL alerts. Star a brand → first new creative drops → Slack ping the same minute KOLens detects it.