adlibrary Changelog, April 8–21, 2026: Meta, LinkedIn, Spend, Download

Two weeks of shipping at adlibrary, and the quiet map of the product changed more than the homepage did. Meta ads now run through a dedicated client. LinkedIn joined Meta as a first-class source. Every ad card carries an outbound link to the advertiser's landing page and a one-click download for the media. Spend estimates landed everywhere we have enough signal to back them. The soft paywall A/B test ended.

This is the adlibrary changelog for April 8–21, 2026 — what shipped, how to use it, and the small decisions behind each ship.

TL;DR: Between April 8 and April 21, 2026, adlibrary shipped native Meta Ad Library support, first-class LinkedIn ads, outbound landing-page links and one-click media downloads on every card, ad spend estimates across Meta, LinkedIn, and federated results, and faster streaming. The soft paywall A/B lost 11-to-3 to the hard paywall at equal retention, so the test was retired.

Meta Ad Library, now a direct source

Until this sprint, Meta ads arrived at adlibrary as one of several federated results — good for breadth, less good when you needed every ad an advertiser was running. That changed. Meta Ad Library is now a dedicated source with its own client.

You can feed it three things:

• A full Meta Ad Library URL (https://www.facebook.com/ads/library/?view_all_page_id=12345…) — we parse the Page ID out of the URL and pull every ad from that advertiser.
• A bare numeric Page ID (10+ digits) — same result.
• A domain or keyword — same search experience as elsewhere.

The meaningful detail is underneath: cursor-based Meta pagination, native to Meta's library. Large advertisers no longer hit a wall at an arbitrary result cap. Paste the Page ID of a brand running 400 ads and you get 400, not 50.

LinkedIn ads became a first-class source

LinkedIn Ad Library is now wired into the same search bar you use for Meta. Same filters, same card layout, same keyboard flow.

The inputs mirror Meta:

• Keyword search across LinkedIn's ad corpus.
• A LinkedIn Company page URL, or a raw LinkedIn Company ID — returns every ad that advertiser is running.
• Geo and date filters, wired in.
• A platform filter that routes exclusively to LinkedIn when you want to strip everything else out.

Two details land deeper than they first read. First, advertiser-name links on LinkedIn cards deep-link into that company's full ad history — one click to "show me everything this company is running", no second search. Second, impressions and days-running are surfaced on company-search cards, so you can sort the feed by how long each ad has been in-market without opening every detail panel.

The enrichment layer is lazy by design. Core results hit the screen the second they stream in. Landing page, domain, follower count — the slower enrichment calls — trickle in behind, and the detail panel lights up progressively rather than blocking on the slowest dependency.

Every card links out and downloads in one click

This was the smallest visual change with the largest effect on workflow.

Every ad card now carries a CTA strip under the creative, showing the destination domain and the advertiser's actual button text — "Shop Now", "Learn More", "Sign Up" — and the click goes straight to the landing page in a new tab. No mouse gymnastics to find the URL. No switching to the detail modal to read the CTA.

Next to it: a download button. One click pulls the video (or the current carousel image) down to your machine, in the resolution we captured it at. The same button is in the detail modal. You can finally build a swipe file of in-market creative without the right-click-save-as dance that every ad library ships with by default.

Works across Meta, LinkedIn, and federated sources.

Ad spend estimates, grounded in transparency data

Every ad with enough underlying signal now shows an estimated spend figure on the card and in the detail panel.

The model behind it is not made up. It's derived from EU political-ad transparency filings — actual spend against actual reach, with a correlation of r = −0.516 between log daily impressions and eCPM. From that, two CPM curves:

• By page follower count (when we have it, usually Meta): 10M+ followers → roughly $1.50 CPM. Scales up as followers scale down. Under 10K followers → roughly $10 CPM.
• By daily impression rate, as a fallback: 50K+/day → roughly $1.50 CPM. Small advertisers → roughly $10 CPM.

Whichever you're looking at, the estimate is explicitly an estimate. Directional, not audit-grade. But enough to sort a feed by approximate spend, or to tell whether a competitor is testing a creative or scaling one. You can cross-check specific numbers in our CPM calculator and ad spend estimator when you need to ground the logic.

The smaller fixes that matter

Five fixes that don't make good headlines, but removed friction you were feeling whether you knew it or not:

• Faster streaming results. The loading spinner now clears on the first streamed chunk instead of waiting for the full batch. Results feel immediate even when the query is wide.
• Credit ownership for team members. Credits on team plans now charge the owner, not the individual member — fixes a billing mismatch that had been silently annoying agencies with many seats.
• Auto-email flow for non-payers. Trigger wired up.
• Marketplace filter fix on federated search.
• Paywall A/B test closed. Soft paywall lost to hard paywall: 11 conversions to 3 at equal downstream retention. The A/B scaffolding is gone and hard paywall is the permanent default.

That last one deserves a note. Paywall UX is the kind of thing team members tend to have strong opinions about in advance, and then the data does what it does. In this case the data was decisive enough (more than 3× on conversions, no retention cost) that keeping the test running would have just been flattery of the losing variant.

What to actually try this week

Three things are worth the five minutes it takes to set up:

1. Paste a Meta Ad Library URL for a competitor you've been half-watching. With native pagination, you'll see the full tail of their testing program, not just the top 50 results.
2. Drop in three LinkedIn Company IDs for the accounts you watch in B2B — direct or indirect competitors — and set a date filter to the last 60 days. You'll see the cadence of what each is pushing right now, and the spend estimates give you a rough sense of budget priority. This fits naturally into a media buyer daily workflow or a competitor ad research loop.
3. Use the new download button to build a clean /ideas/ folder of real in-market creative rather than screenshots. Versions of saved ads help you share that inside a team, but the download button is what makes the folder usable offline.

If you're wiring any of this into automation — Claude Code, n8n, or your own pipeline — the API access feature covers the programmatic path, and Claude Code + adlibrary API workflows shows the agentic pattern we use internally for automating competitor ad monitoring.

FAQ

What's new in adlibrary for April 2026?

Native Meta Ad Library support (direct, not federated), first-class LinkedIn ad library search, outbound advertiser links and one-click media download on every card, ad spend estimates across all sources, faster streaming results, credit-ownership fix on team plans, and the retirement of the soft paywall A/B test.

Does the Meta integration replace the federated Meta results?

For any query that includes a Meta URL or Page ID, yes — it routes straight to the native client and paginates through Meta's own cursor. For keyword and domain queries, Meta still appears as one source among several via unified ad search; the dedicated Meta client handles advertiser-level pulls.

How accurate are the ad spend estimates?

Directional. The model uses CPM curves derived from EU political-ad transparency data — real spend/reach pairs. It is not a Meta reporting number; use it to sort and compare within the feed, not to audit a competitor's media invoice.

Can I pull LinkedIn ads for a specific company into a Google Sheet or Notion?

Yes — through API access, or by exporting saved ads from the LinkedIn tab. If you're automating with AI, ad data for AI agents covers the pattern.

Why did the soft paywall lose the A/B?

11 paid conversions for the hard paywall vs 3 for the soft, at identical downstream retention. Soft paywalls appear nicer and produce fewer paying users; retention was not the compensating signal it's often assumed to be.

Shipping continues. The next sprint leans into the enrichment layer — more sources on LinkedIn cards, more platforms on the native-client roster. Paste the URL of a competitor you care about and tell us what's missing — that's where most of this release came from.