Competitor ad research: how to learn from ad libraries before you build creative
Short answer: Competitor ad research means collecting the ads a rival is running now, taking each one apart, and writing a short brief on what to test for your client. The two free places to start are the Meta Ad Library (the public archive of ads run by Meta, the company behind Facebook and Instagram) and the Google Ads Transparency Center (Google's public list of advertisers and their ads). Neither shows what an ad earned, so treat every finding as a guess to test, not a result.
A folder of screenshots is not research. A one-page brief that says what to test first, and why, is.
I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.
The fastest useful version of this job takes half a day per client. Pick three to seven competitors, pull their live ads from the free libraries, save each one in a sheet, tag the hook, proof and offer, then write down the two or three things you will test. That is the whole method. The rest of this page is how to do each step well, what the libraries cannot tell you, and where tools such as Ploy, an AI platform that builds and runs marketing websites, fit into the work today and where they do not.
This page may contain affiliate links from which Blur Studio earns commissions.
This is information, not legal advice. Platform rules and layouts change, so as of November 2026 check each platform's current rules and screens before you rely on any detail below.
What the research is for
Write one decision at the top of your sheet before you open a single library. For example: "Which two hooks do we test first for this roofing client in January?" If you cannot name the decision, you are building a swipe file (a pile of saved ads), and swipe files grow and never get used.
Two rules keep it honest. First, you are looking for patterns and gaps, not ads to copy. Second, an ad that has been running a long time is a lead, not proof. The library does not show spend or results for ordinary ads, so a long-running ad might be a winner or might be a forgotten ad nobody switched off.
If you want the follow-on step of turning findings into rough visual sketches, see competitor creative analysis with AI sketches. This page covers the research and the brief that come before any sketch.
Where to look, and what each library does not show
| Library | What it shows | What it does not show |
|---|---|---|
| Meta Ad Library | Active ads across Meta products (Facebook, Instagram and others). Per Meta's research tools page, EU-delivered ads are kept for one year after their last impression | Spend, reach or results for ordinary commercial ads. Those extra fields (spend, reach, who paid) are for social issue, election and political ads, which Meta keeps for 7 years. Meta's page does not describe what is kept for ordinary ads that have stopped, so save your own copy |
| Google Ads Transparency Center | Advertiser name and location, the entity that pays for the ads, and the ads served over a time period on Google platforms such as Search and YouTube. You can search by advertiser or website and filter by date and location. Advertisers carry a "verified" or "unverified" label | Results and spend. Targeting details and recipient totals appear only in certain countries or regions, so do not assume you can see targeting |
Check two other places for the client's market: TikTok's Creative Center (a trend board of ads and creative ideas, not a full history of one advertiser) and, for B2B clients, whether LinkedIn (the professional social network) offers an ad library for your target country. Layouts differ by region, so open each one and read what is on screen for your client's country.
Two practical traps. Set the country filter first, because an ad running in another market is weak evidence for yours. And viewing public libraries by hand is normal practice, while automated scraping may break a platform's terms, so read the terms before you automate anything.
Pick the competitors
Use three tiers, five to seven names in total:
- Direct rivals. The businesses the client loses jobs to. Ask the client for the three names they hear most.
- Neighbours. Brands that sell to the same customer but a different thing (a kitchen designer next to a cabinet maker).
- One outsider. A brand from another category that advertises the same type of offer well, such as a free quote, a trial or a booking.
Fifteen to twenty ads per name is plenty for a first pass. In a thin market you may find two ads per rival. That is a finding too: it tells you the field is open and the client does not need to match anyone.
Capture sheet: save your own copy
Ads disappear. Screenshot or screen-record each one the day you see it and log it. Here is a sheet you can paste into a spreadsheet:
Competitor | Library | Country filter | Date captured | Start date shown |
Format (image, video, carousel) | Hook (headline or first 3 seconds) |
Promise | Proof (number, person, demo, review, guarantee) | Offer |
Call to action | Landing page URL | Landing page matches ad? (Y/N) |
Notes | Status: untested
Capture on a schedule, every one or two weeks, because a single capture misses what changes. Keep the sheet in the client's shared folder so the next person does not repeat the work. If you manage several clients, a shared template keeps the columns the same, which makes patterns across accounts easy to see later.
Take each ad apart
For every ad on your shortlist, fill in five things in plain words: hook (the first line or first three seconds), promise, proof, offer and call to action (the button or next step the ad asks for). Add one more that most guides skip, the landing page, which is the web page a visitor sees after clicking the ad. Click through and check whether the page repeats the ad's promise in the first screen. An ad that promises "same-day quotes" and lands on a generic home page is a gap you can use. For how to match page to ad, see ad landing pages that match the ad.
Then sort the ads into clusters. You are looking for lines such as "five of seven lead with a price" or "nobody shows the finished job". The first line tells you what is crowded. The second tells you where your client can look different.
Treat longevity as a weak hint
Many guides say an ad running for 30 days or more is worth a closer look. That is a rule of thumb, not a fact from the platforms. I use it only to build the shortlist and I write "untested" next to every pattern until it has run in the client's own account.
From finding to concept: the one-page brief
Use one sentence format for each idea: "For [audience], test [hook] in [format] because [pattern] may improve [measure]."
Worked example (made-up numbers and a made-up client): a local roofing company. You audited six rivals and logged 84 ads. Findings: 51 of 84 lead with a discount, 9 show a finished job, none shows the crew arriving on time, and 6 of 6 landing pages open with a form and no photos. Your two hypotheses:
- For homeowners with storm damage, test a calm 20-second video of a real finished job, because only 9 of 84 competitor ads show one, and it may improve click-through rate (the share of people who see the ad and click it).
- For the same audience, test a headline promising a fixed arrival window, because none of the rivals mentions timing, and it may improve cost per lead (ad spend divided by the number of enquiries).
Label the numbers as yours to replace. The point is the shape: count what you saw, name the gap, write the test. Every hypothesis is a guess until the client's own data says otherwise.
Hand the client one page: the decision, three to five patterns (mark the crowded ones), one or two gaps, the two tests to run first, the measure for each, and the pass line. Define the pass line from the client's own numbers, for example "keep it if cost per lead is below the account's current average after the agreed spend". Competitor data cannot tell you what a pass looks like.
Where tools fit, including Ploy
The free libraries plus a weekly manual check are enough for most small clients. Paid ad-intelligence tools add search across platforms, alerts and long history, and they start to make sense when you track many competitors across several platforms for several clients. Price and features vary, so check each vendor's page before you commit.
On Ploy, here is what exists and what does not, as of November 2026. Ploy's Web product (the AI site builder, analytics and forms) is documented in detail. Ploy's homepage lists an ad library, creative at scale and Meta campaign management as part of an "Ads" area, but the PloyAds page is a waitlist with the words "coming soon" and no launch date, so I would not plan client work around those features. Ploy's docs do list a Meta Ads connection on every plan, including Free (its docs were updated October 30, 2026; the pricing page (read November 21, 2026) shows ad integrations as not included on Free or Starter, so open Integrations on a Free workspace and see whether the Meta Ads card says Connect or Upgrade), and say it can create campaigns, ads and creatives when asked. Treat that as a connection to test on a client account with approvals, not as a research tool. For the longer picture, see Ploy's ad tools explained.
Where Ploy genuinely fits today is the step after the research: building the landing page that matches the ad you decided to test, quickly, and measuring what visitors do with its built-in analytics. If you want to try it, you can start with Ploy's free plan (500 credits a day, up to 2,000, no card needed per Ploy's pricing page) on a throwaway test page. For plans and limits see Ploy pricing explained.
Rules against copying
Structure is reusable. Footage, scripts, characters, music and a distinctive look belong to someone else. Give your team this list:
- Never paste a competitor's ad, logo, script or named person into an image or video tool.
- Never prompt for "in the style of [competitor]" or use their product photos as a reference.
- Do not reuse their slogan, music, voice or recognisable layout.
- Rebuild the structure (hook, then proof, then offer) in your own words and visuals.
- Do not reuse a claim you cannot support. A rival running "number one in town" does not make it lawful for your client.
Run a copy test before anything ships: cover the logos and show the draft to someone who has not seen the research. If they name the competitor within three seconds, change the structure, not only the colours. For claims about results, health or money, check local law and your country's advertising regulator. This is not legal advice.
What this research cannot tell you
- Whether any competitor ad makes money. You see the ad, not the results.
- Targeting for ordinary ads. Libraries do not show it.
- What the client's own audience will do. Only a test answers that.
- Why an ad stopped. It could have failed, finished or been replaced.
Pair the research with the client's own data: top-performing past ads, sales team notes, and what customers actually say on calls. Those three beat any library.
Who should skip this
- A client in a market with almost no advertisers. The libraries will be empty, and talking to five customers will teach more.
- A team with no way to test. If you cannot run one change at a time and compare results, the brief will not become learning.
- A brand whose edge is a real person or a real place. Look for message gaps, then shoot the real thing.
- Readers hoping Ploy will do the research for them. Its ad library is only a homepage bullet and a waitlist today, so use the free libraries.
Copy-ready checklist
- Write the decision the research must answer.
- Pick three to seven competitors in three tiers.
- Set the country filter, then capture 15 to 20 ads each into the sheet.
- Tag hook, promise, proof, offer, call to action and landing page match.
- Cluster: list what is crowded and what is missing.
- Write two to three hypotheses in the one-sentence format.
- Run the copy test and the claim check on every draft.
- Set the measure and pass line from the client's own numbers.
- Test, then replace "untested" in the sheet with real results.
How I'd approach this
I would treat the first client as a pilot. Day one: write the decision, pick the competitors, capture the ads. Day two: tag and cluster, write the brief, run the copy and claim checks. Day three: build the two landing pages that match the two chosen ads (this is where Ploy's AI site builder can save time, and where a template in any page builder works just as well), set up tracking, and launch both tests with a fixed budget and an end date. Week two: read results against the pass line and update the sheet.
Pick a paid ad-intelligence tool only after you have done this by hand for three clients and know which columns you actually use. If you want a custom version, with AI that reads your saved ads and drafts the brief for your team to approve, Blur Studio builds that. See what a custom build includes.
Related guide: Meta campaign management with AI.
Questions people ask
What is the Meta Ad Library?
It is Meta's public archive of ads running across its products, including Facebook and Instagram. You can search by advertiser or keyword. Extra details such as spend and reach are shown mainly for social issue, election and political ads. Check Meta's current pages for what you can see in your country.
Does the Meta Ad Library show how much a competitor spends?
Not for ordinary commercial ads. According to Meta's research tools page, spend, reach and funding details are part of the extra information for social issue, election and political ads. Treat any spend figure for an ordinary ad as an estimate from a third party, not Meta data.
What does the Google Ads Transparency Center show?
It lists advertisers and the ads they have served over a time period on Google platforms such as Search and YouTube, with the advertiser's name, location and who pays. You can filter by date and location. Targeting details appear only in certain countries or regions. See Google's help page.
Is it legal to learn from a competitor's ads?
Looking at public ads and learning from their structure is normal practice. Copying footage, scripts, characters, music or a distinctive look is not. Reuse structure, build your own expression, and check local law for edge cases. This is information, not legal advice.
How many competitor ads do I need to review?
Fifteen to twenty per competitor across three to seven competitors is enough for a first pass. In a small market you may find far fewer, which is itself useful. Quality of notes matters more than volume.
Does a long-running ad mean it is working?
Not necessarily. Many guides use 30 days or more as a hint, but it is a rule of thumb, not a platform statement. An ad could be running because it works or because nobody switched it off. Write "untested" until you have tested the idea in the client's account.
Can Ploy research competitor ads for me?
Not today as a live product. As of November 2026 Ploy's PloyAds page is a waitlist with no launch date. Ploy's docs list a Meta Ads connection, but use the free libraries for research and consider Ploy only for the landing pages you build afterwards.
Should I pay for an ad intelligence tool?
Only if the free libraries and a weekly manual check stop being enough, for example when you track many competitors across several platforms for several clients. Do the work by hand for three clients first, so you know which fields you actually use. Check each vendor's own page for current prices.
More in this topic: Websites: client reporting, leads and ads for agencies (8 guides) →