Your competitors' advertising is not a secret. Every ad they run on Google, Facebook, Instagram, and LinkedIn is published in a public transparency tool, free to read. What their ads say, how long they have run, which formats they trust, where they point. All of it is sitting in the open.
What is scarce is not access. It is method. This is the full manual workflow we use, across all three platforms, and at the end, what a corpus of 55,085 tracked ads says about the patterns you should expect to find.
Step one: build the competitor list from the market, not from memory
Do not start from the companies you think about. Start from the companies a buyer actually sees. Search your main commercial terms the way a customer would and write down who appears, in the ads and in the organic results. Add the names your sales conversations keep surfacing. Ten to fifteen companies is a working set. The list you build this way always differs from the list in your head, and the difference is usually the point.
Step two: resolve every identity before you count anything
For each company, go to their own website and collect three things: the exact domain, the Facebook or Instagram page linked in their footer, and their LinkedIn company page. This step feels bureaucratic and prevents the two classic false readings: crediting a company with a lookalike page's ads, and declaring a company dark because you searched a name instead of the real identity. Every count that follows depends on this resolution being right.
Step three: the Google pass
In the Google Ads Transparency Center, find one confirmed ad per company, click through to the verified advertiser page, and save that URL. From the advertiser page record: total ads, format split between text, image, and video, regions, and the most recent last-shown date. Read the ten latest text ads and write down each distinct offer. The full Google method is here, including why a domain search alone will quietly lie to you.
Step four: the Meta pass
In the Meta Ad Library, pull each company's resolved page and record: number of active ads, the oldest start date, the format split between video, image, and carousel, and the landing pages the ads point to. Open the oldest ads in detail. Meta shows no performance data, so the start date is your performance proxy: advertisers keep what converts and kill what does not, which makes an ad running for months the closest thing to a public winner. The complete Meta method is here.
Step five: the LinkedIn pass
LinkedIn's transparency surface is smaller but real: on any company page, the posts section has an ads tab listing the ads the company is currently running. No spend, no performance, no history, but for B2B markets it answers the question that matters: are they paying to reach buyers here at all, and with what message? Record the count and the themes. In many markets this tab is empty for almost everyone, and knowing that is worth the five minutes it costs.
Step six: put it in one table and read the market
One row per company: Google ad count, Meta ad count, LinkedIn ad count, oldest running ad, format lean, and the offers, in their words. Now stop collecting and start reading, because the table answers questions no single ad can:
Who is committed? Not who has the most ads, but who has ads that have run the longest. Longevity is conviction backed by money.
What does the market agree on? When several advertisers converge on the same claim, that message is table stakes. You do not win by repeating it.
What is nobody saying? List the claims a customer would care about that appear in zero ads. That list is your opening, and it is usually longer than anyone expects.
Where is the platform gap? A market can be crowded on Google and empty on Meta, or the reverse. The empty platform is often the cheap one.
What 55,000+ tracked ads say you will find
We have run this workflow, plus the continuous tracking behind it, across 55,085 ads from 2,823 advertiser identities in markets from healthcare to industrial equipment. The same patterns keep repeating.
Most markets are darker than they look. In market after market, the loud brands and the advertising brands are different lists, and a large share of the companies everyone assumes are advertising turn out to run nothing at all. Assumed competition and paid competition are not the same thing, and the gap between them is usually the single most useful finding.
Ad volume concentrates. A small number of advertisers typically carry most of a market's ads, while the rest run a handful each. Reading the leaders closely tells you more than skimming everyone.
And the messages cluster hard. Within a niche, most ads make the same two or three claims in slightly different words. The unclaimed messages, the ones search demand exists for but no ad answers, show up in almost every market we read. Finding them is the entire commercial value of this exercise.
Assumed competition and paid competition are different lists. The gap between them is the finding.
The honest cost
Done properly, this workflow takes a few hours for the first pass on a ten-company set, and it decays fast: the transparency tools show today, not history, so a one-time pull is a photograph of a moving object. The teams that get real value repeat it monthly and keep the table alive. Most teams, reasonably, do not have those hours.
That is the work we run as a standing service. Request a free ad benchmark for your market: Muffin Intel tracks the ads, we read them, the written benchmark lands inside one business day.
Frequently asked questions
How many competitors should I analyze?
Ten to fifteen for a first pass. Fewer and you risk missing the advertisers a buyer actually sees; more and the collection work swamps the reading. Build the list from live search results and sales conversations, not from memory, and expect a few names on it to surprise you. After the first pass you can prune the confirmed non-advertisers to a quick quarterly check and read the active advertisers closely.
How often should I re-check the market?
Monthly is the useful cadence for active advertisers, quarterly for the quiet ones. The transparency tools show only what is running now, so anything you do not record is gone once the advertiser turns it off. New campaigns, new offers, and sudden silences are all signals, and all of them are only visible if you look on a schedule.
What is the fastest single signal to read?
The age of the oldest running ad. It costs one click per competitor and carries the most information per second of any public signal: an old ad is one the advertiser has kept paying for, which is the nearest public evidence of an ad that works. Volume tells you who is trying. Longevity tells you who is succeeding.
Do I need paid tools for competitor ad analysis?
Not for the core work. The Meta Ad Library, the Google Ads Transparency Center, and LinkedIn's ads tab are free and are the primary sources everything else builds on. Paid tools add modeled estimates and convenience, with trade-offs we cover in our review of the tool options. What the free tools cost is time, which is a real price. Pay it yourself, or have us pay it for you.
Free
See what your market is advertising
Written by Sarah Mitchell, Muffin Media
Want this run on your brief?
