How to find your real ad competitors

The competitor folder has five names in it. Three came out of a sales meeting, one from a founder who saw something on LinkedIn, and the fifth has not run an ad since March. Every guide on competitor ad research picks up one step after that folder exists: type the brand into the Meta Ad Library, take screenshots. Nobody asks where the five names came from.

That step decides everything downstream. Research the wrong five and you get a very thorough read of a market you are not competing in.

Short answer: Your real ad competitors are whoever bids for the same attention, which is rarely the brand list your sales team keeps. Build it from evidence: pull names out of your own search terms and auction data first, then search the ad libraries by keyword, offer and category instead of by brand name.

The takeaways

• Sales-sourced names come from the bottom of the funnel, where a prospect compares two vendors by name. The auction happens much earlier, against whoever buys the same moment of attention.
• Your own account holds the only hard evidence: the search terms report, plus Google's Auction Insights, which names the domains that appeared in the same auctions and how often you overlapped.
• Prune on activity rather than reputation. An advertiser whose newest ad stopped in March is history, and I keep 30+ days still running as the line for calling anything durable.

Why is the list from your sales team the wrong list?

Because it answers a different question, and answers it well. Sales-sourced competitors are the names a prospect says out loud on a call, after deciding to buy something in this category from somebody. That is bottom-of-funnel information.

The auction sits far upstream of that call. You are paying for a scroll or a query from someone who has decided nothing yet, and the other bidders for that moment include names no one put in the deck: an adjacent product solving the same problem from a different angle, a DTC brand six months old, an affiliate running a comparison page on your terms. None of them appear in a lost-deal report. They never reached the deal stage. They only made your CPM more expensive.

I have run watchlists where two of five names were dead weight and the two that mattered had never come up in a meeting.

Where do you find advertisers you did not know about?

Start inside your own account. It is the only place holding evidence rather than opinion.

The search terms report shows which queries your ads appeared on, including plenty you never targeted. Read it for brand names you do not recognise: a stranger showing up there means the market treats you as substitutes. Google's Auction Insights goes further and names the domains that showed alongside you in the same auctions, with an overlap rate for how often you both appeared. Closest thing to a receipt this job offers.

Then work outward. Run your 10 highest-intent queries in a clean browser and write down every paid result. Ask 3 recent customers who else they looked at, and listen for the surprising answers.

One warning: none of this exists on Meta. No auction report, no domain list, so the Meta half of your list stays inference.

How do you search an ad library without a brand name?

Search by what you sell, not by who sells it.

Brand lookup can only confirm a name you already have, which is why a list built that way never grows. The 3 public libraries accept more than a company name. Meta's Ad Library searches free text, so feed it your offer language, your category, the phrases customers use, then read the advertiser column first and the creatives second. TikTok's library filters by country and industry, which catches someone selling into your category with different vocabulary. The Google Ads Transparency Center is organised by advertiser and domain, so pair it with that manual query search.

This is where the unknowns turn up: names nobody in the meeting has heard of, running your promise at a volume that explains a CPM you could not account for. Adscalr pulls all three libraries daily into one normalized dataset, largely so this stops being three tabs and a spreadsheet.

Which names belong on the list, and which get cut?

Presence is not competition. The list gets shorter before it gets useful.

Cut anyone whose newest ad stopped months ago. A dormant advertiser tells you what the market looked like last spring and nothing about the auction you are in this week. Cut the large brand running three broad awareness spots on a national budget too: they buy reach on a different clock and will not react to anything you do.

What survives is whoever runs volume and keeps running it. I sort the remainder by two numbers: how many distinct ads are live, and how long the oldest survivors have been up. Take two names that both look busy. One has 4 ads live, all launched in the past week. The other has 18 live and its oldest has been up for 71 days. The second is running a program, the first might be drifting. An ad still up after 30 days has cleared a month of its owner's own kill decisions, the only free durability signal a public library hands you, and the same 30-day line Adscalr uses to flag a winner.

What can a watchlist never tell you?

Whether any of them are making money. Public libraries show what runs and for how long. No spend, no ROAS, no way to tell a profitable advertiser from a funded one with patience. A long-running ad proves somebody kept paying for it, which is worth studying and is not proof of anything else.

The second gap is structural. Meta gives you no way to learn who you were bidding against, so your Google half rests on auction evidence and your Meta half rests on inference. Worth remembering which half is which when the positioning argument starts.

A current, active list buys you one thing: a map. 20 live advertisers can be read as positions on 5 messaging axes, price, quality, speed, trust and innovation, and the emptiest of the 5 is your open lane. That clustering is what the competitor intelligence side of Adscalr does with the daily library pull.

Once the names are right, the three-library research workflow tells you what to do with them, and the gap map turns the pile into a positioning decision.