Can You Estimate a Competitor's Ad Spend?
What the Public Data Supports, and Where the Numbers Come From
Short answer
Not accurately. Meta and Google publish spend only for political and social issue advertising, so commercial spend figures in any tool are estimates built on assumed impressions and assumed click prices, where small input errors compound. Use observable signals instead: active ad count, how long ads run, and creative volume over time.
1. What do Meta and Google actually publish about advertising?
They publish what is running and for how long, for free and without an account. They publish spend only for political, social issue and election advertising, and they never publish results for commercial advertisers.
| Data point | Meta Ad Library | Google Ads Transparency Center |
|---|---|---|
| Creative and format | Published for every active ad | Published for advertisers in the repository |
| When the ad ran | Start date on every active ad | Searchable and filterable by date range |
| Placements | Facebook, Instagram, Messenger, Audience Network | Search, Display, YouTube and Gmail surfaces |
| Advertiser identity | Page name and, where applicable, the paying entity | Verified legal or trademark name, plus the paying profile |
| Spend | Political and social issue advertising only | Election advertising only |
| Reach or impressions | Political advertising, plus extra disclosure on EU-served ads | Election advertising, plus EU disclosure |
| Targeting and audiences | Not published for commercial advertising | Not published, and keywords and bids never are |
| Results: clicks, conversions, ROAS | Never published | Never published |
2. How do ad spy tools calculate competitor spend?
They estimate impressions, then multiply by an assumed price per thousand impressions or per click. Impressions come from panels, sampling or crawl frequency, and prices come from category averages.
Neither input is specific to the advertiser being measured. Auction prices depend on targeting, creative quality, competitive density and time of year, so a category average is the wrong price for any particular account by construction rather than by accident.
The impression side is weaker still. A tool that infers impressions from how often its own crawlers saw an ad is measuring its own sampling behaviour as much as the advertiser distribution, and no public data exists to calibrate the result against.
None of this makes the tools useless. It makes their output a modelled index rather than a currency figure, and the mistake is in how the number gets used rather than in the modelling itself.
3. How wrong can a competitor spend estimate be?
Wrong by a factor of two in either direction, because two uncertain inputs are multiplied. An impression estimate off by forty percent combined with a price assumption off by thirty percent produces exactly that.
The compounding is the part that gets missed. People discount a single estimate mentally, allowing perhaps twenty percent of slack, and then multiply two of them without widening the allowance. The resulting figure is still presented as one number with no interval attached.
This is also why two tools routinely disagree about the same advertiser by a large multiple. When the underlying quantity is not published, disagreement between models is the expected outcome rather than evidence that one of them is broken.
4. What can I measure instead of spend?
Three things, all published for every advertiser: active ad count, how long each ad keeps running, and how many new creative variants appear per month. None of them requires an estimate.
This takes about ten minutes per competitor and produces numbers you can defend in any room.
- Search the advertiser per country. Open the relevant ad library, set the country, and search by Page name or domain rather than free text. The same advertiser often runs entirely different creative per market, so one country is one measurement.
- Count active ads and stamp the count with a date. Twenty active ads against three is a real difference in programme scale. A move from three to twenty inside a fortnight is a real change in commitment, and neither observation needs any assumption about price.
- Note the oldest ad still running. Duration is the best available proxy for performance, because nobody keeps paying to run creative that loses money. An ad live for four months is a tested winner. One that vanished after four days probably was not.
- Count new creative variants since your last check. A competitor producing new variants weekly has a team and a testing process behind the budget. One running the same two images for six months does not, and that gap tells you more about operational capacity than a spend figure would.
- Compare all four numbers against your own programme. Put their count, duration, variant rate and placement breadth next to yours. The answer you get is relative rather than absolute, which is what almost every decision behind the spend question actually needs.Record your own numbers on the same day. A comparison against a month-old snapshot of your own account invents a difference.
5. What question are you really asking?
Almost nobody wants the spend figure for itself. The real question is whether a competitor is outspending you, whether paid is working for them, or whether something has changed, and all three are answerable from published data.
| What you actually want to know | What to measure | Where it comes from |
|---|---|---|
| Are they outspending us? | Active ad count, placement breadth and variants per month, versus your own | Both ad libraries, same day as your own numbers |
| Is paid working for them? | How long individual ads keep running, and whether the programme is continuous or bursty | Ad start dates and date ranges |
| Are they about to push harder? | Week-on-week change in active ad count, and the arrival of an unfamiliar angle | Weekly snapshots of the same library search |
| Which message do they trust most? | The angle that appears most often and has run longest | The creative itself |
| Where should we avoid competing? | Markets and placements where they run heavy volume continuously | Country and platform filters |
| Should we be worried? | A step change in count, format or market breadth, not any absolute level | The comparison against your own previous snapshot |
6. When is a spend estimate safe to use?
For ranking, never for planning. Saying one competitor appears to spend materially more than another survives the method error. Putting an estimated monthly figure into a budget case does not.
Label the source and the method wherever the number appears. An unlabelled estimate becomes a fact within two slide revisions, and then somebody plans against it, and the person who introduced it is usually no longer in the room to explain the caveat.
A good habit is to write the estimate as a range with the method attached in the same sentence, for example "modelled at roughly two to four times our spend, based on ad count and category average prices". That phrasing is harder to strip out than a footnote.
Oppira tracks the Meta and Google ad libraries daily for the competitors you follow, recording active ad counts, run duration and creative changes over time. It does not invent a spend figure, because the underlying data does not support one.
Key Takeaways
Commercial ad spend is not published
Meta and Google disclose spend only for political and social issue advertising. Everything else is inference.
Estimates multiply two uncertain inputs
Assumed impressions times assumed prices, where moderate errors in each can leave the result wrong by a factor of two.
The estimate is weakest for small advertisers
Narrow audiences and thin panel coverage widen the error exactly where most SMB competitors sit.
Duration is the best performance proxy
Advertisers do not keep paying for creative that loses. An ad running for months is the closest public signal of a winner.
Count and duration are free and exact
Active ad count and run length are published for every advertiser, need no model, and are directly comparable over time.
Rank with estimates, never plan with them
A large gap between two competitors survives the method error. A monthly figure inside a budget case does not.
Frequently Asked Questions
Sources
- Meta Ad Library Meta Platforms, July 2026.The primary source for what Meta publishes about active advertising, and for the political-only spend disclosure.
- Google Ads Transparency Center Google, July 2026.The primary source for Google ad disclosure, searchable by advertiser with region and date filters.
- Regulation (EU) 2022/2065, Digital Services Act, Article 39 European Union, October 19, 2022.The advertisement repository obligation behind the extra reach disclosure on ads served in the EU.
Explore More
Related analyses, benchmarks, and industry insights
Related Guides
Glossary Terms
Turn reading into a reaction
Oppira watches your competitors, keeps your playbook current, and drafts the response. Start free and see your market clearly by tomorrow.