What ads are YC companies running in 2026? Our complete, systematic study
We checked every Y Combinator company against the Google and Meta ad libraries. The companies with AI in their tagline advertise less than almost anyone else, only a third of the portfolio advertises at all, and it is Google they reach for first.
We took all 6,055 companies in the Y Combinator directory and looked each one up in the public ad libraries that Google and Meta are legally required to keep. Then we pulled the ads themselves, 121,841 of them, to see what YC companies actually run.
The first thing we expected to find was that AI companies advertise the most. They are the batch darlings, the ones raising the most and hiring the fastest. They advertise the least.
The short version
- Only 34.9% of live YC companies run paid ads. Across the full directory, dead and dormant included, it is 27.5%.
- AI companies advertise least (29.7%), below SaaS (47.3%) and E-commerce (58%). Only 8.4% of ad copy even mentions AI.
- Google is the default, Meta is a second step. 33.7% run Google, 8.3% run Meta, and just 1.3% run Meta without Google.
- Headcount predicts advertising better than anything: 16.7% at 1-5 people, 86.5% at 500+.
- Industrials barely advertise (13.3%); Education leads (49.5%).
AI companies are among the least likely in YC to advertise
Companies tagged "Artificial Intelligence" run paid ads at 29.7%, below the 34.9% portfolio average and well under SaaS at 47.3% or E-commerce at 58%. Of the 819 AI companies in the directory, fewer than a third are buying ads.
Ad adoption by company tag. The three AI-related tags (magenta) all fall under the 34.9% portfolio average.
It shows up in the copy too. Across the 73,653 ads with readable text, 8.4% mention AI (we counted any of "AI", "artificial intelligence", "LLM", "GPT", or "machine learning" as a whole word). More than nine in ten YC ads never say it.
The companies leaning hardest on ads are the unglamorous ones: e-commerce, marketplaces, payments, digital health. Businesses with a transaction to drive and a customer who is already shopping. If your growth comes from a sales team, or from being the obvious pick in a new category, paid acquisition is not where your money goes yet, and for a lot of AI companies that is where they are.
Only about a third advertise, and they start with Google
Of the 4,767 YC companies with a live website that are still operating, acquired, or public, 34.9% are currently running paid ads. Across all 6,055 companies in the directory, including the dead and the dormant, it is 27.5%. The rest we found running none.
Both ad libraries only show what is live right now, so a company that advertised last year and paused reads here as a non-advertiser. Read these as "advertising today," not "has ever advertised."
When YC companies do advertise, they go to Google:
- 33.7% run Google ads
- 8.3% run Meta ads (Facebook and Instagram)
- Only 1.3% run Meta without also running Google
Share of live YC companies advertising on each platform.
That last number is the surprising one. For YC companies, Meta is almost never a starting point. It is something a company adds on top of Google, not instead of it. One reading is that Google captures demand that already exists, someone searching for what you sell, while Meta has to create it, and companies selling software to other businesses reach for the former first. We can only see who runs what today, not the order they got there, so treat that as interpretation.
The best predictor of advertising is headcount
Nothing in the data predicts advertising as cleanly as company size.
Share of live YC companies running paid ads, by employee count.
A three-person company almost never buys ads. A hundred-person company almost always does. Paid acquisition is not what YC companies do early to find product-market fit; it is what they turn on once they have a product and revenue to protect. Growth-stage companies advertise at 63.3%, against 27.8% for early-stage.
If you run a small YC company and you are not advertising, you are with the large majority. And that is the opportunity. Paid acquisition is wide open at your size: it is not where your similarly-sized competitors are fighting yet, so the space to win cheap attention is still there for the taking.
By sector, a four-to-one spread
Education leads; hardware-heavy Industrials barely advertise.
Sell hardware and you mostly are not buying ads: hard-tech, robotics and hardware tags all sit under 20%. The platform split by sector is stark too. Meta is a consumer channel here, Consumer companies run Meta at 17.9% and Education at 25.3%, against B2B's 5.6%. B2B lives on Google.
Adoption by batch is roughly flat from 2011 to 2021, then declines for the newest batches: 57% for 2016, 43% for 2021, 26% for 2024, 12% for 2026. Part of that is simply age, since a 2026 company may not have started advertising yet, and part is that recent batches skew toward very small, very early AI companies. We can't cleanly separate the two effects with a single snapshot.
Ad adoption by YC batch year. Flat through the 2010s, then a steep drop after 2021.
What the biggest names run
We pulled the top 100 YC companies by funding (the ycdb.co ranking: Airbnb, Stripe, DoorDash, Instacart, Coinbase, Reddit and so on) and checked them against the same libraries. They behave nothing like the portfolio.
66% of the top 100 are advertising right now, nearly double the 34.9% across all of YC. And they run at a completely different scale: DoorDash has 3,300 live ads, Instacart 1,799, Gusto 1,107, Weave 955, Webflow 935, Stripe 900, Eight Sleep 636. Where a median YC advertiser runs a few dozen ads, these run hundreds to thousands.
They are also on both platforms. Of the top companies that advertise, 26 run Google and Meta together, versus a portfolio where running both at all is rare. Paid acquisition at this level is not a single experiment; it is a standing, multi-channel operation with a budget and an owner.
And the most telling detail is how long the ads have been live. Some of DoorDash's and 1000Memories' search ads have been running continuously since 2021, years of the same creative left on. Nobody keeps paying to serve an ad for four years unless it is quietly making money. A long-running ad is the closest thing the ad libraries give you to a profit signal, and the biggest names are full of them. They are also twice as likely as the rest of YC to send that traffic to a purpose-built campaign page rather than the homepage.
Put together, it reads as a revealed preference. The companies with the most funding, the best analytics teams and the most scrutiny on every dollar treat paid ads as a permanent, growing line item, on both platforms, for years at a time. That is about as clear a signal as this data gives that paid acquisition is a profitable, durable growth channel. The winners are not asking whether ads work. They are compounding on the fact that they do, and the interesting question for everyone earlier on the curve is not whether to be in the channel, but how much sooner to start.
What the ads look like
We pulled every live ad from all 390 Facebook/Instagram advertisers, and a 55% sample of Google advertisers, 121,841 ads in total. The creative findings below are census-complete for Facebook and Instagram; the Google creative mix is from that sample.
Facebook is video. Google is text.
On Facebook and Instagram, the largest format is video at 35%, then dynamic catalog ads at 33% and static images at 25%. On Google, 62% are text search ads, which is close to true by definition, plus 24% display images and 14% video.
Worth one note on method. Our early 150-company pilot showed static images as the leading Facebook format. At full scale that reversed and video came out on top. The pilot was too small to see it, which is why we ran the whole portfolio.
The copy is short and literal
The median YC ad headline is 5 words; body copy runs 15. The most common opening words are the plain ones: "get", "start", "try", "free".
Buttons are narrow too. "Learn more" is 33% of all call-to-action buttons, ahead of "Install now" (16%), "Shop now" (11%) and "Sign up" (11%). On Meta the button is a preset dropdown rather than free text, and "Learn more" is often the deliberate top-of-funnel choice, so read this as convention, not as evidence anyone is or isn't testing.
Real headlines, verbatim. Notice how many open with a plain command verb or a question, and how many just state the offer.
One curiosity for the data nerds. The most common opening string across all 121,841 ads is not a word, it is {{product.name}}, in 10,185 ads. This is not thousands of broken live ads. It is how the Meta Ad Library displays dynamic catalog ads to a logged-out viewer: the advertiser's product feed fills that token in at delivery, but the public library has no feed to draw from, so it shows the raw template. Ninety-one percent of these are dynamic-format ads, which is the tell. Real customers saw a product name. It is an artifact of how the library stores catalog ads, not a mistake anyone is making.
Where the click goes
Of the 75,445 ads where we could see the destination, 41.9% point at the company homepage and 34.7% at a specific product page. Only 3.6% go to a purpose-built campaign landing page.
For cold prospecting traffic, especially on paid social, a homepage is a weak destination: the visitor arrives with the specific question the ad just raised and lands on the page built to answer everyone's. For a single-product company running brand-term search, the homepage often is the right page, so this is not a universal mistake. But the share pointing at a generic homepage rather than a page matched to the ad is high enough to be worth a look if it describes your account.
Purpose-built campaign pages (the ones doing it right)
Article and advertorial pages (heavy in fintech)
Straight to a demo or trial
The tracking behind the ads
A quick caveat first, because the honest answer matters here. The public ad libraries are not built to expose tracking. Google's Transparency Center usually shows the cleaned final URL, and even Meta only surfaces the click URL, so query strings and UTM tags are stripped or missing on most records (only 11% of the Google landing URLs we captured have a query string at all). So we cannot say how many YC companies tag their traffic. But on the ads where the full tracked URL does come through, the conventions are worth a look.
The plumbing is standard: utm_source is usually google, facebook or meta; utm_medium is cpc, paid_social or paid. The interesting part is utm_campaign, where two completely different schools show up.
The first is the enterprise taxonomy: rigid, delimited, machine-parseable, encoding channel, region, segment and creative in one string.
The second is the human-readable, fiscal-year style, common among the growth-stage SaaS companies:
And then, inevitably, the same dynamic-template leak we saw in the ad copy shows up in the tracking. A slice of ads ship UTMs with the tokens never filled in, so the analytics receive the literal placeholder instead of the value:
If nothing else in this piece is actionable, this is: pick one consistent campaign-naming convention, and check that your templates actually render before the ad goes live.
Two smaller things
Ad records outlive companies. For 8.6% of the inactive-but-still-online companies (61 of 706), a live ad record still sits in the transparency indexes. We can't see whether money is still being spent, only that the record is there.
The ones that made it keep spending. Of the 21 YC companies that went public, 14 still run ads, 10 of them on Meta.
1,000 real YC ads for your inspiration
Enough numbers. Here are 1,000 actual creatives, pulled straight from the Meta and Google ad libraries, run by 722 different YC companies. Product shots, founder photos, text-on-color, search units, the lot. Filter by platform or industry inside.
Creatives are shown as served in the public Meta and Google ad libraries, credited to the advertiser. A representative sample, not the full set.
1,000 real YC ads for your inspiration
Method and data
We joined the Y Combinator company directory (6,055 companies) against two public, legally mandated ad databases: the Google Ads Transparency Center and the Meta Ad Library, which covers Facebook and Instagram.
How a company counts as advertising. For Google, the company's own web domain appears in the Transparency Center's index; domains are unique, so this match is exact. For Meta, we found a Facebook page that is provably theirs (either their website links to it, or its ads point back at their domain) and confirmed the page is running live ads.
What these numbers are not. Both libraries show only currently-live ads, so every figure is "advertising now," not "ever advertised," and all of them are a floor. The Google creative-mix figures come from a 55% sample of Google advertisers; the Meta figures are a full census of all 390. We excluded LinkedIn, whose public library only lists recruitment ads. And this measures who advertises and what they run, never how much they spend, which no public source discloses.
How we pulled the ads. The ad data comes from Apify scrapers run against the public ad libraries: the Meta Ad Library page resolver and Meta Ad Library ads scraper for Facebook and Instagram, and the Google Ads scraper for the Transparency Center.
The data is open. Every number here traces to two files: companies.csv, a per-company table of who advertises on which platform, and aggregates.csv, every summary statistic in this article.
Built by Adside. We help companies run better paid ads. If you recognized your own account in the homepage-destination number, we can help with that.