Companies Using Heroku

We detect Heroku on 675 domains, including Harvard Business School Publishing Corporation, Northwestern University, Honda Motor Co Ltd. A detection means the technology is present on that domain, which for large organisations is often a single storefront or subdomain rather than a company-wide deployment. Among domains we matched to a company with a resolved industry, the most common is Software Development (10.84%).

We detect 675 domains running Heroku across our monthly crawl of live business domains. Company data could be matched for 476 of them (70.5%), and every company-size, industry and country breakdown on this page describes that matched subset rather than the full detected base. A further 4 domains carried a Heroku detection previously and no longer do.

Heroku holds 0.03% of the PaaS category, ranking #23. This is a share of the domains we detect running a platform in this category — not a share of all websites, and not a share of spending — so it is not directly comparable to figures published on other denominators.

The top alternatives to Heroku include Amazon Web Services (39.98% market share), Vercel (16.64% market share), WordPress.com (8.24% market share), Azure (7.5% market share) — measured as a share of the domains we detect running a platform in this category.

United States leads with 272 matched Heroku domains, followed by United Kingdom (34), Canada (26), Australia (20), France (11). These are counts within the 476 domains we could match to a company record, not the full detected base.

The most common company size is 1-10 employees, representing 55.67% of the Heroku domains we could match to a company record and size — a subset of 476 matched companies, not of all 675 detected domains. This is followed by 11-50 employees (25.48%) and 51-200 employees (9.21%).

Among the Heroku domains we could match to a company with a known founding date, the largest group was founded in the 2010s (63.66%), followed by the 2000s (11.34%). That matched subset is drawn from 476 companies and is not representative of all 675 detected domains. This suggests Heroku is most popular among relatively young companies.

The ideal Heroku customer is: Company Size: 1-10 employees, Location: United States, Founded: 2010-2019, Tech Stack: Cloudflare, Google Analytics, Rails, Typical Deployment: One app, often on herokuapp.com — based on our analysis of 476 enriched companies at TechnologyChecker.io.

Heroku is identified from three narrow signals: a Via response header naming vegur, which is Heroku's HTTP router, a .herokuapp.com URL, and a DNS TXT record. A detection means one of those was visible on that domain at crawl time. Nothing about running a Heroku app forces any of them to be visible from outside, and a CDN in front of the app can remove the header entirely, so 675 domains is a floor for production apps on custom domains rather than a count of Heroku accounts.

Count every domain where a Heroku detection currently stands and the total is 1,290. Of those, 617, about 48%, are *.herokuapp.com subdomains. That is the default hostname every Heroku app receives before anyone attaches a custom domain, so a large part of that set is demos, staging environments, internal tools and hobby projects rather than a business running a website. It is the most important thing to know before quoting any Heroku usage figure, including ours.

We cannot tell you, and neither can anyone else working from a three month series. Our current window has three points: 606 active domains in June 2026, 660 in July and 675 in August. Three readings on a base this small cannot separate real adoption from ordinary turnover in the set of domains crawled. Presenting that slope as growth would be reading noise, and so would presenting it as decline. Ask again when there are more points.

Our data supports no claim in either direction. Only four domains have dropped their Heroku detection across the entire tracking window, which is far too few to describe a direction of travel, and it is why we publish no switching corridors on this page. That is a statement about what we can currently measure, not evidence that migration is not happening.

Two reasons. The detection rule here is narrow, so any Heroku app that does not expose the vegur header, a herokuapp.com URL or the DNS record is invisible to it. And this is a count of domains, not accounts, apps or dynos. One Heroku account can run many applications, and most applications never get a domain of their own, so they never appear in a crawl of live business domains. Figures published elsewhere often count apps or accounts, which is not the same measurement.

Because each chart drops the rows it cannot resolve. We detect Heroku on 675 domains. 476 of those matched a company record, which is the base for the country chart. 467 of those had an employee band, and only 344 had a known founding year, which is the base for the age chart. Every percentage is of the base named on that chart, never of 675, and the smaller the base the more cautiously it should be read.

Sharply. Of the 344 Heroku companies with a known founding year, 63.66% were founded in the 2010s and only 10.47% in the 2020s. In the same category, Supabase draws 39.90% of its matched companies from the 2020s and Render 38.05%. Heroku is running the applications of companies that started during its own peak years. It is not appearing under the companies starting now, and that gap is the clearest signal in this dataset.

Because Heroku's original audience never left. Ruby on Rails is on 13.95% of sampled Heroku domains, the highest share of any framework in that stack and well above React at 7.6% or Vue.js at 4.88%. Heroku launched as a Ruby platform, and more than a decade later the Ruby cohort is still the most distinctive thing about who runs on it.

It runs, but it is not being developed. In February 2026 Salesforce announced that Heroku is transitioning to a sustaining engineering model focused on stability, security, reliability and support, with no new Enterprise Account contracts, while stating there is no change for customers using Heroku today (per heroku.com/blog/an-update-on-heroku). Our own detection data cannot confirm or contradict that, since three monthly readings on 675 domains say nothing about a platform's direction.

No. Heroku is owned by Salesforce, which acquired it in 2010. The confusion comes from the fact that Heroku's infrastructure runs on Amazon Web Services, but Amazon has never had an ownership stake in the platform.

Heroku is a PaaS. It abstracts the infrastructure away, so a developer deploys code without configuring servers, load balancers or operating systems. AWS is an IaaS, which hands you those lower level pieces to assemble yourself. Heroku runs on top of AWS but operates a layer above it, which is why the two appear in the same category table here despite being very different products to work with.

Heroku's infrastructure runs on Amazon Web Services, but the technology stack figures on this page will not show you that and should not be used as evidence of it. Those percentages describe what a visitor's browser loads from the site: CDNs, analytics and frameworks. Cloudflare tops that list at 40.54% because site owners put Cloudflare in front of their Heroku apps, not because of anything Heroku does underneath. Where an application is hosted is a separate question from what it serves.

Ruby, Java, Node.js, Python, PHP, Go, Scala and Clojure through official buildpacks, with community buildpacks extending that to languages such as Rust, Elixir and Crystal. It began as a Ruby-only platform, which is still visible in the data: Ruby on Rails is the most common framework on Heroku domains at 13.95%, ahead of every JavaScript framework we detect alongside it.

Technically yes, but almost nobody in this data uses it that way. Of the 467 matched Heroku domains with an employee band, 6 are at companies of 10,001+ employees and 3 more at 5,001-10,000. The large organisations present, honda.com, harvard.edu and northwestern.edu among them, appear as a single hostname inside an estate running thousands of applications. A detection tells you the platform is on that hostname. It tells you nothing about whether it is a company standard.

G2 reviewers are consistent about both sides. The praise is for ease of use: a streamlined setup, straightforward deployment and multi-language support. The complaints are about the service rather than the product, and they cluster around support quality, billing disputes, account deletion with no recovery path, and difficult debugging. It reads as a product people like using and a service they struggle to get help from.

Heroku removed its free dynos, free Postgres and free Redis in late 2022 (per devcenter.heroku.com/changelog-items/2461). For more than a decade the free tier was how most developers first met the platform, and its removal is the event people usually reach for when explaining Heroku's trajectory. Our data cannot measure the effect, because the current crawl window starts in 2026, years afterwards.

Our data cannot answer a product question, but it can frame one. The companies on Heroku skew heavily to those founded in the 2010s, 63.66% of the 344 with a known founding year, while newer platforms in the same category draw 39.90% (Supabase) and 38.05% (Render) of their matched companies from the 2020s. That describes where teams starting today are going, which is a different question from whether Heroku still works. It does what it always did. What has changed is that Salesforce has stopped adding to it, so the ecosystem around it will keep getting older relative to the alternatives.