Review sources
How to add Product Hunt reviews to your website, not just upvotes
Product Hunt hands you an embeddable badge, and that badge counts upvotes. We fetched it on 6 September 2026: 2,063 bytes of SVG, 250 by 54 pixels, carrying the words “FEATURED ON Product Hunt” and exactly one number, which is the upvote total. There is no rating in the file, no review count, and not one word written by a customer. To put actual Product Hunt reviews on your website you have to import the text from your product page and render it on your own domain, and this post is about what that text turns out to look like once you do.
Everything below is a measurement. The live requests were made on 6 September 2026 and the counts come from our production database on the same day. The picture is not flattering to Product Hunt as a review source, and it is not especially flattering to us either, since we hold 243 Product Hunt reviews in total. Both halves are here.
The badge is an upvote counter
We asked the badge API for two different posts and compared the files.
| Request, 6 September 2026 | Result |
|---|---|
featured.svg?post_id=1&theme=light | HTTP 200, 2,063 bytes, image/svg+xml |
featured.svg?post_id=100000&theme=light | HTTP 200, 2,079 bytes |
| text rendered in the first | “FIND US ON”, “Product Hunt”, “1” |
| text rendered in the second | “FEATURED ON”, “Product Hunt”, “104” |
occurrences of rating, review, star | 0 in both files |
| cache header | cache-control: public, max-age=14400 |
The number moves with the upvote count and nothing else. The wording flips between “FIND US ON” and “FEATURED ON” depending on whether the launch ranked. That is the whole payload. It is a small, well cached file and we have no complaint about how it is served, but it is worth being clear about what a visitor learns from it: that you launched, and that some number of people clicked an arrow, on a day that may have been four years ago.
A Product Hunt reviews badge does not exist
The obvious next question is whether Product Hunt offers a second badge, one that carries the rating. We probed the badge API for it. The answer is legible in the shape of the failures rather than in any documentation.
Endpoint, valid post_id, 6 September 2026 | Response |
|---|---|
featured.svg | HTTP 200, SVG |
top-post-badge.svg | HTTP 404, 85 bytes, application/json, {"error":"not_found"} |
top-post-topic-badge.svg | HTTP 404, 85 bytes, application/json |
reviews-badge.svg | HTTP 404, 461 bytes, text/html, “Product Hunt: Page not found” |
daily-reviews-badge.svg | HTTP 404, 461 bytes, text/html |
review-badge.svg | HTTP 404, 461 bytes, text/html |
product-review-badge.svg | HTTP 404, 461 bytes, text/html |
Those two 404s are not the same 404. The JSON one is an endpoint that exists, understood the request, looked for a top-ranking post and did not find one. The HTML one is the site’s generic “page not found” screen, which is what you get when the route was never built. Every review-shaped name we tried landed in the second group. If you want a rating from Product Hunt on your own page, there is no first-party image that will give it to you.
Product Hunt’s own review page hides from browsers and answers scripts
While measuring, we hit something we did not expect and repeated it until we trusted it. We requested the same public reviews page three ways.
| User agent, same URL, 6 September 2026 | Result |
|---|---|
default curl, 3 attempts | HTTP 200, 264,854 bytes, 3 of 3 |
| desktop Chrome string, 3 attempts | HTTP 403, 5,774 bytes, Cloudflare challenge, 3 of 3 |
| Googlebot string, 2 attempts | HTTP 200, 264,854 bytes, 2 of 2 |
That is backwards from how bot protection usually behaves, and we are not going to pretend we know why it is configured that way. The consequence matters more than the cause: your Product Hunt reviews live behind a bot filter you do not control, on a page whose behaviour changes without telling you.
There is a second thing inside those 264,837 bytes. The page carries two blocks of structured data,
a BreadcrumbList and a WebApplication describing the product. Neither contains
aggregateRating, ratingValue or reviewCount. We searched the whole document: zero occurrences.
So even on Product Hunt’s own reviews page, the ratings are not published in a machine-readable
form. Anyone hoping that linking out to Product Hunt will feed a rating into search results or into
an assistant’s answer is hoping for something the markup does not offer. If star markup is your
actual goal, that has to happen on your page, and we walked through how it works in
review schema markup and stars in Google.
The reviews arrive in a burst and then stop
This is the part that changes how you build the section, and it is the part only our database can tell you. We hold 243 Product Hunt reviews across three accounts. They span 3,011 days, from October 2017 to January 2026. They fall on 63 distinct calendar dates.
Five of those dates carry 154 of the 243. That is 63.4% of every Product Hunt review we have ever imported, landing on five days out of eight and a quarter years. The year 2021 alone accounts for 165 of them.
The clearest version of the pattern is inside a single account, where nothing else varies:
| One account, our database, 6 September 2026 | Reviews | Distinct dates |
|---|---|---|
| Product Hunt | 161 | 9 |
| 139 | 64 | |
| 13 | 11 |
Roughly the same number of reviews from Product Hunt and Facebook, and the Product Hunt ones are compressed into a seventh of the days. Launch traffic behaves like a firework. Across all three accounts, the last 365 days produced two Product Hunt reviews and the last 30 days produced none.
So the honest framing is that Product Hunt is an archive, not a stream. That is fine, and an archive is still worth showing, but it means the section on your site should not be built to look like it updates. We ranked every source we import by whether it still produces anything in which review sources still work in 2026, and Product Hunt sits firmly in the archive half of that list.
Two thirds of them contain no words at all
The second surprise is density. On Product Hunt the rating and the write-up are separable, and most people stop after the rating.
| Source, our database, 6 September 2026 | Reviews | No text stored | Share |
|---|---|---|---|
| Product Hunt | 243 | 159 | 65.4% |
| 388,104 | 125,055 | 32.2% | |
| 49,145 | 8,660 | 17.6% | |
| Booking.com | 3,260 | 506 | 15.5% |
| Capterra | 6,697 | 53 | 0.8% |
| G2 | 15 | 0 | 0.0% |
| Yell | 2,831 | 0 | 0.0% |
Product Hunt is the emptiest source we import, by a distance, and it beats Google’s already high share by more than two to one. Of the 84 reviews that do carry words, the average length is 120 characters. For comparison, on the same day and counting only reviews that have text, Google averages 198 characters, Facebook 190, G2 469 and Capterra 489. A Product Hunt write-up is typically one sentence of launch-day enthusiasm.
The ratings are strange too, and we would rather flag it than smooth it over. The 243 break down as 167 five-star, 13 four-star, 50 three-star, one two-star and 12 one-star. Three stars being the second most common score is unusual, and 39 of those 50 have no text, as do 108 of the 167 five-star ones. We are not going to invent an explanation for a shape we can only see from the outside.
What this means when you build the section
Sorted by how much usable material each source actually hands a testimonial widget, Capterra and G2 are at the top, Google and Facebook are in the middle on volume, and Product Hunt is at the bottom on both density and recency. That ordering, not alphabetical order, is how we would decide what goes on a page.
The practical consequence is a mixed section rather than a Product Hunt wall. Pull the handful of Product Hunt reviews that have real sentences, put them next to a source that still produces something this month, and let the launch material do what it is genuinely good at: showing that early adopters showed up. If you sell software, the two sources worth pairing it with are the ones that write paragraphs, and we measured both in how to add G2 reviews to your website and how to add Capterra reviews to your website.
How to add them
Connect your public product page, in the form www.producthunt.com/products/your-product. Not the
/posts/ launch permalink, and not a topic or collection page. Our
Product Hunt reviews widget reads that page,
imports the reviews that have text, and renders them as HTML in your own markup, so the words are
indexable, quotable by an assistant, and present for a screen reader. Then choose the source mix per
widget: pool Product Hunt with a livelier source on the pricing page, and use a single source on a
case study where that fits better. You can have the first widget on a page in a few minutes from a
free account,
and the step-by-step version with screenshots is in the help centre.
Three caveats, because they are real and we would rather you hear them from us. Product Hunt sits on
our Ultimate plan at 39 EUR a month, alongside G2, Capterra and the other sources that need a crawl
rather than an API; the current tiers set out what that costs. Our sample is small: 243
reviews across three accounts, so read the percentages above as the shape of the source rather than
as a census. And crawling it is not reliable. We hold four Product Hunt connectors, and when the
health check ran on 5 September 2026 exactly one reported complete, while two reported failed
and one invalid_url.
What to do about it
If your product launched well and the proof is sitting on Product Hunt, move the words onto your own site, keep the expectations calibrated, and pair them with a source that is still adding reviews this quarter. The badge is not a substitute: it is a picture of an upvote count, and there is no review badge behind it to graduate to. The sentences a hunter wrote about you on launch day are the only part of Product Hunt that can still do work for you, and they only do that work where a reader, a crawler and an assistant can all reach them, which is your page and not theirs.
FAQ
Is there an official Product Hunt reviews badge?
No. We probed the badge API on 6 September 2026. The endpoint featured.svg returns a working image, and top-post-badge.svg returns an 85-byte JSON error saying the object was not found, which is what an endpoint that exists looks like when your post does not qualify. Every review-shaped name we tried, reviews-badge.svg, daily-reviews-badge.svg, review-badge.svg and product-review-badge.svg, returned a 461-byte HTML page titled 'Product Hunt: Page not found'. That is the response of a route that was never built.
What does the Product Hunt badge actually contain?
2,063 bytes of SVG at 250 by 54 pixels, the words 'FEATURED ON' or 'FIND US ON', the Product Hunt wordmark, and one number. The number is the upvote count: post 1 renders a 1 and post 100000 renders 104. The file contains zero occurrences of the strings rating, review or star. It is an upvote counter with good typography, and it is not evidence about your product's quality.
Which Product Hunt URL do I connect?
The public product page in the form www.producthunt.com/products/your-product, the page a visitor reaches from search and the one that carries the reviews tab. Our only Product Hunt connector currently reporting a healthy crawl uses that form. Two of the others point at /posts/ launch pages and neither is crawling today. That is four connectors in total, so treat it as an observation rather than a rule, but start with the product page.
Why do so many Product Hunt reviews show a rating and nothing else?
Because on Product Hunt the rating and the write-up are separable, and most people stop after the rating. Of the 243 Product Hunt reviews in our production database on 6 September 2026, 159 have no text stored at all, which is 65.4%. That is the highest share of any source we import, ahead of Google at 32.2% of 388,104 reviews. A testimonial widget has nothing to print for those, so plan the section around the 84 that do carry words.
Which plan includes Product Hunt?
Product Hunt sits on the Ultimate plan at 39 EUR a month, the same tier as G2, Capterra, Sitejabber and Yellow Pages, because those sources need a crawl rather than an official API. Google and Facebook are on the free plan. The current tiers and limits are listed on the pricing page.