About AI Product Hunter
Overview
AI Product Hunter is an AI reviewer that evaluates products launched on Producthunt (PH) everyday.
Although its initial motivation is the dissatisfaction with Producthunt, it doesn't aim to spoil Producthunt.
It aims to enhance the Producthunt experience, making it more exciting for users and makers without stealing contents and traffic from Producthunt.
Evaluation Process
To improve the transparency and validity of its evaluations, evaluation process of AI Product Hunter is maintained in the following repository.
ai-product-hunter-evaluationMotivation: LastCapsule's terrible launch experience
On 2024 June 6 We launched our products "LastCapsule".
We recklessly adopted "no external upvotes strategy" because we have confidence of our product.
Results of the launch
7 upvotes(1 from us, at least 5 from likely fake users maybe operated by the "Producthunt Promoter")
0 views on the Youtube video(created over a month)
2 visitors to our website(1 seems a sales person who says he can sell 70 upvotes)
Postmortems
We found that our product had never been featured and concluded that "we never even participated in the race in the first place". However, others would probably just take this as loser's talk.
Of course, we have some responsibility for our ignorance. Even so, we felt indignant, not only we had wasted our time, but had burdened LastCapsule with negative social proof.
Was this really a valid indignation? What exactly is a fair race? We started thinking about these things, and as a result, we came up with AI Product Hunter!
Current Issues with Product Hunt
80% non-featured products likely not exists on Producthunt
Becoming "featured" is extremely important when launching on Product Hunt.
At what rate are products featured?
The following statistics cover the period from 2024-01-01 to 2024-07-15 (197 days).
- all: 21,915
- featured: 4,480
It can be said that only 20% of products are featured on Producthunt.
On Producthunt, users typically only see featured products.
By default, the "featured" tab is displayed on the web. Non-featured products can be seen in the "all" tab, but it requires infinite scrolling and long load times. And on the mobile app, there is no measure to view all products today.
Let's compare the upvote counts of featured and non-featured products.
The following statistics are aggregated data from 2024-01-01 to 2024-07-15.
- average upvotes of featured: 193.55
- average upvotes of not featured: 10.91
And based on our experience launching LastCapsule, fake users generate about 5 upvotes per product.
Considering the quantity of friendship upvotes and failed fake user boosting, it's almost impossible to gain recognition through Producthunt without being featured.
Whether a product is featured or not is determined at the time of launch, and it generally doesn't changes after the launch.
Producthunt says it decides which products to feature through an "algorithm", but the criteria are not disclosed.
Though we are somewhat skeptical about this claim, at least the feature status is an uncontrollable element for the makers.
From the above, it can be concluded that on Producthunt, "unless you pass the luck test of becoming featured 20%, you can't participate in the race, and the effort to launch is almost wasted."
The race is too dominated by the snowball effect
In the current race, there is the enormous advantage of sitting high rank among the featured products.
Everyone is busy. It's natural to want to check out only "promising products."
And it takes a bit of courage to vote for lower-ranked products. It's human nature to want to back a winner and avoid a loser.
https://pw2.akkio.com/As you can see from this, the product that wins in the first 10 minutes basically snowballs.
In other words, on the current Producthunt, if you don't get a good start, you can't expect an influx of "Producthunt users." You win if you can get a voting bloc early. The influence of gathering friendship votes (or fake user boosts) in the initial stage is significant.
Now, considering these aspects, doesn't the Producthunt race start to feel like boring bullshit?
Is Producthunt guilty?
At least for 80% of products, launching on Producthunt ends up being "a wasted effort to get minus social proof". This is disgusting.
However, we can somewhat understand Producthunt's policy of picking up promising products and having users focus on those.
In reality, there are many products that are difficult to find any value in(even from AI). For the Producthunters, watching such product is waste of time.
And for now, products that get featured tend to be of higher quality compared to those that don't.
The AI Product Hunter's score is one evidence of this. Let's compare the AI Product Hunter scores of featured and non-featured products.
- average score of featured: 155.96
- average score of not featured: 137.44
It seems that Producthunt's featuring strategy is not completely wrong.
If there are too many options, visitors may lose the desire to check them out.
And the more options there are, the greater the impact of friendship votes or fake user boosts.
However, as mentioned before, this policy results in 80% of products being just nothing. This is not known to ignorant but innocent people like us. So it's frustrating that Producthunt presents itself as if it were a prestigious race.
Fundamental issue: Evaluating products is a hard work
Considering that YouTube views are often about 1/5th the number of upvotes, even non-fake users might not be evaluating products "properly". We guess many genuine users may upvote based on a vague first impression without understanding the details.
To properly evaluate products at a level where you can say you've done it correctly, it takes about 10 minutes of intellectual effort.
A human might not think of doing that a lot every day if it is not a job.
How AI Product Hunter make Producthunt Exciting Again
Feedback for All Products
There are many invisible products on Producthunt. All new products may be hungry for feedbacks. Even if the feedback comes from AI, having some form of feedback would be good for the mental health and beneficial for considering the "additional investment" in the product.
Discovering Promising Products Through AI Evaluation
Can generative AI really evaluate products? You may question of this.
For now, our answer is "Yes."
AI Product Hunter has the "eyes" at least more than entry-level Producthunter.
For the understanding of the products, AI Producthunter showing very few misunderstanding.
The evaluation seems reasonable in most cases and the scores are generally convincing and exhibit reproducibility.
When you actually try the top-ranked products on AI Product Hunter, you might find that some of them have poor UI or implementation. As you may have noticed, AI Product Hunter have some aspects that cannot be evaluated such as UI/UX. However, many of the top products of AI Product Hunter tends to have attractive concepts. (In addition, it seems that few PH users thoroughly try the products before upvoting them.)
Generative AI is based on something existing that it has learned. As widely known, it is not good at proposing something new.
But because of this constraint, we believe it might be able to recognize something truly new. (If you want to challenge this, it might be an interesting experiment for you to evaluate each product yourself and create your own rankings!)
AI Makes Discovering Products Easier
On Producthunt, the taglines are often designed by makers to be impressive, which can make it hard to quickly understand what the product is.
There are many products that still remain unclear about what they actually do even after reading the description and the comment and HP.
AI Product Hunter generates taglines for each product that allow users to instantly understand what the product is.
This reduces the time needed for you to find the product that worth a closer look.
Additionally, since there is no need for long load times to display all products, it becomes relatively easy to skim through and check all of today's products.
Yet Another Exciting Ranking!
Why does Producthunt itself present products in a ranking format? It must be simply because rankings are fun.
Currently, it must be said that Producthunt's rankings effectively handle only about 20 featured products.
However, the AI Product Hunter's rankings average around 100 products, based on highly transparent evaluations.
Both rankings have their own significance. We hope you enjoy each of them.
To reiterate, AI Product Hunter doesn't intend to pollute Producthunt with AI-generated content. It is designed to be the complement of Producthunt. It aims to make the experience more exciting for those watching the race and those who have launched their products. We hope you have fun!