Category Page Experiments


#1

Today we’re announcing the launch of a new experiment for the default display of results on category pages.

Over the last few months we’ve been rolling out changes and updates to the item layout on category pages. All marketplaces are now using the new ‘card’ layout and we’re very happy to report that the results are showing an increase in conversion.

With these new cards now in place we are going to be running a new experiment to learn more about how the display of category results impacts buying behavior. In particular, we are going to be running an A/B/C/D test to look at the different display methods currently available.

For a limited time, users will see one of four listing options as the default; best sellers, newest items, trending items and best rated. Users will still be able to switch between methods of sorting through the dropdown menu in the top right corner.

As always, our goal with these experiments is to learn. We do expect that some default methods will perform better than others and we’ll be closely monitoring the performance of each listing method, adjusting the experiment as we go. We’ll post any updates or future changes here as they are confirmed.

We’ll be on hand to answer questions for the next seven days, responding in batches as necessary. Please remember our Community Guidelines as you post.


Best rated items are shown by default. Where is the newest items?
Have you seen the "Top new files"?
Themeforest category filter
Recent search results - bugs?
There is a problem in sort by in themeforest
Themeforest news
The default view By : Best Rated !!! Why ????
The Sales Monitor 2
#2

@jamesgiroux You should mention what is the current before-experiment listing.

Also I thought there experiments were ongoing some time ago… also you should give exact dates for authors to know when what sorting is applied. We need to track our sales too.

Shouldnt we know dates when which sorting is applied?


#3

It was newest by default before.


#4

Hi James! The last time you started a similar experiment, all new tracks were without sales for 1.5 months. I remember there were a huge number of dissatisfied non-elite authors, I hope the story does not happen the same way (I’m talking about March 2017).


#5

Hasn’t this “experiment” been done numerous times over the years already? What more is there to learn here?

But sure, tweaking the “sort by” default is more important than to work on the pressing issues the community has with Envato. Well done!


#6

Thanks for keeping us informed about these experiments and it’s great that the conversion rated has been increased.

I don’t know about the other markets but I can tell for sure that in AudioJungle, listing best rated and best sellers in the category pages are a nail in the coffin for new tracks. Literally nobody rates tracks at AJ and the “AJ best seller museum” have way to much monopoly in the search engine already. These category pages is where new items have their short moment in the sun of visibility.

25% of all the music tracks at AJ were added just the last year. Over saturating has never been more extreme. Pluss count in the aggressive Elements banner which increases the over saturation even more. This is not just bad for authors, but also not improving the customer experience. I’m talking about those customers who got tired of being force feed the same old popular tracks list 2 years ago.

Yes customers are king, and what they want is what should happen right? Or is it like Steve Jobs said; customers don’t know what they want until you show it to them? It’s time to think new. Try showing them music playlists. Every week a small new playlist of great handpicked music is added to a “featured genre themed album” along with featured single tracks. This is something that many regular customers at AJ would love for sure + it would attract new customers who get panic attacks just by looking at the current chaotic search engine.

Regarding experiments to the search engine (for both customers and authors) take a look again at these 2 AJ threads because I really believe this is where the money is at long term:
https://forums.envato.com/t/ajs-search-algorithm-is-not-sustainable-the-popular-files-list-is-going-stale/
https://forums.envato.com/t/idea-for-search-algoritm-update/


#7

Ouch! :grimacing: 2 years ago it was a disaster for new projects. I hope this does not happen again this time :crossed_fingers:


#8

@jamesgiroux
Hi, James!
Since the Envato team began all these experiments, the incomes of many ordinary authors (not the elite) have gone down. New items, almost has not sales. Lately, it also doesn’t matter how much items you have in the portfolio… What is the use of this experiments for the most ordinary authors? (not top level). You write about the conversion… But for example, I see the opposite last few months, - sales are getting worse and worse… What does this “increase in conversion” give us?


#9

Thanks for letting us know.
But please keep in mind, if Envato change default sort from newest items to something else… it will dramatically influnce on new items sales.
So in the end authors will upload less than usual, and Envato will earn less than usual in long term.
IMHO before changing something, you should find a different way to promote new items.

I will not upload anything till the end of this experiment. Sorry, it was too expensive for me previous time, when you did it.


#10

Definitely. Just wondering, like music playlists or let’s just call them music collections, could something similiar work for Videohive and other marketplaces? Where the follow feed used to be, there could be each week a new collection of handpicked high quality items, a mix between newer and older items in a themed genre or aimed at specific use?

You just started a thread looking for Sci Fi cyberpunk tracks, so you know how time consuming and hard it is to find something in a niche category or for a specific use. These collections would only need a nice icon for the collection and let’s say at least 10 items that fits together for the purpose (Synthwave, Sci Fi cyberpunk, Stomp/Percussion, Smooth Jazz, Cooking shows, Extreme Sports etc etc). When Envato is out if ideas they could just start over again, like Extreme Sports collection 2.

I really think this would attract more customers and be an important addition to (at least for AJ) a search engine that lacks good ability to search for usage and themes. I also think this would help for the over saturation because authors would see the possibility of getting into these themed collections and possibly focus more on quality and different genres.


#11

Something needed is “Similar Items” from other authors at the bottom of a product page


#12

It will not be a replacement.

First. Category pages with sort by newest items are highly visited. It’s first thing you see when you click on category. It attacks a lot of views. And you offer to build another “categories” calling them “collections” and put it all somewhere near “follow feed”? Answer this questions:

  1. How many times do you visit “Follow Feed”?
  2. How may times do you visit “Top New Authors”, or “Public collections”, or “Top New Files”?

Those pages don’t attract so many views. We need full replacement, it’s mean something that attracts the same amount of views.

Second. my problem in AJ search is that I can’t filter and exclude some authors from search result. So basically I want to have instrument that cutout everyone who do not meet my personal quality standard.


#13

No, I do not mean collections as a replacement for new items search in the category pages. To sort anything else than new items in the category pages is simply crazy and counter productive in the long run. Customers can’t be entertained forever if you just offer them the same old items everywhere in the search engine.

Not much, except I check Top New Files now and then to see how overall sales of new items are doing.

The point with collections is to have them as an addition to the search engine for customers. Of course they need to be of super high quality, not just some random collection. The idea is to have them prominently visible (just as featured files). So the question is really how much views does featured files have? If it’s significant, we can expect collections to have significant views as well.


#14

Thanks for info! I support tests and careful conclusions :slight_smile:

I just want to point a tiny issue: on AudioJungle ratings do not work at all :slight_smile: Almost all our rated items have 5 stars because buyers see what they get in the preview before the purchase. Other thing is that only very, very small percent of our items gets rates (for exactly the same reason - if buyer knows the quality before purchase, why would he spend time on rating it after the purchase?).

Cheers!


#15

i tend to believe this is better to have author’s own other items …


#16

Hi @jamesgiroux

I realise I have not asked a single question on topic here, so I will try again :slight_smile:

  1. Do you understand authors concerns about how these experiments can affect new items negatively?

  2. Do you have any ways to estimate the long term effects of short term experiments?

  3. Are there any other experiments planned that are aimed to help new extra high quality items gain visibility?

Thanks!


#17

Hello,

We have seen that the tests started in one of the categories. The sorting indicator indicated that the products are sorted by the best rated ones. In practice, however, the products were sorted according to the “old default” order, that is, the newest ones.

Regards,
Tangible Design


Recent search results - bugs?
#19

Hi

@jamesgiroux I salute the popular categories sections, in which all categories are now visible permanently. Also, the card layout is cool, however, I personally need time to adjust as I was used to icons among which I could navigate faster, but as I said this is a welcoming change.

However, regarding conversion statistics, I would not jump to conclusions as these months are known to be better. I presume you are measuring in comparison to the same timeline with last year?

I still advocate more control or automation regarding discounts and sales as this is what is killing it all, but
this is not part of this topic :slight_smile:

Cheers


#20

It’s happening already: 60% drop. And it’s almost 2 months like this.


#21

Hey @jamesgiroux it looks like category pages are ordered by “Best Rated” (for one of the testing scenarios) - that’s totally fine apart from the “Best Rated” algorithm is completely broken.

  • Item 1 has 50 ratings at 5 stars, therefore a perfect 5 stars.
  • Item 2 has 400 ratings at 5 stars, a few at 1 star, and the odd 4 star, its rating is around 4.98
  • Item 3 has 3 ratings at 5 stars, therefore a perfect 5 stars.

Whilst there’s an argument that item 1 is therefore better rated than item 2, it’s really not fair to say that overall, item 3 is rated better than item 2, as only 3 users have rated item 3 as apposed to 400+ for item 2.

Yet in the current “Best Rated” sorting algorithm, Item 3 (and hundreds of others like it) would be shown before item 2.

What this creates is a rating system where a 4 star review is hugely damaging, whereas a 4 star review is actually meant to be a good review, not perfect, but above acceptable.

Lets take the https://themeforest.net/category/wordpress/corporate page for example. The top item there “MF” has 123 5 star ratings as of today. If tomorrow that item receives a single 4 star rating, boom, it’s completely gone from this page.

In fact you need to go back to halfway down page 9 just for this category before you get to items that have nothing but 5 star reviews: https://themeforest.net/category/wordpress/corporate?page=9

Take a look at this screenshot. This was taken from halfway down page 9 of the category mentioned above. An item with 60 sales and 3 reviews ranks above an item with 1.7k sales and 83 reviews. Of those 83 reviews, 82 of them are 5 stars, and 1 of them is 4 stars.

It’s totally broken, can Envato say that this is going to be improved? How can we be promoting items based on overall rating when what we’re actually saying is “This item hasn’t run into a 4 star review yet”, it’s pointless, there’s 5 possible review scores available, but only 1 of them is worth anything.