WMA - Pornhub's Natalia Hampton's Presentation - Comments
 

Yesterday, Pornhub's Natalia Hampton gave a great presentation on how best to use your metadata to optimize for more views. Thank you Ms. Hampton for taking the time to compile that.

The focus was mainly on models uploading content, but most of the points apply to Content Partners. As a paysite owner, I wanted to make a few comments and observations.

Localization and Translation

Pornhub added an option for you to add translated titles to your videos and they encourage it. Xvideos has had this option as well, for quite some time. I've used it extensively, and while there is some uptick in views, the clickthrough sales were no better than not translating at all. I believe this is because my landing page and site overall, is entirely in English. So a non-english viewer, coming from a translated video, would be confronted with - to him/her - unintelligible text. Therefore, I believe having your site translated, should be a priority, before sending foreign traffic to it. A little bit of "captain obvious".

It's well known that users search for things similar to themselves. So a user from Spain would search for "espanol" or "spanish". Natalia makes it a point to geo identify your content in this manner so you have a greater chance of appearing in search results for the users of that geo. But I'm fairly certain it also applies to what is displayed to the user on the homepage's geo targetting:

https://i.imgur.com/oOhJNVm.jpg

It does, however, appear to be checked by the compliance team. For example, you can't randomly label something as "Canada", if it's an adultlab clip with Russian girls. I've been recommending to paysite owners to title and tag where the actresses, featured in the videos, are from.

Votes Don't Really Matter

Natalia recommends not being deceptive in titles and tags, so you will get a better rating and views, but it's not the main driver. You want your clip to be favourited, and to a lesser extent, playlisted.

https://i.imgur.com/CtCA0Eq.jpg

I aggregated data when I was working at BLACKED, and aggregated my own data, and found these two metrics were the strongest indicators for view performance. That's not to say votes/rating are completely useless. As Natalia says, if your video is not accurate to it's title and is just bad, your video will get a low rating, and that hurts its performance. But there appears to be a threshold. A video that is 90%, does not perform signficantly better than 70%. Their algo seems to take into account a portion of users who just downvote and are jerks.

Every CPP video gets an equal shot of success when it is featured and posted on the homepage. The more favourites it gets within the first few hours of featuring, the longer it will stay on the homepage and get more views. It also greatly increases the chance of being refeatured.

Video length (aside from the standard minimum) and comments, do not seem to matter.

My suggestion: long cuts of prime action that users can jerk off to, instead of quick, snippet-style editing. If a user can jerk off to the clip, he is more likely to favourite it and want to rewatch it. This is not counter intuitive thinking. A user who is satisfied, is more likely to buy - maybe not in the moment, but later (buying habits do appear to concentrate on weekends, suggesting less impulsive buy behavior). Most importantly, more favourites mean more views, and more refeatures, and your content gets exposed to more potential customers.

Focus On Trending

Natalia makes a great suggestion that certain trends can be predicted, such as movie releases and holidays. But I think she downplays how important trending terms are. I think the algo gives it considerable weight.

I notice when I upload a video that features a trending model or subject, it gets featured much faster, often bypassing the 1 week queue.

BUT - HUGE BUT - don't rely on what you see listed on the site. That data is not current. For example, the top trending models listed, are probably a few days or weeks old. If you try to produce content around that list, you would be screwed. Just ask your rep what is trending, and they will tell you.

Don't Make Content Outside Your Norm. Stay Consistent

Natalia brings up this point and I wanted to emphasize it even further.

This is another huge "BUT" for chasing trends - don't do something outside your norm, for the sake of catching a trend. Your users will not like it. For example, if anal is trending, don't make a bunch of anal videos if you are a foot fetish site.

Example from personal experience: When I worked at BLACKED, I watched other paysites start shooting interracial content or shoe-horning interracial themes, into their paysites, during the time BLACKED was blowing up in popularity. This didn't help their sales at all and likely alienated some of their existing customers. Unless they could deliver a product consistently as good as BLACKED, they would always look inferior by comparison. It's no wonder so many interracial starter sites have disappeared, and those left, are imprints of networks.

Keep doing what you do best and be the best at it. Look at X-art. They had 50k members at one time, but stopped making good content, even so much as buying content from adultlabs. It made it much easier for VIXEN to completely take away its market share.

The Rise In Kink Interest

I forgot her exact words, but I disagree with Natalia's reasoning for the rise in kink and BDSM interest on pornhub.

I believe it's more to do with the implosion of kink.com, and to some extent FOSTA/SESTA, which have drove fans of the content to look for new sources to satisfy their preference. Pornhub essentially being the porn search engine of the world, these users have gone there to discover new content. It just makes more sense to me.

Some Other Things I Wanted To Add

• the CPP featuring system only features one video per week and I believe, it still only takes the longest video in your queue, not by upload date. So watch out.
• Show your best content. I keep seeing paysite owners posting only their old or crappy videos. The logic is that keeping the best stuff in the members area, gives the users reason to join. But think of it like a movie trailer: if you saw a trailer full of horrible, boring parts, does that give you the impression the full movie will be good? Of course not. Post the good stuff, and your potential customers will think you have even better stuff in your members area.
• Don't wait to post a video of a hot model. Same reason as above and if you wait months later and the girl is no longer trending, you lose all the potential traffic from riding that trend.
• 10 years ago, Michael Pahl told me SEO for paysites, doesn't really matter, and he's mostly right. Big tubes like xvideos and pornhub, spend millions per year to get all the juice for short and longtail google search terms. You can't compete with that. Instead, treat them as search engines of porn and optimize there. That's the gist of Natalia's presentation. SEO nowadays is really about compliance and user experience. Google will reward you by making sure your brand/domain appears at the top of searches for your brand/domain.

What an excellent post! Thank you for taking the time to write this breakdown for us. As a fellow paysite owner I agree 100% with your observations and suggestions (and, by extension, Natalia's).

:thumbsup:thumbsup

Interesting post, thanks for sharing!

1000% agree. Too many people here are trapped in the 1990s and think that they can protect their "money shot" clips as if it's some magic that will compel a viewer to pay. Fans pay.

Great post movieguy!

Nat from PornHub shared this from the presentation yesterday, check it out:

http://webmasteraccess.com/extrafile...-category.xlsx

Thanks for writing it all down, I also really enjoyed Natalia's presentation.

Great post and thanks for taking the time to share it

Outstanding post indeed :thumbsup !

Thank you for watching and your summary! Responding to different aspects of what was discussed above:

My caveat for my presentation - my data was exclusively pulled off interactions of models/users and might not apply to studios/channels. Therefore featuring algo and trending context is different than for CPP.

Trending can bring more views for model videos - just a matter of finding out exactly what's on trend and tagging appropriately, which can help views. Obviously if flooded with same content, you've missed the sweet peak of getting content in time.

Looking at x-art vs vixen etc - I think an important part of that discussion is also CONSISTENCY: consistently uploading to your profile increases traffic, we see a general "sweet spot" of about 2-3 video posts a week. This keeps fans interested and coming back.

I can definitely take a look at the time line of kink + niche content growth vs implementation of FOSTA/SESTA and kink.com - caveat is that a lot of the kink growth is actually not USA based, which is why I don't think it has to do with FOSTA/SESTA.


If anyone has other questions I'll hop on to answer as I can too, again thank you for watching! :)

Very useful, thank you :thumbsup

Yeah..say her yesterday..Great presentation..a lot of great stuff

Great post, thanks!

Are you single ? ?

LOL.

I am an unfeeling robot in a human suit.

This is amazing Thank you Thank you Thank you!!!!

Aren't we all :pimp ? ?

Yes yes but a SEX robot in an unfeeling human suit?
That could be useful. :)

PS: Robots in human suits new niche on PH?

Great thread!

Great sharing .. many thanks keep going :thumbsup

Christian ended up going into more detail about the CPP and it's recommended to watch his presentation which was a day after Natalia's. But I wanted to touch on the incest-trend: it doesn't seem to be slowing down.

Here's a list of the top ten pornhub premium channels. My best friend and I, are only 2 of 5 channels that ARE NOT faux-incest-themed, in the top 10:

https://i.imgur.com/D3Yiuq2.jpg

I'm proof of what Christian said about trying cool new ideas and being consistent, working. We didn't have major studio backing like the others. But if you want guaranteed success, it seems you should make a step site.

Just look at Spank Monster. He smashed into the top ten within months, after a friend recommended he shoot POV and label the girls as step sisters and step daughters.

I also think this shows pornhub's algorithm (for the most part) is working honestly. Mindgeek would prefer the content not be so incest-heavy. Just look at Christian's body language when he touches on the topic. Everyone I spoke to at Mindgeek, reacted the same way. I think there is almost no manual manipulation of the rankings and content.

So if you want success, you can improve your chances by making incest-theme videos, but put it on another channel as Christian suggested, and not part of your main brand. There are ways to screw it up. Stepsecrets.com just took joymii-style videos and labelled them as "Step". That doesn't work. The girls have to look young and innocent and milfs look like milfs. Warning: if you use your own gateway, you may not be approved, but all the ISPS's I talked to, said it's all clear.