Ebbe Altberg on user retention now and in the future

A part of one of the new user experience social islands
A part of one of the new user experience social islands

While at the TPV meeting on June 20th, 2014, Ebbe Altberg didn’t only address questions on the new virtual world platform the Lab is developing. The session actually kicked-off with a discussion of user retention in Second Life and the new user experience.

Some eight minutes were devoted to the discussion, of which around five and a half were taken-up by Ebbe talking through ideas, and the remaining time either with questions or with reminiscences about mentors and the old mentor programme. The audio presented here represents all of Ebbe’s comments on the subject, together with a transcription for those who prefer to read his comments.

 

The first question was whether the Lab had any further ideas for trying to improve user retention.

Well, there are lots of different types of users, and I think we can make it easier for some users without losing the power for other users. But yeah, that is a tremendous challenge, allowing for this incredible flexibility, openness, freedom, and at the same time make it simple.

I am convinced there’s a lot of things we can do, not necessarily in the short-term, but in the medium-to-long term that going to make things a lot easier.

Ebbe Linden (LL CEO Ebbe Altberg) Talked about user retention at the June 20th TPV Developer meeting (image: Strawberry Singh from the VWBPE conference, April 2014)
Ebbe Linden (LL CEO Ebbe Altberg) Talked about user retention at the June 20th TPV Developer meeting (image: Strawberry Singh from the VWBPE conference, April 2014)

I mean, just coming in and getting dressed is a major undertaking, and it shouldn’t have to be. There’s a lot of stuff that is geekier than it need to be; what the hell does it mean to “detach” something from your avatar and stuff like that?

So I think the user interface can improve a lot. and obviously, a lot of the work that you guys have just discussed here [Experience Tools, etc] will make retention improve. When the quality improves, performance improves, that will do a lot as well. It’ll just take users a while to notice and appreciate, and therefore get sticker. It seems to me, the metric seems to indicate, you know, some potential extra stickiness is going on. We’ll have to see; we’ll have to wait a little longer to be sure about that.

But when I talk about making in “simpler”, I’m not suggesting I’m trying to make it dumber or less powerful. So it’s going to take some smart people quite some time to solve for, but we have to try really hard to continuously make it easier.

Hopefully we can have something with the power of Second Life, but appealing to hundreds of millions, and not just a million; we have to figure out how to get there.

The next question was whether Ebbe was aware of the official mentor programme, and whether he would consider bringing that back.

[1:57] I was actually in-world yesterday [Thursday June 19th, 2014] with a group of people, many of whom have been part of that, and so we started conversations. There’s a lot of sub-groups participating in that conversation, of helpers.

I think the difficulty is, for our perspective, or why it might have been discontinued, is how do you manage it at scale? Who do you trust? Who’s behaving? Who’s not behaving? I think that might have been a part of it.

But I think that also a lot of those efforts were unfortunately discontinued as a part of the big layoff back in the day, when a lot of things were thrown out of the window whether it made completely sense to stop those types of things or not.

[2:43] Personally, I’m in big favour of a – what was at the beginning, what was it called? Community Portal? … Community Gateway programme – to enable creators to attract their own audience into their experiences. I think ultimately, if we’re going to scale way beyond a million users but tens of millions or hundreds of millions, we have to allow creators to be able to attract an audience from the outside world directly into their experiences.

There we have to think about what does that mean? Does it become some kind of co-shared, co-branded on-boarding experience for people? Because there are too many unique communities and verticals and experiences, that we can’t correctly advertise and drive traffic to those experiences, and it would be much more powerful if the creators could sort-of attract their own audience, and that’s much more scalable. That’s the way I think about it, but how long it will take to get there, I’m not sure.

[3:43] So we’re actually experimenting right now, where we’re doing A/B tests with a welcome island as we know it, and a second welcome island with a live helper so we can A/B test the conversion rates and the stickiness of the product. Obviously, one person is not going to do it, but we’re starting with one person as a very consistent way of treating people coming in. and then we can see what the actual performance differences [are] between those two experiences. We’ve also tried with audio on and audio off.

So we’re doing these little A/B tests to see what works, and if we find that having a live greeter there has a meaningful conversion rate improvement , then we just have to figure out how we can scale that, so we can have people able to meet everybody.

I haven’t seen statistics come out of that yet. I saw some stats of audio versus no audio, and it was like a tiny, tiny, plus for no audio, so we’re now trying with no audio, or no speech, so that … but other things we can do is maybe only certain people can get to the welcome island … and [not] having cuckoo folk show up there … Having a way to … we can technically potentially solve it so that we don’t have to have people kicking people out, but you could be sort of, only new users or … and if you’re a brand-new user, it’s a little harder to grief than if you’re and existing user.

Anyway, it’s stuff we’re actively playing with. I’ll try to figure out when we’ll have statistically relevant numbers from the tests we’re doing, and then we’ll go from there.

[At this point the discussion moves to PR as a means of reaching out to new users, and then moves into the discussion of the Lab’s next generation platform.]

The Snow Lion, Oceanside dAlliez; Inara Pey, May 2014, on FlickrCommunity Gateways offer perhaps the most flexible means to meet the needs of incoming users (image: The Snow Lion, Oceanside dAlliez (Flickr))

Comments

The first interesting point of note here – and with the benefit of hindsight – is the first clue is given that the Lab is looking towards another platform for mass adoption:

“Hopefully we can have something with the power of Second Life, but appealing to hundreds of millions…” [my emphasis]

The idea of enabling established users to be able to build tailored on-boarding experiences which not only get newcomers involved in their activities, but also provide them with sufficient guidance to be able enjoy the rest of what a platform has to offer, could be a powerful move in the right direction. Not only does it resolve issues of scalability, it is also something that delivers the on-boarding process more squarely to users within a platform, something many in SL want to see.

It will be interesting to see if the Lab actively pursue such an approach with SL, or whether it is something they’ll look more towards implementing in-depth on the new platform. Perhaps we’ll see a little bit of both; the Lab initially “trialing” such an approach on SL, prior to enhancing  / expanding it with their new platform.

The further A/B testing with the current on-boarding process is interesting. Even if one discounts 90% of the current 350,000 (ish) monthly sign-ups as alt creation or spambots*, that’s still potentially 35,000 new users month who are coming into SL, of whom around 80% have apparently evaporated within a month. So any attempts to increase on this are to be welcomed.

That said, whether user retention can be signficantly increased by tweaking at the edges of the current on-boarding process is debatable. I tend to still be of a mind that without finding the means to connect incoming new users on a more “social” level with others within SL, attempts at increasing user retention will be limited in success.  Nevertheless, it will be interesting to see what comes out of the current A/B testing – and in seeing just where else the Lab is willing to invest time and effort in order to try to increase SL’s retention levels.

*Footnote: some claim that 99% of this number is the result of spambots which never actually go on to download the viewer, much less log-in. While I agree that the potential number of sign-up hits by bots is likely to be high, placing it that high is perhaps a stretch, and points toward’s Marcello Truzzi’s statement that such an extraordinary claim requires extraordinary proof.

Advertisements

13 thoughts on “Ebbe Altberg on user retention now and in the future

  1. I’m glad you highlighted this. I think this part of discussion has suffered very much from the “witch in the water” effect that I talked about with Jessica in our discussion at SL11B CC – that a big shock overshadows other valuable information. The fact that this is all going on is a very strong indicator that Second Life is not (yet again) dying – that Linden Lab are looking to gain and retain new users in Second Life.

    But – as Carl Metropolitan has repeatedly said – while the Community Gateways was a really good programme, with much better overall stats in many cases, the lack of institutional memory here is a real handicap.

    I can’t remember exactly when the Gateways closed – sometime in 2010, I seem to recall, and soon after we did a HUGE article on them in Prim Perfect’s March 2010 issue. 2010 was also the year of the mass lay-offs and M Linden’s hasty departure too (it took place, if I recall, while we were waiting for him to come online and make a keynote speech at that year’s birthday – SL7B). What I do remember is that they closed with 18 hours notice – much to the shock, disbelief and fury of the landowners who had invested thousands of US dollars in holding the required land, creating the necessary elements in an appealing way, and staffing them – sometimes with volunteers but often paying a small stipend to have them properly manned (which, I believe, was a requirement).

    The trouble is, while Lindens change, residents tend to stick (this is the still, I believe, only virtual world where people make regular incomes, after all). That means that even if the Lindens don’t remember, residents do – and reviving the much-loved community gateways could be a hard sell for the Lab, unless they are prepared to sweeten the pill.

    .

    Liked by 1 person

    1. Your recollection squares with mine. When I came in-world for the first time, it was a totally different experience, and went smoothly. but I do wonder if I got lucky, that first day.

      If we have multiple providers of new-user service, how can the quality control be managed? Different languages is something pretty obvious. But could I get an English language portal that worked well for me? British English or American? I remember I used to go one place that had a lot of regulars who had come into SL on the back of some US TV show. They got a special lastname, and maybe that common interest encouraged retention, until the show stopped.

      Like

      1. I think that was the Gossipgirl people. Not only did they have a special last name, they also had a set of special sims and regular activities – such as coffee morning meetings, and there was a team of people who mentored them and organised events (and also brought in characters from the show perioducally – much to the fans’ delight). That was a very full package – different to the CSI experience which – if I remember – brought in new people and dropped them in a murder mystery CSI New York scenario (I’m not sure how they handled training for all those newcomers).

        The Gossipgirl package was, I think, reasonably successful but not very big. We remade a cafe for their New York on a VERY early Designing Worlds show (we were still MetaMakeover and pre Elrik – http://treet.tv/shows/designingworlds/episodes/ep020-0). The CSI one was generally seen as not being a huge success.

        Like

    2. Yes, there was a mistake in the past in cancelling the old Gateway programme. There’s an admission to the mistake here in Ebbe’s comments. Question is, are we willing to forgive what was done and move on, and see what the Lab is willing to propose (hence why I opted to not rake over old coals by bringing up the old programme here).

      Yes, the Gateways did evaporate around the time of the internal changes at the Lab in 2010 – although I think they occurred ahead of the “big” layoffs (which people tend to forget also came after a 18-month period when Mark Kingdom had expanded LL’s workforce by some 50% with around 118-125 new hires, but that’s another story).

      One of the issues with the old programme that the Lab has pointed to is that of scalability – the implcation that as the old programme grew, due to the way it was set-up and run, it required increasing overheads on management by the Lab. I’m not entirely sure why that should be, but I doubt it’s merely an excuse that gets rolled-out when people ask “why was it cancelled?”. This time, the onus would appear to be something that is perhaps more self-regulating while at the same time building an apprach that leaves the Lab with far less of an onus on having to “manage” or “administer” things. Again, how this might be achieved, I’ve no idea. but i do still maintain it is potentially the best approach there is.

      Liked by 1 person

  2. The Lindens must have the raw data to put the numbers on retention, and the drop-outs at different stages. Things such as A/B testing, and using that label, suggest they are trying to use the numbers they have.

    But there is more to this than the numbers. It’s easy to warp survey results by choosing how questions are worded. There’s more to this than just counting.

    Like

    1. Yup, the Lab has a lot more data that at times people are prepared to credit them as having. This isn’t the first bout of A/B testing they’ve done either; Rod Humble pointed to much the same approach being taken a year ago when they made the initial changes to introduce the current round of learning islands (which still see to be called “welcome islands”) and social islands.

      My problem isn’t so much with the idea that numbers / surveys can be warped. It’s more a case that everything is being baselined off the same approach / criteria, ergo, the results all tend to indicate the same thing, encouraging inertia when it comes to thinking more broadly about matters.

      Like

  3. Has LL ever performed and shared a detailed assessme t of exactly who / what make up all these current supposedly new sign ups? I agree that likely a large percent of the signups are bots which any LL retention and engagement strategy has no value to improve. I also believe another manor portion of these signups are existing user alt by griefers ,harassing / sexually manipulating users , and victims from those that harass. I know of several jerks that have 30+ alts they use to havr multiple relations and to spy and harass victims. I would guess most existing users have at least one or two alts. So a strategy to increase or retain these users is futile as they already exist in sl and know the grid. So… how many of the daily signups are the TARGET POPULATION of real new users that LL wants to retain?

    A detailed assessment is to analyze new account activities, metrics (what viewer what IP ) … to see if there is a more accurate scope of who really makes up these new signups.

    If LL cannot figure how to attract and retain new users into the industry leading virtual world, what hope do they have to attract retain net new users to the new SL2??

    Like

      1. So if they do, would they be willing to share with us what percent of the daily / monthly signups are NEW HUMAN USER ACCOUNTS, EXISTING USER ALTS, BOTS & OTHER?

        My suspicion is that if they really did know the actually ratio to any level of accuracy, they would not want to share the numbers because it would be revealed that likely only 10% – 20% of all new signups are real new human user accounts (the ones that Ebbe and LL’s strategy wants to retain). The exposure to these numbers would show that actual NEW SL USER interest is far lower than LL has been posting over the years. It would be real but the truth would be what the media would pounce on – that SL vw popularity with NEW users is much worse than has always been reported.

        I don’t think it would surprise most SL Bloggers and SL Residents that know most SL accounts are either bots and alts – not actual user counts.

        The reason I bring this up is, if the daily new user signup as far smaller than the grand numbers LL reports, is fixing retention from such a small population require so much effort and sweat? Maybe fixing this problem is more costly and consuming more LL resources to fix than what would be solved. Maybe LL should focus their resources and development priorities and better understanding SL resident frustrations (like broken existing features, and continued sim lag, and mesh bake lag/fails, and improvements on missing/weak basic functional).

        Only LL believes they have solved much of the sim / mesh lag offered major new shiny features like pathfinder that the SL community has been begging for and resolving steep learning curves for real new users. Sl Residents want basic features fixed or improved, but most of these are ignored by LL as they think they know better.

        Like

Have any thoughts?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.