July 9, 2011 by George Eberstadt
I just got an invitation to Google+. After a brief time with it, I’m making a prediction: the Circles thing isn’t going to work. With Circles, Google+ is making the play to become one network spanning many types of relationship and purpose by letting you restrict your sharing and filter your view of the global feed by sub-group. Work. Friends. People I follow. There are two reasons I doubt this will work.
Reason #1: I blogged a while ago about something we learned (the hard way) at TurnTo about granular privacy control. It doesn’t work. There are people who will use your system, and there are people who won’t use your system, but there are very few people who would-use-your-system-if-only-they-had-more-granular-privacy-control. Early on, we built a very similar privacy model to Google+: it provided groups to enable users to restrict what got shared with whom. Like Google+, we offered a set of starter groups and allowed user-customization. Later, we ripped the whole thing out. We came to understand that (most) people want to manage privacy at the level of the network, not sub-groups within the network. People do their work-related networking on LinkedIn and their personal sharing on Facebook. Twitter is great for following people you don’t know personally (and therefor also useful for businesses to promote themselves). People decide who to connect to on each network based on what they plan to share (or read) there, and then they share fully with all their connections. That’s as granular as it gets. An item that isn’t suitable for someone’s whole audience on a particular network doesn’t get restricted to a sub-group, it just doesn’t get shared at all (or it gets shared with everyone with whatever consequences…).
Reason #2: Different system services are optimal for different types of network; there’s no one-size-fits-all collection. As a professional network, LinkedIn provides a great structure for exhibiting your work history. As a personal network, Facebook has great photo sharing. As a network of mostly-nonpersonal-following, Twitter provides great link sharing. The network services and the community co-evolve and specialize over time. If Google+ members really do try to maintain many different types of relationships within the system, they’ll end up unhappy with the tools the system provides. Either the tools will be too sparse, or the tools to support one type of network will be clutter to the others. (Would an elaborate resume system be appropriate for your Facebook profile?) In order for a social network to provide relevant services, there needs to be some level of focus to the type of relationships the network supports. And once the network has that focus, groups become irrelevant.
If I’m right, someone buy me a beer. If Circles works out, the drinks on are me.
June 8, 2010 by George Eberstadt
It’s a big day at TurnTo: we’re introducing our Social Commerce Suite. (Yes, we know that it’s ambitious to call it a “Suite” with just 2 products – please humor us. Also, there’s more in the pipeline…) Official press release here.
So what’s new? 1. We’ve done a nearly complete overhaul of our current product, now branded “Social Merchandising” and 2. We’re introducing a new product called “Social Purchase Sharing”.
Social Merchandising. We’ve made improvements top to bottom.
- Shoppers who open the widget but don’t personalize it by checking for friends will now see a range of other customers and their purchases designed to give the site that buzzing busy-store feeling and to encourage consideration and purchase of more items. (The goal is to address one of the big limitations of the shopping online: lots of stuff in the stores, but no people.) We’ve built a ranking engine that selects which customers and which items to show, ensuring the greatest relevance given limited data.
- We’ve made the value and process of personalizing the widget a lot more transparent to the user, so many more of those who open the widget will go the next step and personalize it to see their own friends in place of those the system picks. Underlying this is a simplification of the sharing rules to a vanilla Twitter-style “follow” model. (See our last post about the importance of simplicity when it comes to privacy and sharing.) We’ve also switched to delegated login for most of the friend list sources we support, including the newest Facebook protocols. (The short explanation: it’s better.)
- The widget now shows big, attractive product images throughout, so not only are shoppers seeing which of their friends also shop at that store, the purchases those friends made look particularly inviting. Good for cross-sell and order size improvement.
- The comment mechanism has been redone to both capture more input from buyers and to show it more visibly to shoppers.
- We’ve made significant enhancements to the guts of the system to provide greater speed and reliability. These include use of a Content Delivery Network as well as a range of server-side caching and summarizing strategies. The design point was to be able to support the largest ecommerce sites out there.
- We’ve added new tools for optimizing the button that calls up the widget. It doesn’t do stores any good to have a fabulous social merchandising tool if only a few shoppers use it. We now provide a range of more interactive button designs as well as tools for doing rotation tests (randomized A/B/C tests) of alternatives. In its initial use, we’ve already seen large engagement rate improvements.
In a nutshell: you have to see it. So here’s the first screen shot we’ve released:
Social Purchase Sharing. Our partner merchants have been telling us how valuable it is when a customer posts to their social network (most often Facebook and Twitter) about their purchase. So we’ve added a simple tool to significantly increase the amount of purchase sharing online stores can generate. It’s an overlay that appears on the order confirmation page right after a purchase and makes a clear, persuasive appeal to share. The permission obtained from the buyer is also used to power the Social Merchandising widget, so the “sharing” appears both on the social networks and on the store site itself. Here’s an example of the overlay – just picture it on top of your order confirmation page. (See also our blog post on “Like” vs. “Bought”.
The TurnTo Social Commerce Suite will be generally available to online retailers at the beginning of Q3, 2010. If you are in Chicago this week for the Internet Retailer show (IRCE), please come by booth #431 and we’ll give you a full demo. If you’d like more information on the thinking that went into these products, please have a look at the white paper we just released: Onsite Social for Online Commerce.
June 4, 2010 by George Eberstadt
In short: keep it simple.
In the first version of TurnTo, we were determined to set a gold standard on privacy control. We provided a multi-level model for authorizing purchase information sharing. We had forward and reverse models for specifying friend relationships. We let users create groups of friends then share with groups while excluding individuals or sub-groups. We provided time-based controls that let users specify review periods. And that’s just the stuff we implemented; our plans went even further.
You know how that movie ended: no one used these functions. And we’ve been stripping them out of the system one by one ever since.
Here’s what we learned: when it comes to sharing purchase information, there’s them that do, and there’s them that don’t, but there’s no one in the “I would if only I had more granular controls” group. The best way to serve your users is to keep the model very simple so that it’s obvious at first glance what sort of information sharing is going to happen. It’s OK to be very open, very restrictive, or anywhere in between, as long as the rules are obvious. Granular controls don’t help you increase your audience. At best they’re ignored, and at worst they cause confusion and bad feeling.
In contrast, Facebook has been moving in the opposite direction. They wanted to make their environment more open to enable functions that would be valuable to their members. But they felt a significant part of their membership might prefer the old, more restrictive model. So to keep everyone happy, they added granular privacy controls. “Everyone can have it just the way they want it.” But in trying to keep the old and the new at the same time, what used to be simple got complicated. And that hasn’t worked. People get their settings wrong and are surprised. People feel duped if their settings change without warning, or they feel coerced if pressed to change settings they were happy with. Or they feel burdened by having to learn complicated rules for something that used to be simple. Or they lose confidence in the system and back away. And what about those conditions where A meant to share only with B, but B shares everything with everyone, and A didn’t see that one coming? Now Facebook has added the Bandaid of bundling those granular controls into higher-level preferences. “You want it small, medium, or large? Don’t worry about the details.” That might help – we’ll see.
But if Facebook had asked us, we would have told them this: it’s OK to change, even radically. (You of all companies know that and have shown the guts to do it.) Decide on the basic approach to privacy you think is best for your users and your business. And throw everything else out. Some users will gripe about the changes (like they did when you introduced the news feed). But then they will see the wisdom of your new model, their behavior will adapt (some may share less, others more), and they will thank you thank you thank you for keeping it simple.
June 19, 2009 by George Eberstadt
I had fun during the Internet Retailer Conference this week chatting with Ina Steiner, Editor of the AuctionBytes blog. We covered a lot of topics in a short time. She has posted the conversation as a podcast. Enjoy.
February 4, 2009 by George Eberstadt
Here’s our second presentation from day 2 at the OnMedia conference. This one is a straight-up product demo and company backgrounder without the “theory” from yesterday. The TurnTo part runs from min. 36-46. (As with yesterday’s, we’ll swap in the individual video once we get it from the conference.)