Social Search Blog

From Beta to Launch

The journey from beta launch to full launch will be an interesting one. I really have no idea what the product will look like when it finally launches. The feedback already has been staggering. We have been able to add and change quite a few things immediately. And we have lots more already in development. I have a feeling though the initial launch product will look significantly different as we currently have a major redesign underway. More on that to follow…

Desktop Search and Community Answers

I read this New York Times article yesterday on Tacit software’s Illumio product “Tacit plans to start testing the service, called Illumio, next month. The service allows the user to mine the data on the computers of friends, business associates and others with shared interests on any subjects”. I paritciularly liked this line “[Illumio] is preparing to pick the brains of friends and colleagues for opinions and expertise”. The service uses a question-based system and a reverse auction to find the best answer source for user-generated questions. I think this is an interesting Social Search approach as it combines two techniques currently gaining popularity. Namely desktop search and a community ‘question and answer’ --similar to Yahoo’s Answers (which is very good by the way). I personally have had problems with both of the desktop search products I have tried and subsequently had to uninstall them both but I like the idea and think this method has promise. We shall see.

How We Define Social Search

Everyone defines Social Search differently so I’ll start out with my own definition. I define social search to be the intersection of Search and Social Networking.

Search is reasonably well understood. Basically as more information becomes digitized we need efficient tools to index and access this information. The Social Network part effectively takes the way we commune and interact in the real world and amplifies this by moving online.

The interesting thing is that the intersection of these two areas should be reasonably well defined. For example, if the space was ‘Search Social’ instead of ‘Social Search’ we would expect to use more efficient index and access tools to effect the way we socialize online. I think we already see that happening extensively. We use search tools to obtain composite digital profiles before and after we meet people. Also, think about some of the dating sites and the amount of information (increasingly psychographic) that is stored and filtered prior to ‘dating’

So if social search (or social networking search) followed the same definition rules, it would logically take the way we commune and interact offline, move this online and then apply this result to enable more efficient information indexing and retrieval for each of us.

So the logical result is that search becomes:

·         More communal (people-based)

·         More personal

This would lead to a set of potential social attributes where search results are based on:

1.     your previous searching history

2.     your stated preferences

3.     The search history of ‘trusted’ relationships

4.     The stated preferences of ‘trusted’ relationships

5.     The search history of individuals ‘similar’ you

6.     The stated preferences of individuals ‘similar’ to you

7.     The search history of an ‘affinity’ group that a person belongs to

8.     The stated preferences an ‘affinity’ group that a person belongs to

9.     The stated preferences of ‘experts’ in a particular area

10. The search history of ‘experts’ in a particular area

11. The stated preferences of a ’group of experts’ in a particular area

12. The search history a ’group of experts’ in a particular area

13. The universe of search history

14. The universe of stated preferences

Lower numbers are the more ‘personal’ search attributes, higher numbers are more ‘communal’ search attributes.

We would then use these social search attributes or dynamic pattern recognition to make search results increasingly ‘meaningful’ in each specific context.

As social search evolves, I think we will start to see models emerge that will have increasingly accurate weightings of these and other social search attributes. In addition the user will have the ability to adjust the attribute weightings to reflect context or individual preference. Finally social search algorithms will be able to adjust attribute based on the term that it being search for. For example, a search for ‘movies’ might be more heavily weighted to stated preferences, people with similar tastes and experts. However, a search for travel might skew towards stated preferences and trusted relationships. The former is a relatively low risk endeavor but the latter has greater consequences and therefore you may only trust your own preferences and the preferences of those individuals with whom you have a trusted relationship.

Here at Customerforce, we use all of these attributes. We also go further and filter all of the content based on either:

·         Affinity group

·         Trusted relationships

·         Your stated preferences

·         The Customerforce universe

The connection to a person’s preferences can be either active or passive and implicit or explicit.

In summary, it will be fascinating and exciting to see how social search evolves. We are committed to an extraordinarily high level of innovation in this nascent space. This will be a great place to document Customerforce’s approach to social search and over the next few weeks I’ll compare and contrast this with the approach of our peer group…

Launch

Customerforce.com goes into beta launch tomorrow. Everyone knows how gruelling launches are.

I got my first glimpse of the future today. I saw my default Customerforce search page. After using the new portal for a few weeks it had finally figured me out. When it presented my default search I was completely shocked. It had three of my favorite TV programs (well actually the only three I have time to watch) and a couple of my favorite artists. Right then I knew I had seen the future. Here's the press release.