Something happens, let’s call it “small bang”.

In order to “real time” search it; you have to know about it, right ? You can only search for something that you already know about.

But if you already know about it, that’s no longer real time. Hence,  real time search is an oxymoron.

We should rather name “the need to be on the top of something as soon as it occurs”,  real time listening or real time discovery.

And this opens several interesting challenges:

1) One can listen to so many sources at a given time.

Although one can talk/ tweet to millions in one click, listening, and even automated listening does not work the same way.

The consequences is that people and brand have to prioritize and organize who they are listening to.  That’s what Facebook Friends, Twitter followers, RSS lists are all about.

I remember attending one Robert Scoble talk where he was mentioning having a 45 mn advantage over CNN learning of events like plane crashes, earthquakes …

To do that, whether on purpose or not, he was relying on a network of people spread across the globe and counting the repetition of the same tweet to validate the information and judge its importance.  He also had trusted sources he could count on to double-check.  Even implicitly, there was an organization around his listening effort.

If let’s say, 95% of his contacts were in Silicon Valley, then he would have a hard time spotting a new event happening in Italy or Brazil.

For a brand, the implication is that  they have to map their target communities and align their listening efforts and resources to their strategies and to “who matters”.

They also have to give up the dream of real time listening to a few billion people real time.  The “more is better” approach of listening  is a root cause for “systemic intelligence failure” as Predident Obama just named it.

2) Any discovery process starts with a definition of what is supposed to be discovered:  “patterns”.

Pattern can be simple ones

• mentions of brand & product names, executive names, competitors … (yes/no, number, frequency)
• lack of mentions of brand name in their target communities… which is usually a much bigger issue for mid-size brands.

or more complex ones.

• expressions that have surfaced recently  in a given community
• changes in the structure of the community (let’s say top influencers start linking back to your competitor blog, following an outreach effort this competitor just performed)
• changes in the tonality of the conversations around your brand in a given community

Poincare (early 20th century French Physicist) once said: ” to find something without looking for it, one must have looked for it without finding it for a long time” (“Pour trouver sans chercher, il faut avoir beaucoup cherche sans trouver”).

Back to Robert Scoble example, he was not just looking for “anything” but discovering potential already known phenomenas. He also had to have an idea of  normal versus abnormal frequency of events.

This is the same for brand. They need to get an idea of what they should be searching for in order to instruct the discovery process.

This is best done with a non directed listening for a while; to get an idea of the events that are usually of interest and a generalization into “patterns” that can be looked for.

After community mapping and strategy definition, listening is a mandatory step before setting up any “automated” discovery.

They’re many solutions out there, and  no “one size fits all” one. Thus, ahead of starting your social media initiative, it’s key to develop a strategy aligned with your business objectives.

Roughly, the “find” technology powering social media marketing solutions fall in two different categories:

1. Keyword Search – you’re asked to input a few keywords and then an engine crawls for conversations that matches the search strings. Results are aggregated, sometimes attempts are made to identify the most influential people for this bag of keywords and to provide a social interpretation  of the results (graph analytics, mashup with social sites), which can be leverage for future searches.
2. RSS Readers -  the user is left finding places, people on his own and over time collects the people and places that make sense. He/She can dive in the conversation zooming with keywords on conversations of interest from people in target.

There are pros and cons in these two approaches.

Keyword Search

• +simple and easy to use, (at least from a first sight), it’s like Google and people have experience searching with keywords.
• + present extensive results. since you don’t focus first, you shouldn’t miss any conversation about a topic, brand, to the extend that your provider can go deep. (note Google itself only index 69% of the internet, deep search engine cost 100′s ok K’s - here is a very interesting study).
• - tend to produce a lot of noise. Even looking at presentations/demo  from providers  you usually can “feel” the noise.
• - very difficult to establish relationship with such approach, from one day to the other you end up getting to different authors
• - almost impossible to segment your audience. i.e: when communities using the same keywords differentiate on dimensions that are not easily captured  with keywords ( commercial versus community, expert versus novice, solution provider versus independent)
• +ideal if your brand is known and if your main focus is brand monitoring
• – More challenging if you research on a given topic (let’s say you want to understand how bailout is perceived in the experts blogs of the democratic side) or a mid range brand (high probability that you get nothing returned, or nothing more than what a google search would give: sporadic posts but no real conversations).

RSS Readers

• - this approach means you focus first so you will miss conversations. you can compensate by having an open search along with your RSS approach but this calls for “two level of reading” (the relevant and the rest)
• + you can calibrate your effort towards a limited set of people and places and you spend more time on each of the places and people you select. This gives you more chances to establish a relation that can be activated for word of mouth, recommendation, affiliation.
• + you can go deep. whether for influence measurement, trend identification. A few conversations about “sarah palin” become very visible in IT security blogs when her email was hacked, they were almost impossible to spot or discover in the open web.
• - it takes time and it adds an additional level of complexity.  One you have to identify who’s and where is relevant for you and only then you can start monitoring and engaging
• + you can segment your audience along your target market and business initiatives.
• this approach is better for research, relationship builiding and growing influence in a niche community. it is also better when you want to do lasershap monitoring on specific topic.