Nov 182007

In Search of Automated Relevance

In Search of Automated Relevance

A bunch of us had a free form meeting last week that started out as an Email Summit focused on protocols and ended up, as Brad put it, with us rolling around in the mud of a much broader and amorphous Messaging Summit.  The participants (and some of their posts on the subject) in addition to me were Fred Wilson (pre, post), Brad FeldPhil Hollows, Tom Evslin (pre, post), and Jeff Pulver (pre, post).  And the discussion to some extent was inspired by and commented on Saul Hansell’s article in the New York Times about “Inbox 2.0″ and how Yahoo, Google, and others are trying to make email a more relevant application in today’s world; and Chad Lorenz’s article in Slate called “The Death of Email” (this must be the 923rd article with that headline in the last 36 months) which talks about how email is transitioning to a key part of the online communications mix instead of the epicenter of online communications.

Ok, phew, that’s all the background. 

With everyone else’s commentary on this subject already logged, most of which I agree with, I’ll add a different $0.02.  The buzzword of the day in email marketing is “relevance.”  So why can’t anyone figure out how to make an email client, or any messaging platform for that matter, that starts with that as the premise, even for 1:1 communications?  I think about messaging relevance from two perspectives:  the content, and the channel.

Content.
  In terms of the content of a message, I think of relevance as the combination of Relationship and Context.  The relationship is all about my connection to you.  Are you a friend, a friend of a friend, or someone I don’t know that’s trying to burrow your way onto my agenda for the day?  Are you a business that I know and trust, are you a carefully screened and targeted offer coming from an affiliate of a business I trust, or are you a spammer? 

But as important as the relationship is to the relevance of your message to me, the context is equally important.  Let’s take Brad as an example.  I know him in two distinct contexts:  as one of my venture investors, and as an occasional running partner.  A message from Brad (a trusted relationship) means very different things to me depending on its context.  One might be much more relevant than the other at any moment in my life.

Channel.  The channel through which I send or receive a message has an increasing amount to do with relevance as well.  As with content, I think of channel relevance as the combination of two things -  device, and technology.  For me, the device is limited to three things, two with heavy overlap.  The first is a fixed phone line – work or home (I still think cell service in this country leaves a lot to be desired).  The second is a mobile device, which could mean voice but could also mean data.  The third is a computer, whether desktop or laptop.  In terms of technology, the list is growing by the day.  Voice call, email, IM, Skype, text message, social network messaging, and on and on.

So what  do I mean about channel relevance?  Sometimes, I want to send a message by email from my smartphone.  Sometimes I want to send a text message.  Sometimes I want to make a phone call or just leave a voicemail.  Sometimes I even want to blog or Twitter.  I have yet to desire to send a message in Facebook, but I do sometimes via LinkedIn, so I’m sure I’ll get there.  Same goes for the receiving side.  Sometimes I want to read an email on my handheld.  Sometimes a text message does the job, etc.  Which channel and device I am interested in depends to some extent on the content of the message, per above, but sometimes it depends on what I’m doing and where I am.

So what?  Starting to feel complex?  It should be.  It is.  We all adjusted nicely when we added email to our lives 10 years ago.  It added some communication overhead, but it took the place of some long form paper letters and some phone calls as well.  Now that we seem to be adding new messaging channels every couple weeks, it’s becoming increasingly difficult to get the relevance right.  Overlaying Content (Relationship and Context) with Channel (Device and Technology) creates a matrix that’s very difficult to navigate.

How do we get to a better place?  Technology has to step in and save the day here.  One of the big conclusions from our meeting was that no users care about or even know about the protocol – they just care about the client they interact with.  Where’s the ultra flexible client that allows me to combine all these different channels, on different devices?  Not a one-size-fits-all unified messaging service, but something that I can direct as I see fit?  There are glimmers of hope out there – Gmail integrating IM and email…Simulscribe letting me read my voicemail as an email…Twitter allowing me to input via email, SMS, or web…even good old eFax emailing me a fax – but these just deal with two or three cells in an n-dimensional matrix.

As our CTO Andy Sautins says, software can do anything if it’s designed thoughtfully and if you have enough talent and time to write and test it.  So I believe this “messaging client panacea” could exist if someone put his or her mind to it.  One of the big questions I have about this software is whether or not relevance can be automated, to borrow a phrase from Stephanie Miller, our head of consulting.  Sure, there is a ton of data to mine – but is there ever enough?  Can a piece of software figure out on its own that I want to get a message from Brad about “running” (whatever channel it comes in on) as a text message on my smartphone if we’re talking about running together the next day, but otherwise as an RSS feed in the same folder as the posts from his running blog, but a voicemail from Brad about “running the company” (again, regardless of how he sends it) as an email automatically sorted to the top of my inbox?  Or do I have to undertake an unmanageable amount of preference setting to get the software to behave the way I want it to behave?  And oh by the way, should Brad have any say over how I receive the message, or do I have all the control?  And does the latter question depend on whether Brad is a person or a company?

What does this mean for marketers?  That’s the $64,000 question.  I’m not sure if truly Automated Relevance is even an option today, but marketers can do their best to optimize all four components of my relevance equation:  content via relationship and context, and channel via device and technology.  A cocktail of permission, deep behavioral analysis, segmentation, smart targeting, and a simple but robust preference center probably gets you close enough.  Better software that works across channels with built-in analytics – and a properly sized and whip smart marketing team – should get you the rest of the way there.  But technology and practices are both a ways off from truly automated relevance today.

I hope this hasn’t been too much rolling around in the mud for you.  All thoughts and comments (into my fancy new commenting system, Intense Debate) are welcome!

Filed under: Email, Technology

Tags: , ,

  • http://intensedebate.com/people/bfeld bfeld

    Good mud. I've had a bunch of interesting inbound from people. The consistent theme is that they are only focusing on one channel; therefore, you insight about channels is an important one that I think most people are missing.

    There's a "meta-problem" here that – when you extend this to messaging – and then do 1:1, 1:many, and many:many – your brain starts to hurt a lot.

  • http://intensedebate.com/people/jpmorgan jpmorgan

    What are your thoughts on Xobni? Many people seem to think their product is very cool and makes email way more useful. Of course, I can't tell too much from their homepage.

  • http://intensedebate.com/people/matt_blumber478 matt_blumber478

    Xobni – I like the concept, but it kept crashing my Outlook, so I disabled it for the time being. But I think things like this ultimately need to be baked into the core piece of software to really have their impact.

  • http://intensedebate.com/people/deva_hazari2084 deva_hazari2084

    Love your post, Matt. You lay out a lot of the reasons that make this challenge both really interesting and really complex/tough. Some of the first ideas we had at ClearContext a few years ago were around integrating other channels. However, in our research on that, we realized that neither the data structures not user interfaces existed to really handle even just email communications well, much less pulling in lots of other channels. So we've concentrated on first focusing the existing information within the email client (Outlook first) using elements of both context and relationship. Once a layer of structure is applied to existing data, the foundation is in place for much more interesting things to be done. That's what's coming next. Here's some more on this, you'll probably find the blog post before it relevant as well: http://www.emaildashboard.com/2007/11/the-four-as

  • http://www.fickennacktsex.com/ erotik
  • http://blog.gist.com T.A. McCann

    Great post. We are taking a stab at some of these concepts at http://www.gist.com. We are using the inbox as a starting point for both the "list" of people and companies and we have started with a simple, though evolving, algorithm that determines importance. We have a long way to go and not sure if we will ever go down the full "email client" direction, but when it comes to aggregation, focus and prioritization, we think we are doing some pretty cool stuff.

    Come check it out and let me know what you think.

    T.A. McCann (CEO and Founder of http://www.gist.com)

css.php