February 2012 Archives

Props to AT&T. Really.

(The useful title for this post is, "How to setup a temporary international data plan with AT&T.")

It's easy to jump on the anti-AT&T bandwagon. Typical complaints  usually center around crappy voice coverage, a crummy website, outrages fees, or unlimited data plans that are actually limited.  Me?  Admittedly, I've been an AT&T defender and apologist at times -- I cut my teeth as a programmer at the fabled Bell Labs, so it has a special place in my heart.  So, I've refrained from complaining too much.

International Data Roaming Plans Are Complicated
I've always been pained by egregious fees for data services when traveling out of the United States -- about $20/MB.  And choosing the *right* plan was difficult and confusing. So, I always signed up for the (most expensive) data plan at $199.99 for 800MB.  AT&T would prorate that for the number of days you actually used the plan within a billing cycle.  So, if you traveled for 3 days, you'd get about 80MB of data (3 days * 800MB/30 days) for about $20 (3 days * $199.99/30 days).  And if you went over 80MB, you could always tack on more days at about 27MB/day at $7/day, after you returned.

Whew.  That was still a bit complicated.  And, even more so if your travels crossed a billing cycle.  But, if you carefully managed your data utilization (painfully using "airplane mode" or turning off cellular service), you could get by.  It was worth it for the cost conscience traveler.

New and Improved! Monthly Usage Prorated.  Data Usage Not.  Hooray!

On my recent trip to Mexico, things got surprisingly better.  A lot better, especially for short term travels.  I found that AT&T changed its policies so that they will still prorate your service based on the number of days you are traveling abroad.  BUT, they don't prorate the amount of data you get.  For the $199.99/month plan, you'd get a budget of the full 800MB, even though you only paid for 3 days of service.  So, for $20, I was able to get (all most) all the data I wanted without worrying too much about going over my allotment.  (Note that I wasn't able to watch unlimited Netflix movies but you get the idea.)

Sure, I'd like to have "unlimited" data wherever I go, whenever I want.  But,  turning on an international plan for the number of days I'm traveling  for up to 800MB is a pretty easy and effective solution in the mean time.

Good job AT&T, with this policy change!

Caveat:  I'll be singing a new tune and writing an update to complain if my bill this month doesn't work out this way!

97 Failures Is the Key to Success.


As noted in a few blog posts, when Ben and I co-founded MerchantCircle, we were met head on with skepticism...."You know nothing about media and local...or the Internet."... "Look at all the dead bodies pursuing the holy grail of getting to small businesses.  How could you possibly succeed?"..."You guys are naive and crazy....but, possibly, if it all works, brilliant!"

All of this was true.  But we believed in the space.  We believed we had what it took to be successful.  And, while we didn't have the obvious answer to what was going to work, we believed we'd be able to figure it out.  Because....

We knew we could build a machine to figure it out.  We'd use two tools -- statistics and an experimental methodology.  And luck.

Hopes and Dreams Failure.

Figuring out how to acquire merchants as customer at low cost and scale was a key concern. Our initial hunch was "build a social network of small businesses and go viral."  It was the pitch  when we raised the series A -- a hopes a dream story.  We raised on a Powerpoint presentation and had no really proof points that this would work.

And, it didn't.

New Big Idea.

So we brainstormed over other ways to get customers.  We cycled through a bunch of ideas.  Some were reasonable but hard to proof.  For example, we hypothesized that, even though it was 2005, there was still an opportunity build traffic through organic search.  This was a fairly technical exercise.  And, we knew building traffic through SEO was a long game, requiring patience.  And it might not pan out.  We rolled the dice anyway and went down this path.

Fortunately, this worked out in the long run. We acquired boat loads of merchant customers organically over the years with an added bonus of large consumer traffic that converted into significant ad revenue.

Crazy Ideas.

chicken.pngBecause SEO is a long game with risk, we started to think about other ideas.  Some of it was obvious.  Like affiliate marketing.  SEM.  Email campaigns.  Direct mail.  Telemarketing.  Some it was absurd.  Radio advertising.  Creating coupons on the fly for local businesses.  Dressing up as a singing chicken delivering a  telegram.  (I can't remember if this was Mark, Daniel, or Ben.)  Give away fishbowls to local merchants (Don't ask).  And a bunch of others that were even more crazy.  (DM me if you want to know.)

We probably cycled through 500 ideas.  Nothing was too absurd.

100 Experiments. 97 Failures.  3 Successes.

We then formulated some hypotheses.  Which ideas were truly absurd?  Which might work?  We weren't certain any of them would work!  Ultimately, we decided to execute on 100 experiments to test which ideas were good.  We needed to build a machine and culture where we could quickly setup and execute our experiments.  For each one, we wanted to do as little work/implementation as possible and spend less than $100.  We figured it was no use spending a lot of time or money on a failure.   Fail fast, in modern parlance.

Our goal was to find 3 approaches that worked.  We figured that three of them would.  But the problem was -- it was kinda hard to tell which ones did.  It was easy to get fooled by false positives and false negatives.  Or maybe sometimes we weren't so rigorous statistically.  Or someone wanted an outcome so badly he tortured the data.  If I never hear a faulty argument about the importance of the  R-Squared being greater than 80%, I won't shed a tear.  :-)

Regardless, when we saw a glimmer of hope of an experiment with a good result, we ran with it.  There were probably 10 of these.  We went deeper.  Spending $1000 each.  When we whittled it down to three, we spent around $10,000 each.

So, all in the cost to run all the tests to get to something that worked was less than a $100,000.  It felt like a pretty good use of money.

3 Successes?  Maybe.  One Exit?  Definitely.

Did we really choose three winners?  Were there better ideas that we falsely rejected?  It's kind of hard to tell.  But we were able to run with 3, in which at least one of the three worked well.

Coupling these with a working SEO strategy, we were able to successfully bootstrap our customer acquisition efforts.  10,000 other steps after executing these three ideas, we guided the company to a successful exit.


Fast forward to 2012.  Would this strategy would today?  From one perspective, it seems it could.  I just read how PunchTab cycled through a bunch of widely different ideas before settling down on the current idea.  It also seems to be vaguely compatible with the idea of building a minimum viable product within the context of a lean startup.  However, our experiments were narrowly focused around specific customer acquisition goals within a set strategy to win on the local Internet.  We weren't experimenting to find the "big idea." Second, we weren't so interested in iterating and refining an idea after an experiment.  Either it was working or not, and we were moving on if not.  Last the cost of doing an 100 experiments in search of a success was not without cost.  At $100,000, if you are early stage, this might not be affordable. 

If you are boot strapping, maybe this doesn't work because of the capital requirements.  You can't afford to do 100 experiments.  With 97 failures.  And, raising enough money at this early stage might be difficult.

But for MerchantCircle, this all worked out fine.

Why Local Sucks. Or Maybe Not.

Chris Devore, a long-time friend, writes an interesting blog post on why you shouldn't do  a startup in the local space.  Or why it was a bad idea a few years back.  I've exchanged a few comments on facebook with him and Andy in a friendly spar.  Ben Smith weighs in and so does Jason Culverhouse with a comment on Chris' blog.

In short, local sucks if you are an Internet entrepreneur.

And yet, there have been a few  big successes (OpenTable and Groupon), a few moderate ones (Topix, urbanspoon, and  MerchantCircle -- a company I co-founded), and a few that I'm pretty sure are (going to be) hits. Peixe Urbano, for example.  Maybe it depends on the metrics you use to define success and therefore YMMV depending on your viewpoint.

My conclusion: There was and still is money at the end of the local road.  It's a matter of taking the right path and choosing the right angle of attack. At MerchantCircle we believed similarly to Judy's Book but executed differently. 

Merchants First.  We knew our customers were the small local businesses.  While others went down the path of on-boarding a large audience first, we knew where the money was.  The local business owners.  While we never grew a ginormous visitor base, getting to the mom and pop operators was more important.

Scale Free Customer Acquisition.  We believed that low cost customer acquisition was important to success.  High cost of customer acquisition and high churn rates would be death.  Unlike Groupon and ReachLocal, we were fundamentally against building a large sales force.  We did not think that would scale.   Knowing that the merchants were our targets, our goal was to acquire a customer at $1 a pop.  We spent countless of hours experimenting and deploying different tactics to get in front of customers at low cost and at scale.  Automation and software were key.  Special kudos to Doug Kilponen for being the business driver behind all of this.

Hit the Hinterlands. 
Unlike many, we did not focus on the big cities (San Francisco, New York, Los Angeles, Chicago, and Seattle).  We actually feared that successes in any single geography (big or small) would yield false positives.  Tailoring a special plan of attack for each city or zipcode would not scale.  Again, automation and software  allowed us to get access into 20,000 cities, 40,000 zipcodes, and 15 million businesses, inexpensively, quickly, and effectively.

Go Lean.  Before the Lean Startup movement, we decided build a company with few people.  We called that lean.  It was over a few beers in 2004, a friend was on pedantic rant:  "People suck.  You should build a company and product without people.  A product should market itself and sell itself.  Everything should be driven out of engineering."   I took it to heart (even though the person revealed later that he didn't quite mean it).  We set out drive everything out of product and engineering -- a product that marketed itself, sold itself, and was fully self service.  We had no sales people, no marketing folks, no customer support, no separate IT organization, and no QA group.  Eventually of course this changed.  But the company never grew much larger than 10 people in the first two years and it was smaller than 20 nearly 5 years later.  Because labor was the most expensive cost to the company, the small employee foot print was paramount to conserving cash.  Further, the small team size made communication much more efficient.


Many of our assumptions were the same as Judy's Book. But we acted on them differently.  For example, while "turning on many cities simultaneously" didn't work for Judy's Book, it was important to our success.  Execution is/was everything. 

And what about MerchantCircle in the end?  Well, it would have been nice to have achieved the success of OPEN or GRPN.  While we sold the company to Reply, I still have hope that, as part of a larger family, MerchantCircle will make a massive positive impact on local businesses and be part of a great success.