Patrick McKenzie on SEO & AdWords for Bingo Card Creator

 

Patrick McKenzie is the solo founder of Bingo Card Creator. BCC makes bingo cards for elementary schoolteachers. After achieving traction trough scalable use of SEO and AdWords, Patrick was able to quit his job and start selling software full time. He explains the exact strategies he uses and how they are repeatable in a variety of niches.

For more, check out the Traction Book site.

Update: Patrick has also outlined the interview on his blog. He's also answering questions on HN.


For more interviews, check out the Traction Book site.


Interview Transcript (transcribed by Patrick--thank you!)

Good morning Gabe.  I'm Patrick McKenzie and I'm the founder of Bingo Card Creator.

Gabe: For those of you who don't know -- I guess people who are not on Hacker News -- Patrick is a solo founder.  Bingo Card Creator pretty much does what it says -- is that fair?

Patrick: That's correct.  I wrote it four years ago, originally as a Java Swing app.  These days the main product is a Ruby on Rails web application.  It is mostly used by teachers and occasionally by other folks in the social services industry.

Gabe: Unlike most of my other interviews, you're a solo founder, right?  You just quit your job to do this full time, and it's enough to sustain yourself, basically.

Patrick: Right.  I'm not exactly rolling in it.  I publish stats -- it is on the order of $30,000 in profits this year [note: estimate for 2010].

Gabe: But nevertheless, people like to hear that kind of thing, because they feel they could also start something and quit their jobs.  

Patrick: Right.

Gabe: You started slow, since you were part time.  I'd like to run down what the whole operation of Bingo Card Creator looks like.  If you could just walk through the whole universe of your sites, that would be great.

Patrick: Well, earlier we were discussing SEO strategy.  The obvious place to start is that there is one main website, bingocardcreator.com .  It is a Ruby on Rails application.  The website has been up for about four years [since June 2006], the Ruby on Rails portion for about 3 years [since October 2007].  The core of the site was originally a custom CMS written in Rails.  [It takes as input a list of words and creates PDFs, GIFs, and web pages featuring those words as a bingo activity.  For example, see http://www.bingocardcreator.com/bingo-cards/english-language/homonyms ]

The reason I originally created the CMS was because I wanted to target the Long Tail of traffic in the bingo card search space.  The core need for a teacher who wants to play bingo with her class tomorrow is that she needs one randomized bingo card per student, on the topic which she wants to teach.  For example, if you're doing an American History lesson tomorrow about the names of all the presidents for your fifth graders, she will go into Google and type [american presidents bingo cards], and if you typed that in four years ago, you would find *nothing*.  

So, I wanted to have an answer for someone who was searching for [american presidents bingo cards], but I didn't want to have to create all the pages manually.  I actually tried it once and it took me about 1 hour for one page.  You can't create content at scale making HTML pages in Notepad by hand, so I wrote the CMS.  The actual content creation is mostly farmed out to freelancers.

I currently have about 800 pages so for "something bingo cards" you will find about 800 activities on my site -- everything from American presidents to owls of East Asia to addition exercises.  Basically, I want to cover every bingo activity that an elementary schoolteacher could possibly want to run a lesson with.

Most of those are Long Tails lessons.  Long Tail is a reference to the 
Chris Anderson book of the same name -- it is probably the most influential book I've ever read about Internet marketing, and I recommend that everyone read it.  In addition to the Long Tail, there is the fat head of the distribution.  For example, Halloween bingo cards.  Halloween is a holiday that kids like to have a game to play on, but it is a (mostly) secular holiday so kids will actually be in class, so it causes a traffic spike that goes like [Patrick draws a vertical asymptote with his hand] in the last week of October.

It was worth doing a bit more to capture the Halloween searches.  In my case, I created a mini-site called halloweenbingocards.net (I couldn't get the .com, unfortunately).  Halloween Bingo Cards .net is a very simple site which says "You are looking for [halloween bingo cards].  Here are freaking Halloween bingo cards."  This acts as a landing page to get people into the free trial version of my service, and hopefully some of them actually go on to convert and pay me money.  This site will be seen by on the order of 100,000 people in one week in October.  [Patrick notes: That number is inaccurate.  In 2009, it was closer to 30,000.  Sorry, quoting stats from memory.]  

Christmas sees a bit less than that.  There is a zipf distribution [Patrick notes: the technical term for the type of distribution now popularly referred to as the long tail] among holidays.  Halloween bingo cards is at the top of the head of the distribution for holiday bingo cards, Christmas a bit further down the tail.  One of the key concepts from the Long Tail is that there are tails within tails: for example, there is a tail of genres among bingo cards and, stepping out one level of abstraction, bingo cards are themselves way down the tail of the data that Google gets to look at.  It is tails, tails as far as the eye can see.

I also run a blog.  The blog is mostly not for business benefit -- it helps to keep me sane.  I enjoy sharing ideas with people and getting feedback on them.  The act of writing also helps me to crystallize my thinking.  A couple of times I've been writing up my "great new idea" on the blog and thought "Hmm, as a writer, what I'm saying here sounds like total BS.  That is because what I am doing *is* BS.  Why am I doing this?"  I've saved myself a lot of programming time this way.

Gabe: So how many of these microsites do you have?

Patrick: That's a good question.  When I started the project I pulled 14 site ideas off of a SQL query and farmed them out to freelancers.  As to how many are functional at the moment, hmm, six or seven off the top of my head.  [Patrick notes: five and two halves.]  Eventually I hope to finish the others and also add a few more targetting head search terms, such as [wedding shower bingo cards] and [bridal shower bingo cards].  These see high search volumes year round.

One of the things that I totally didn't realize when starting out is that the bingo card market is highly seasonal.  You can tell that if I'm getting 100,000 [sic] people in the last week of October every single year that it causes a visible difference on my sales graph.  Unfortunately, there is not a visible difference on the "I would like to eat ramen noodles" graph, so I would like to smooth that out.

Gabe: So how much traffic do the microsites attract as opposed to the main site?

Patrick: That's a good question.  Typical traffic levels on the main site are on the order of 2,000 visits on a slow day (weekend) spiking up to 3,000 on a typical weekday or  4,000 to 5,000 on a very busy weekday, for example, right before Valentine's Day or St. Patrick's Day.  

The Christmas bingo card site would get about 25 people for today.

Gabe: So what is the aggregate for the whole year?  

Patrick: For the Christmas site, there would probably be a total of about 60,000 people in December.  [Patrick notes: I misremembered.  60,000 pageviews, but only 20,000 people.]

Gabe: In the aggregate then, the microsites yield more than the main site?

Patrick: In aggregate, they're probably on the order of how many people visit the main site.  The value per visitor is a fraction of the ones on the main site.  The reason is that my segmentation of the free users and the paid users has been, for the longest time, "If you want to create less than 15 cards, you can use the free trial.  If you want to create more, you must purchase the software."  For example, if you have an American classroom with 22.7 kids in it, you have to pay money.  

When a parent -- and when I say parent, I mean mother, because 95% of my customers are women -- is looking for bingo cards to play with her 2.5 kids and husband or the people coming over for Christmas dinner, she'll find ChristmasBingoCards.com and say "Wow, this is exactly what I needed!" and hopefully mention it on her blog.  A teacher finding the site would see that it doesn't do exactly what they need but it accurately promises that they would accomplish their goal for tomorrow if they bought the software.

It works out that a searcher on the main site is worth ten times what a searcher on a microsite is.

Gabe: You still find the microsites are worth it, given that you're farming out the work.

Patrick: Right.  I just did the math last night for Easter.  Easter was nowhere near my most successful site, but it sold maybe $650 worth of software this year.  Putting up the site once took $100 for content creation -- the server is free at the margin and I used an open source template.  Keeping the site up costs $30 a year because I own the .com, .net, and .org -- always do that, by the way, if you have the opportunity.  

Unless something grieviously changes in the Christian religion, there will still be an Easter next year, [Patrick elaborates: which means site will still be equally effective or more effective next year, without generating substantial marginal costs over what I've already paid.]

Gabe: Why do you want to own the .com/.net/.org?  Is that to prevent competition for the keyword value of the domains?

Patrick: Yep, that is the main reason.  Owning a domain at the registration fee -- that is what you pay if you get something new from GoDaddy or similar, $6 to $9 per domain per year depending on who you buy from -- is so little money that you shouldn't even spend time thinking about it.  Buy it, set it to autorenew, forget about it.

Gabe: So the value of the microsites is the keyword value of the URL, and that is more valuable to you than putting the site on a subdirectory of your main site, such as (NOT A LINK) www.bingocardcreator.com/easter ?

Patrick: Right.  I don't want to say that microsites are the right solution for everyone.  The problem with microsites is that they cause you to split your domain equity and link equity.  The basic idea behind PageRank (the core of Google's original search ranking algorithm) is that any link to a page is a vote that that page and site is trusted and should appear higher in the search engine rankings.  Separate from PageRank is what is called the exact match domain bonus -- if your domain (.net, .com, .org, or country-level TLD like .co.uk if the searcher is in the UK) exactly matches -- and I mean character-for-character match when you strip out the spaces -- you get a super bonus.

For example, if you registered patricksfunnydomain.com , and tried to rank for [Easter Bingo Cards], you might need hundreds of links or dozens of links until you're competitive for that search result.  f I was putting it on Bingo Card Creator, which has hundreds of links, it would probably rank on the top page but near the bottom of it, until I built a significant amount of links directly to that page.  The exact match bonus causes easterbingocards.com to rank in the top three with a handful of links in a matter of weeks -- a link from my blog, a few from directories, that is all it takes really.   (Nota bene: it might not be in the top 3 when you look.)

Gabe: So are you worried about some sort of link-farm penalty as a consequence of doing all these microsites?

Patrick: Some people who create microsites try their level best to hide the fact that they're part of the same Internet business empire.  I go totally in the opposite direction -- my sites are all registered under my name, the About page prominently says it was written by me, they link prominently to my site, the subheading says "from Bingo Card Creator."

The boogyman for SEOs is Matt Cutts, a Google engineer who is in charge of search quality and also outreach to the webmaster community.  If you hear SEOs speaking of their darkest secrets, they'll look over their shoulder for Matt Cutts.  I am totally OK with Matt Cutts looking at my sites.  They're totally above board: if you're looking for Halloween bingo cards, then my site on them gives you exactly what you are looking for.

Gabe: So they pass the Matt Cutts Test.  *laughs*

Patrick: Right.  They pass the Google Remote Quality Rater guidelines.  Google employs or used to employ about 10,000 people who would manually check what sites were ranking for real searches on Google and grade them for quality.  Their guidelines have been leaked.  For example, if you're searching for [toyota motors], you would expect #1 to be the official site for Toyota Motors corporation because it is probably a navigational search.  The rest of the sites should be about Toyota cars, as opposed to the Lincoln Nebraska Toyota dealer if you're not from Lincoln Nebraska.  Certainly, you shouldn't see a spam site.  

If you look at the guidelines, which are sort of extensive -- about 100 pages long -- I'm pretty sure my business is on the up and up side.  

Gabe: So how much of your traffic comes from organic search?

Patrick: About 50% of my traffic and 70% of my sales comes from organic search.  The lion's share is from Google: Google is synonymous with navigation on the Internet these days, particularly in the English-speaking world.  Microsoft and Yahoo only add up to about 15% of my organic traffic to Google's 85%.

Gabe: So aside from organic search, you mostly get AdWords and direct traffic?

Patrick: Mostly AdWords, particularly the AdWords content network.  Many people tell me they haven't had much success with the Content Network, but I'm an odd duck in that I haven't had much success with the search network.    People who Google [halloween bingo cards] will find me on the left side of the screen (i.e. the organic search results) but not find me on the right side of the screen (i.e. AdWords) because I haven't been able to do that in a cost-effective fashion.  

Are you familiar with Demand Media?

Gabe: Yep.

Patrick: Demand Media is making major headway on the Internet with a later, more sophisticated version of about.com.  They write content about topics people are searching for and then monetize it with AdSense ads.  For example, they'll have a page on eHow about How to Write Bingo Cards and it will have four ads on it: one from me, one from a direct competitor, and two from people who do not know how to configure their AdWords campaigns properly.  

Since Demand Media content is less than inspiring, people who read the article think "That's great, but I don't have actual bingo cards.  Aha, this ad promises actual bingo cards.  I will click on the ad."  I have mixed feelings about it this but I subsidize their business model to the tune of hundreds or thousands of dollars a year, but it works out to a heck of an ROI for me.

Gabe: How much do you pay for cost per click for these ads, and how to they convert?

Patrick: My CPC is 3 to 5 cents, on average.

Gabe: Wow, that's low.  

Patrick: Right.  If you've worked in Internet advertising, you've heard this term "remnant inventory".  If Upton Sinclair were writing about the Internet, it is what he'd write about: it is what is left on the floor after you've carted away all the parts of the cow that people would actually want to eat.  All you can do is shred it up and feed it to other cows.  

Most of the people who are bidding the good keywords to heights where it is not beneficial for me to be bidding on cannot make money from the housewife in Licoln Nebraska who is on a Demand Media page looking to make bingo cards.  Happily, that is the only person in the world I *can* make money from, so I get the click for like 3 to 5 cents.  Typically, between 18 to 20% will sign up to the free trial on the landing page, and 2% of trial users will convert, so that works out to a CPA of about $10 ~ $12 to generate a sale worth $30 for a product which has no marginal cost.

Well, $1 for the Paypal fee.  $30 - $1 - $12 = $17 margin per sale, so I'm happy to do that as long as Google has traffic for me.

Gabe: So you're not doing anything special, you're just bidding for keywords other people are not bidding for?

Patrick: Other people might be interested in the keywords or pages I'm placing ads on, but they can't bid as high as I can -- and when I say as high as I can, I mean as high as five cents.  There are many business models for which even five cents is too much to pay -- for example, if my conversion rates were such that my cost per action (CPA) were $40 for a $30 product, I would not be able to sustain an AdWords campaign.  

[Patrick notes: Very savvy Internet marketers would say "Aha, but you might have more off the back end -- selling additional products/services to the email list you've generated from previous customers."  I have no back end, but this is something you'll want to keep in mind.]

I have a lot of competitors who perhaps out of a lack of savvy or desire for a better competitive position lower their price -- if he charges $30, well, I'll charge $20.  If you have fixed costs of $1 for Paypal and have a $15 CPA on your ads, you can't really afford to compete with me, because I could spike it to $20 CPA any time I want and then you're out of business.  Don't compete on price, and *especially* don't compete on price when doing CPC advertising.  

Google will love you if you do bidding wars, though.  

Gabe: You talk a lot about A/B testing.  How high did your CPA start at, how did it come down, and was that all a result of A/B testing?

Patrick: Well, I've been doing AdWords for a long time, since about August 2006.  At the time, my product cost $25 and the cost per action was... $25, sometimes as much as $30, but I was just spending $1 a day trying to find the magic code to make AdWords work.  I kept going, refined my creatives (the ad text), worked on keywords, etc.  I got to profitability, in the sense that I was spending $90 a month and making $100, but if that took 5 hours of experimentation a month that clearly wouldn't sustain my desired wage.  

Eventually, Google Conversion Optimizer came out.  Instead of bidding on a per-keyword basis, you bid on a per-conversion basis, telling Google "OK, I'll pay you up to $15 for a sale."  Actually, that isn't how I structure it.  "I'll pay you 25 cents for someone who downloads my trial."  Google's magic algorithms will go try your ads on all sorts of websites and keywords, and only continue the ads on sites that appear to work.  

For example, if there is a spam site with a million people clicking on it but not purchasing, Conversion Optimizer will quickly decide "OK, this site sends crud traffic, I won't pay for money on it".  On the other hand, if you put an ad on about.com -- which always works out well for me -- you'll have that ad seen by real English teachers searching for real English lessons that they want real English bingo cards to run.  So Google will say "Alright, we'll put more ads on about.com."

After starting to use Conversion Optimizer, my AdWords campaign got rather shockingly more effective.  It was because of that and because I was an early adopter that I got a call from Google about doing a case study with them -- one of the weirdest moments of my life.  "Can WE use YOU as an example of a successful business on the Internet?"  WHOA.

I really, seriously got into A/B testing sometime after starting AdWords.  I wrote my own A/B testing software for Rails: A/Bingo, look for it on the Internet (http://www.bingocardcreator.com/abingo) or if you want a twitterable URL http://abingo.org (Patrick notes: remember what I said about domains costing $6 a year?).  

One of the nice things about A/B testing is that if you have a business that works even a little, then A/B testing makes it better.  If my business successfully drives folks from AdWords to trial to purchase than A/B testing any part of that pathway makes my business better.

Gabe: I've found in my experience with A/B testing that sometimes the smallest changes have the largest effects.  Can you talk about your experiences?

Patrick: That's true.  The first thing anyone learns in A/B testing is that everything you know is wrong.  Even changes which you think are clearly, obviously slam dunks will sometimes turn out to negatively impact conversions -- your opinions are figments of your imagination, reality does not agree.

One example: There is this thing in copywriting called "the scent".  You're supposed to "maintain the scent": don't send people to pages which would cause them to perceive a discontinuity in their interaction with your business.  So, using an example from my microsites, if you're on Halloween bingo cards .net I have to pass you over to the main site to make the conversion to the trial.  You can look at the site, but it has a black and orange color scheme, with black cats and pumpkins.  Then you go to bingocardcreator.com, which was created by a very talented young lady in India to be blue and inviting to a non-technical audience.

So I thought "OK, I'll take a graphical element from the Halloween site and use it as the H1. The text is orange, the background is the same, you can clearly see that you have not accidentally been redirected to some random site on the Internet."  I thought this would work very, very well, since everybody says that maintaining the scent is very, very effective.  Not so much: conversions down by 20%.  (Patrick notes: Actually, less.  http://www.bingocardcreator.com/abingo/results )  So, that test got reverted.

In terms of things that are very effective: if you look at a company like Netflix, and think "Wow, this is a great business model.  They spent SO much money on the Netflix prize.  They have hundreds of thousands of man hours invested into making their service the best possible thing for movies."  However, probably the single biggest lever in the Netflix business model is... the button on their free trial page.  It is absolutely insane how much difference buttons make to your business.  Test the text, the color, the positioning, the button-ness.

For example, just this month I spent $400 on new buttons and thought, clearly, this is going to work well because these are much prettier than my old buttons.  Apparently folks saw it as less clickable than the old button, so it got clicked on less.  Set fire to that $400, try something new next month.

Gabe: So talk a bit about how you manage your freelancers.

Patrick: I've had about 15 freelancers over the years, most of them folks I met over the Internet.  Some I met on freelancing sites, and after my blog had a bit of a following I would just post an article saying "I need somebody to do this.  Do you have a nephew who would like some money?"  That works fairly well.

My bingo cards are created by a lady that lives in New Mexico.  I put out a call for freelancers on my blog and gave five folks a (paid) trial project, then picked the one who produced the best work.  She is a teacher, so she already has to do the work: she just copy/pastes the vocab list for the week into my CMS, and then tells me that she is done so that I can send the check.  I pay her $100 a month for 30 word lists.  

Management is all via email.  I know a lot of folks use Basecamp, but I focus on small, easily checked tasks.  If someone sends me an email saying "I'm done", I check the attachment quickly and say "Yep, it is done."  Since I cultivate long-term relationships with my freelancers it is usually not a question of someone cutting corners on work.  I'll send an email back at about 3 AM in the morning saying "Thanks very much for the work, I'll see you again next month, your check is in the mail."  

That is about the extent of my freelancer management skills.  My hints for other people: pay regularly, don't be unreasonable, and give specific enough direction that the person can follow them without needing to ask you about every little thing.  Also, find the people who can follow a task to completion without asking about every little thing, because that skill is not possessed by everyone in the world and it is worth paying for.  Your time is very, very valuable.  You don't want to outsource to a low-wage, low-initiative country and have to answer six emails a day about "When you say you want Halloween bingo cards, what sort of words do you want about Halloween bingo cards?  Do you want words about pumpkins?  Or words about jack o' lanterns?"  My answer to that would be "I want you to figure out what kind of words I need, and I want to have about five seconds of involvement in this project."

Gabe: What sort of external tools do you use for analytics and SEO?

Patrick: I use Google Analytics -- everyone in the world will eventually have it on their site.  It is really, really good for giving you a firehose of information, but it is not good for time efficiency.  Analytics will let you find the percentage of visitors from Germany using Firefox who converted from the trial to the sale, Google Analytics will let you drill down to that information very well, with nice pretty graphs.  Unfortunately that bit of information is useless  to every business in the world, or substantially every business in the world.

Analytics and analysis is about getting actionable insights: figuring out facts which drive decisions that matter for your business.  Pulling actionable insights out of Google Analytics requires a lot of drilling and a lot of work.  You will often end up just not doing things that require a lot of work: it becomes the flossing your teeth of running a business.  (Maybe some people floss their teeth -- you're better men than me.)  

I tend to focus on other products.  One of my favorites is CrazyEgg, which does heatmap visualizations of where people are clicking on your website.  For example, you could set it up to track where people are clicking on your landing page.  You might discover "Aha, half of the people clicking on this page click on the left sidebar... which is not, actually, a link."  That would very quickly tell you that you need to write an A/B test -- because you always write an A/B test -- testing the current page against a left sidebar which was clickable.

I did that and it increased the conversion in a modest but statistically significant way, a good job for 3 minutes of work.

I also use Mixpanel, a Y Combinator company.  They do funnel tracking.  If you've ever tried to do this in Google Analytics, you may have discovered it is less than fun, particularly if you're doing something more sophisticated than "I want the number of people who sign up for my newsletter that go on to sign up for the free trial."  

Google Analytics treats everything as a page view.  If you want to track anything else, you have to write a lot of Javascript.  [Patrick notes: Events which can be represented cleanly by page views map to GA very well, but everything else suffers a painful impedence mismatch.]  

Mixpanel lets you track via a server side API, which lets you do much more sophisticated tracking.  For example, I have a funnel which tracks everything from trial signup to printing out their first bingo card.  I call that my "core interaction loop" -- people who succeed at tend to buy my software a lot more than people who fail at it.

If you are writing a World of Warcraft clone, people who don't get as far as killing their first ten rats and completing their first quest probably don't go on to pay you money, so you'd spend gratuitous amounts of time making sure those first ten rats were right in front of someone dropped into the world.  Similarly, I spend a gratuitous amount of time optimizing the first five minutes of interaction with my site, getting people to the point where something comes out of their printer and they say "Yay, I succeeded!"  Then they know they will continue to succeed if they use this for their class tomorrow.  

There are 6 steps in that funnel.  Mixpanel tracks the entire funnel and tells me where people are having problems.  For example, I have a customize your bingo cards page which was giving folks a bit of trouble.  There are about 15 elements on the page you have to mentally parse to get to the next step: do I need to change the color?  Do I need to change the font size?  etc, etc.  It is like Microsoft Office, overwhelming the user with 5 zillion options.

So I made a heuristic to determine whether you are sophisticated user by the standards of my user population or whether you are a newbie to the world of bingo cards.  If you are a newbie, it doesn't show you 15 options, it shows you 2 options: how many cards per page, how many cards total.  This is a fairly simple change which increases the number of people who get through the page by 20%.  Since everything is multiplicative, that increases the conversion of the whole funnel by 20%.  Since everything is multiplicative, that increases my sales by -- theoretically -- 20%.  

This whole idea that all improvements are multiplicative is something we see over and over and over again in Internet businesses.  [Patrick notes: It is even more powerful in the context of viral loops, where scaling goes from geometric to exponential.]  I think the original insight is from Steve Pavlina, who was a shareware developer who went on to other weirder stuff.  He has some old articles (http://www.gamedev.net/reference/business/features/shareprof/ ) where he explains how a 20% improvement at one stage and a 20% improvement at the next stage causes them to compound in a multiplicatively effective fashion (i.e. 44% increase not 40% increase), and that is one of the biggest insights that helped me build my business.

Gabe: Mixpanel, CrazyEgg, anything else?

Patrick: That is it for Analytics.  [Patrick: I forgot Clicky.  http://getclicky.com]

Gabe: You also recommend SEOMoz and SEOBook?

Patrick: Yep, I've been reading them since before I even had my business, back when I was an underemployed salaryman.  I'd read about them during my down time and even gave a presentation on SEO back at my ex-ex-employer.  They pretty much taught me everything I know in the 2004 to 2006 era.  Check their archives, they're very very good for beginners and intermediates.

Uncompensated plug here: SEOBook has a community which costs about $300 a month and I found it very valuable.  (I've since been made a moderator.)  SEOMoz also has a pro option where you can ask questions directly to their staff.  I never actually asked them questions, but I was a subscriber for a long time.  For folks on the lower to intermediate end of the spectrum, they're quite possibly worth your time.

Gabe: Did you ever use their tools?

Patrick: I underused their tools -- for the year I was with them, I only consistently used rank tracking.  Many of their tools are focused on a particular use case: the core of SEO in competitive markets is collecting links to your site.  One way to get links is to see who ranks for what you want to rank for, identify who controls that page, and send them an email asking them to please link to your website.  There are 10,000 ways to phrase that such that they might actually link to you.  [Patrick notes: focus on the benefit that they or their readers get out of linking it you.  See Peldi's email template.]

For example, Harvard University might have a page about English bingo as a pedalogical tool.  I could send them an email saying "You have a page about English pedagogy, I have a page about bingo cards used for English pedagogy, why don't you link to it?"  I basically never do that, because I find it is not efficient time-wise and I get about 100 of these emails a day from people who want links to their gambling bingo sites.  "You have a site about bingo, we have a site about bingo, why don't you give us links?"  I have always found this an annoying thing to get so I don't inflict it on other people.

Gabe: Do you ask your customers to consider linking to your site?

Patrick: Yes!  This is an easy, low-risk strategy.  In sales confirmation emails it takes ten seconds *once* to say "Hey, I'm a small business and would really appreciate if you helped get the word out by talking about me on your blog or website."  

Gabe: Has that worked?

Patrick: It has a non-zero impact for zero work.  The computer sends out emails automatically, without my intervention.

One non-obvious thing about the link graph is that some people make many, many more links than everyone else does.  SEOMoz calls them the Linkerati -- they have a million blog posts on this subject and they're all worthwhile.  For example, people on Hacker News probably have an average of 6.2 blogs per person.  They link out to things very frequently.  On the other hand, a teacher in Normal Bloomington probably does not have a blog, and probably does not share content that often.  So if you're trying to get links you would try to appeal to the interests of people who are linking a hundred times a day rather than the people who link once per year or less.

Since my customers are not the world's best source of links I don't typically focus my link acquisition activities on them.  However, I have gotten some fairly surprising links from folks directly in my niche.  For example, there is apparently a large community of Jane Austen friends in the world.  One of the wonderful things about the Internet is that if Jane Austen is your passion in life, you can find other folks who share that passion online.  There is an entire web ring of Jane Austen blogs where you and your other Jane Austinites get together, and one day one of them discovered Jane Austen bingo, and said "We have to do a Jane Austen bingo cards link."  That got me like 6 links to a page that cost $3 to create, so yay.

Gabe: What do you think about widgets with backlinks at the bottom as a linkbuilding technique?

Patrick: I have.  If you go to my popular card list, you'll find a widget that lets you embed the popular card list on your site or blog.  That is not a very good widget and it has not been successful.  The key of widgets, and linkbuilding in general, is that you need to come up with a way that giving you the link helps the person that is linking.

For example, if you have a widget for some game which allows someone to embed their avatar on their site, that fulfills psychological needs for the person doing the embedding.  It lets them show off "I have this character that I spent all this time building" and lets them brag to their friends "My Sword of Magic Destruction +10 is better than your Sword of Magic Destruction +8.  HAHA."  Things that let people demonstrate social proof typically spread quickly.  Also, someone who sees that sees "Wow, he has his character on his site, I want my character on my site", so they tend to spread in a viral fashion.

My underimplemented widget might cause someone to think "Well, I like this guy because he helped me out that one time, so I'll put his widget on my site" but most people won't do that.  If I were redoing the widget today, I'd figure out some way to implement the advice from Kathy Sierra: figure out a way to make your user awesome.  

Also, widgets were very popular about two years ago, but they got a bit of negative attention from Google, because people were using them to get scalable links to e.g. payday loan companies.  Anything that gets scalable links is basically Google's working definition of spam.  If you've got a zombie apocalypse widget that succeeds in getting 100,000 links for a payday loan company, then that technique is not going to remain effective for much longer.  If on the other hand you have widgets which are more focused on users, I think it has the potential for working.

If I remember right, I think you had a widget where you could embed a HN or Reddit karma tracker on their website?  

Gabe: Right.

Patrick: It occurs to me that I probably shouldn't have mentioned this, but let's run with it.  This is a great idea for a widget, because it lets peoppeople demonstrate to others how much they've spent on these sites and how central they are to their online personas.  [Patrick notes: social proof AND psychological fulfillment, it is like Mazlov's hierarchy of needs in a nutshell.  If only you could eat HN karma ...]  I would assume that it works well.

Gabe: So I have found that anchor text works amazingly well for influencing rankings.  Have you found the same thing?

Patrick: Well, yeah.  The problem is the techniques that I use for linkbuilding don't really lend themselves to suggesting anchor text.  Obviously I can suggest anchor text if someone copy/pastes a widget code, and I have instructions to "copy/paste this blue text to make a link on your site" on my site, which has a positive ROI because it takes one minute to do once.  However, this hasn't set the world on fire.  The vast majority of links on my site are links to "Patrick McKenzie" or "Bingo Card Creator."  I rank very well for my own name, for all five people in the world who search for it.  Link text works: figure out a way to get it to work for you, I have not quite found a way to get it to work for me.

Gabe: Widgets would be a good way.  What about guest blogging?

Patrick: I haven't done much guest blogging.  Guest blogging is one of a variety of ways to create content for other people on the Internet.  For example, guest blog for someone else's blog, create articles for someone else's directory.  The reason "someone else" accepts this relationship is that it creates value for them -- they get to keep the evergreen resources that you've just spent your time writing.  [Patrick notes: Evergreen means "content which will be good without modification until the end of time", as opposed to say a TechCrunch post which will be dated in a matter of hours and only 1% will be useful as a historical reference.]  

Up until quite recently, I had very little free time, and what little time I had was not put to best use making evergreen resources for other people as opposed to making evergreen resources for me.  I spent most of my time building my own blog.  If you have no blog and are creating one today and have no way of getting traction for it, then blogging for other people might help you jumpstart that, but don't replace your own blogging strategy with blogging for other people.  

Certainly, guest blogging has been successful for some people I know.

Gabe: That's about it for SEO.  Any other remarks about it?

Patrick: The one thing I'd say about SEO is the same thing about everything else: figure out a way to make it scale.  Especially for single founders like us, we don't have the luxury of a 20 person marketing team that comes in every morning and spends 8 hours on SEO every single day.  Figure out what you can automate.  Automate it.  Figure out what you can outsource to people whose time is cheaper than yours.  Outsource that.  Spend the majority of your time on strategic direction questions.  That is about all I have to say on the subject.

Gabe: A couple of personal questions: where does your nickname Patio11 come from?

Patrick: That has been my username on the Internet since about 6th grade.  The reason is my best friend in 6th grade was Puerto Rican, and he had a bit of a Spanish accent.  Spanish doesn't have a hard K sound, so after struggling with saying "Patrickkkk" for a while [Patrick adds: and refusing to use "Pat" because it sounded too much like "pato" (duck)] he said "Alright, your name is Patio" and it stuck.  So when I signed up for AOL back in the day, Patio + 11 (my favorite number) = Patio11.

Gabe: Why did you move to Japan?

Patrick: I have something I like to call the Venn Diagrama method of ensuring job security.  Up until college I was intending to do the standard career path rather than being an entrepreneur.  When reading the Wall Street Journal I saw several articles claiming that all programming jobs were moving to low-wage countries, so to increase my competitiveness I thought to combine programming with one other skill.  90% of programmers think they're in the top 10%, but that isn't me: I'm above average, but very very modestly above average.  

But modestly above average INTERSECT modestly above average in something else results in a very small set, meaning I compete with less people.  So if I go into a language, which is a difficult skill to acquire, that would improve my competitive position for getting a salaryman job somewhere.

So looking at the list of languages at my university, Japanese looked like a good bet.  It is a large country with a high-tech economy that does a lot of business with America and has few speakers of English.  So with the programming side of the Venn Diagram and the Japanese side of the Venn Diagram, I figured I'd have a nice safe job for life at some place like Microsoft.

When I graduated college, my Japanese was not good enough to work at Microsoft.  So I thought I'd go to Japan for a little while, learn to speak enough Japanese to function in an office environment, and then come back and work for Microsoft.  So I went to Japan and wound up in a little town called Ogaki in Gifu Prefecture, about 30 minutes north of Nagoya by train.  I really really like Ogaki, and thought "I'll stay here for just a little while longer."  Six years later, here I am.

Gabe: What in your previous experience helped you get traction for Bingo Card Creator?

Patrick: Probably, the single thing that helps me best is that I write fairly decently.  I consider myself more of a professional communicator than an engineer.  I did technical translation and teaching prior to programming, and have been blogging for the last couple of years.  I really like writing -- writing my blog, writing copy on my website, etc.  I don't have the reflexive engineer "What, you mean I have to *talk* to people in this business?  Can't I write computer code?" mentality.  So that has helped me a great deal.

Gabe: Do you think other people could repeat your strategy used in Bingo Card Creator?  Is it unique or are there a thousand other niches where it would work?  Could they quit their jobs, too?

Patrick: First of all, I think absolutely nothing I've done in the last four years is particularly inspired or out of the ballpark for anybody.  I think anyone on e.g. Hacker News could open a small business on the Internet: if you can program, that is all you need to be able to do.  I think there is a rich, vast, almost limitless sea of niches: many of them are bigger than bingo cards for elementary schoolteachers.

One of my friends from the Business of Software boards, Andy Brice (http://www.successfulsoftware.net) runs a site (http://www.perfecttableplan.com) which does seating plans, primarily for weddings and bar mitzvahs, etc.  He is my business idol.  Peldi from Balsamiq does Mockups -- and that's all.  I used to know a guy -- unfortunately he exited the business in favor of more lucrative opportunities -- that wrote scheduling software for chimney sweeps.  

If you can make a decent living doing bingo cards or seating plans or software for chimney sweeps, then clearly there are additional niches out there for you if you take the time to look for them.  [Patrick notes: talk to real people with real problems.  That is literally all there is to it.]

Some parts of my niche were ridiculously convenient for me.  The built-in long tail SEO strategy, for example, has helped me a great deal.  I think with enough creativity, domain expertise, and empathy with your customers' needs, though, you can create the moral equivalent of that in almost any arbitrary niche.

Gabe: Anything else you'd like to add about getting traction?

Patrick: It isn't a "flip switch and it works" kind of thing.  I know many folks in your audience are on the Silicon Valley model, where traction means opening a website for business, getting covered on TechCrunch 10 days later, and having your user graph immediately go vertical.  That is not exactly the model I've run my business on.  It has been four years of very part-time (average of 5 hours a week) hard slogging, just a million little things that combine to make a business that actually works.  If you're doing a full-time thing and have six months until your cash runs out -- and I'm not criticizing folks that do, that is a valid decision for some folks running businesses -- you need to accelerate the timetable, but that doesn't fundamentally change the strategy.

Try things, collect data, find out what worked, repeat the things that did work, iterate iterate iterate.  Collect the little 1% marginal improvements that are multiplicatively effective.

One of my goals when I was part-time was to run four A/B tests a month.  Three of them will have a null result, which is A/B testing jargon for "no statistically significant difference" in A/B.  One of them might pick up a 5% improvement.  If you pick a 5% improvement you'd be getting about a 70% improvement in revenue every year.  70% is pretty good compared to the raise you'd get every year working at a 9 to 5 job.  That is a very sustainable model, and that is what you can do in your part-time: feel free to do more than 4 A/B tests a month.

So that is my advice in a nutshell.

Gabe: Thanks very much for the interview.

Patrick: Thank you for having me, Gabe.  It is always a pleasure to talk about my interests with people.

 

If you have comments, hit me up on Twitter.
I'm the Founder & CEO of DuckDuckGo, the search engine that doesn't track you. I'm also the co-author of Traction, the book that helps you get traction. More about me.