Hacker News new | past | comments | ask | show | jobs | submit login
Patrick McKenzie (patio11) interviewed for an hour on SEO & AdWords (gabrielweinberg.com)
212 points by epi0Bauqu on Apr 6, 2010 | hide | past | web | favorite | 48 comments

Thanks to Gabe for the interview and, in particular, for letting me write and edit a transcript of it. I have always preferred reading to listening myself.

If you guys have any questions, feel free to ask.

Thank you for doing it with me.

Anyone running a Web business would benefit from consuming this interview. It's probably one of the best ROIs for an hour of time.

And I'd love to do more like it. So if anyone has any good interviewee suggestions (for the subject of getting traction), please introduce me!

What I'd like to know is what frame of mind allowed you to slog through that for the last 4 years (losing money early-on). Did you believe in the idea? Were you just constantly tinkering (mental auto-pilot)? Were you doing other startup activities in addition? Did you sometimes take a break from the Bingo Card Creator to try other things?

I find that even ideas that I believe strongly in, I eventually get distracted and wind up doing other ideas... until distracted again. Rinse, repeat.

losing money early-on

The business put more money in my pocket than it took out since, hmm, about 30 days after I threw the doors open. (I think I know how you came to this misconception, so sorry for the unclear phrasing in the interview: I had a few hundred a month in sales from SEO/etc when I was burning $1 a day trying to figure out how AdWords worked.)

Did you sometimes take a break from the Bingo Card Creator to try other things?

I put it on the backburner to focus on either the day job or on living rather frequently.

As to the frame of mind, I always say that business is a lot like WoW. People express huge skepticism that you could play WoW for 30 solid days of your life and not feel bored out of your mind... but you don't get 30 days of /played all at once. You smash a goblin here and complete a quest there and organize a raid for your next weekend and maybe think about offing that dragon eventually and then collect some mushrooms and kill the dragon and then... whoa, 30 days.

I passed the last four years on a thousand and one intellectually engrossing mini-projects. They never got to the point where they'd lose my interest -- after all, if I just hold out until Saturday, I'll have A/B test results to look at. If I just put in another hour, I'll have a new feature. If I just write this email, I'll have a happy customer and maybe get another postcard for my fridge. etc, etc.

The loot is a bit better than WoW's, too.

Thanks for the transcript, it's easier for those of us (supposedly) at work :)

Pretty interesting for those of us that don't follow you that closely :) - I liked the practical advice and examples.

Also for those that just prefer reading :-)

So, yeah, thanks for the transcript.

What was your first job in Japan and how did you get it?

My first real job in Japan was as technical translator for the prefectural technology incubator. (I am not comfortable telling you who, exactly, although given that I live in the equivalent of a small town in Kansas it isn't exactly hard to guess.) It was arranged through the JET Program -- I applied as a Coordinator of International Relations, they saw my engineering degree, the rest is history. Note that I'm told there was exactly one technical translator placed that year, so I wouldn't suggest that as a high-percentage route to your dream job in technical translation.

My next job was the job I recently quit. I got that by overdelivering on a project for a client of the incubator. When my contract was up I called him and essentially called in the favor. He arranged for my recently ex-bosses to hire me (they were clients of his), and came with me for the job interview, which at that point was a formality.

Good stuff! So you use Paypal/Ejunkie... That's obviously easier than getting a merchant account and jumping in with something like BrainTree. Can you run thru why you made that choice and why you're sticking with it? Any plans to change?

When I started my business, my level of web programming expertise was "I can create a page in HTML in Notepad." Dealing with Paypal callbacks causing key generation, accounting, etc was way the heck out of my capabilities. Enter e-junkie. As I gradually improved in my web programming abilities, it always made more sense to extend e-junkie than it did to rip out and replace, particularly since they wrap both Checkout and Paypal in a consistent interface.

It works, costs $5 a month, is reliable like you wouldn't believe, and their lead developer literally sends me Christmas cards. Why would I ever change?

It is likely that my next product will not use e-junkie, though -- subscriptions, blech.

Interesting-- So assumption no subscriptions, you'd use ejunkie for a new product? I always assumed that it would ultimately carve out a bit more of your revenue than a standard merchant account.

By the by-- good post on subscription commerce options here: http://www.untitledstartup.com/2010/02/accepting-payments-on...

I noticed you support Google Checkout. Have you thought about adding Amazon FPS? From my experience, about 20% of people choose Amazon versus 10% for Google. It seems like a really small change to capture a few more percent on conversion.

You're an inspiration to me.

Quicky: From reading your blog, I see you had a desktop programme in java at the start. Do you still have a desktop programme now, or is it all web based?

If you have a desktop programme, do you have any advice on choice of language so that non-techies can download and run with as little fuss?

Thanks, that means a lot to me.

As of April 6th I have a desktop program. If that phrasing suggests to you that I have a very consequential A/B test planned for this month, you're very perceptive. :)

choice of language

Forget language, focus on the packager/installer. Java, .NET, and that RIA/Flex stuff all work fine if you file off all the edges in your installer. For Java, I strongly recommend you take a look at Launch4j (OSS for Windows) and possibly Excelsior JET ($$$ for Windows). The Mac side is easier since Macs are a very predictable environment for Java these days -- use the mac jarbundler task and you're pretty much home free.

I read that you are using paypal (and also ejunkie from clicking buy on bcc.com) . What type of paypal account are you using?(Personal, business, premier)

Do you use any plugins/gems to integrate the payment?

I use both Paypal (Premier) and Google Checkout. Both ping e-junkie when a transaction is completed. e-junkie then does two HTTP posts to me -- one to grab the Registration Key and one confirming the details of their transaction.

I don't use any plugins/gems to handle this. e-junkie's documentation is good and it is mostly a matter of copying parameters they pass into the right models for my home-grown sales tracking, then calling sale.save. This takes all of 77 lines, counting whitespace, and about 60 of them are assignment statements.

Ok. if you have a pure web app, will you still use e-junkie? Any advantages in that case?

And a Big Thanks for the transcript! I wish every video interview had transcripts like these.

That was one of the best hours of my life I've spent all year. I've been researching adwords and online advertising for about six months now and I've learned more in that sixty minutes. Thank you!

Question for Patrick (or any other SEO experts, I suppose): You mention the SEO benefits of exact domain name matching (easterbingocards.com for a query for "easter bingo cards") -- is it also useful to have a partial match (patrickseasterbingocards.com for the same query)?

Right now I have tarsnap.(com|net|org), and I'm wondering if I should consider buying tarsnap<foo>.*, for appropriate values of <foo>, and just making them redirect to tarsnap.com.

There is a similar "Gray hat" SEO strategy, which involves buying domain names that include keywords you want, and then 301-redirecting them to your site--without updating their whois information. I wouldn't do this kind of thing unless you can devote a lot of attention to it.

You can create content-light pages like the ones Patrick does, for, e.g., encryptedbackupsonline.com--each page on the site would have another title from the same keyword cluster, and each one would end with a call-to-action like "Tarsnap offers [encrypted backups online] for security-conscious power users." (Where the brackets indicate a hyperlink.

Bonus points for having a landing page on your own site that targets those keywords.

(As is probably clear from the above, I do SEO. Patrick's advice is extremely good, so if you have the time, you'll probably do best by just doing what he says. However, if you'd like to know what you might pay to outsource that process to a third party, well, my email address is in the profile.)

Do those shadow domains need to have anything link to them for them to be useful?

I think if you do a redirect there is no advantage to that. Travel to http://www.easterbingocards.com, there is content there for the search engine to index and return as a result (content that also of course has a lot of the keywords that are being targetted).

(Unless you're talking about doing a reverse proxy to the content of the tarsnap site, but I believe there is also not much advantage to that since the search engines pick up on duplicate content)

[EDIT: this is coming from things I've read not controlled experimentation. I don't think you can be an 'expert' without the latter so maybe I should not have responded]

Ok, that's what I thought -- I started doubting when I read Patrick's comments. Thanks!

The strategy that you have outlined has vanishingly little to recommend it. You lose the exact match domain bonus and split your link equity across your main site and the mini-sites.

I hadn't seen http://tractionbook.com/ before... glad that I have now though, looks like a lot of great interviews.

Awesome! 2 of my favorite HN startups, both of them got the "Make something users want" part right.

Very helpful. Thanks. Just wondering, why you decided on going with RoR and if you considered any other frameworks such as Django?

It is what all the cool kids were using. No, seriously, there was that "try Ruby right in your browser" site and I fell in love with the language in a matter of minutes.

Can anyone please explain that scalable links stuff? I tried searching for it, but I guess the term "scalable links" is very insider lingo.

Which part of it? Why you would want to get them, or why Google does not want you to get them?

Links are the primary determinant of rankings in competitive searches. (Competitive searches are high value ones with lots of savvy folks angling to win, or ones where there are other players for the search term with built in advantages: for example, [credit cards], [buy viagra online], or [facebook] are competitive searches, [patio11], [elementary reading bingo cards] or [how do i sort an array in ruby] are not.) Thus, you want links.

Most methods of actively acquiring links are extraordinarily labor intensive: they boil down to sending emails to folks asking for them to link to you, possibly with a sweetener to the deal (such as money, content, a reciprocal link, etc etc). Google likes this state of affairs, because as long as links are hard to get, then the number of links you have is a good proxy for either quality or prodigious effort expended in ranking. Either works for Google -- it makes spam less cost-effective than buying AdWords ads, and there might be gains to the user experience, too. (A less jaded individual might flip that sentence.)

If you're on a campaign to get links, ideally you'd want that to scale out of proportion to the time invested, for the same reason you want users, sales, etc etc to scale out of proportion to the time invested. Google wants exactly the opposite, for the above reason. Accordingly, virtually any tactic which repeatedly results in scalable link generation will eventually be discounted, regardless of whether it improves the user experience or conforms with Google's guidelines, which they enforce in an arbitrary and capricious fashion.

One example of this is widgets, which do result in scalable link acquisition. See the video for what I think of them.

Thank you Patrick for the explanation. I'm trying to think of the ways in which a widget can create a link on a site where it's installed. I'm thinking:

1) Javascript, buy making a GET back to the widget server, and then inserting the link into the DOM of the widget host page.

2) Iframe.

3) Flash or something like that.

What am I missing? Can't Google easily ignore a part of a page that's generated by Javascript (their bot doesn't process it anyhow, right), and also ignore Iframes or Flash?

Copy this magic spell to put my widget in your page:

<iframe ... > <a href = http://www.example.com >Anchor Text Here</a>

Empirically, most people do not strip the link.

You can add my 'empirical' confirmation of your datapoint. We did the same thing with a webcam-in-a-frame using the webcam component, the link was left in-tact in almost all cases.

MP3 feed: http://traction.blip.tv/rss/mp3/

(Posted as I can't see it linked anywhere)

Thanks for the interview, Patrick. I am also curious about what led you to switch from the Java Swing app to the web app. If you could do it over again would you start with the desktop app and move to the web or start with the web?

I have a blog post titled "Why I Am Done Making Desktop Apps" and it answers this at length:


Short version: web kicks desktop's booty from perspective of developer in conversion rates, support burden, maintainability, and ability to extract data. My next app will be a web app, no question. (If it were a do-over, would I be knocked back to my programming competences as of mid-2006? Because that was Java desktop apps, only -- I couldn't write Ruby and couldn't even spell SQL at that point.)

Patrick - is Google Analytics or Google Conversion Tracker able to track conversions that may occur across many sessions/days apart.

For example, tracking the conversion from landing page to registered user is easy. But what about the conversion from registered user to paying user. That conversion might occur a week later. Do you know if that is tracked, or does google only count a conversion within the same session.

good interview, but the format just looks weird to me. i.e. when the guest isn't talking, you are just watching someone listening.

With both people on the screen, you can change focus between each person, as they speak.

well you got me to sign up for mixpanel.

This is all mildly interesting and nice, but honestly $30K projected for this year is nothing. And that number shows to me more limitations of this particular method of SEO & AdWords than anything else, i.e. it is not a way to build company.

My first crappy product made $8K a year in profit 0 advertisement. My second made $72K a year no advertisement no SEO. My third product is making over $500k a year (not projected) no advertisement no SEO.

I'd ask what your product is but someone who created an account an hour ago and pulled out some unsubstantiated large figures probably won't give specifics.

Sure certain products can pull in stupidly large revenues, but this is an example of a person with limited time applying a small amount of time to a niche and being nicely profitable.

By that logic, doesn't your lack of success with no advertisement or SEO on the first product prove your second and third products don't exist?

I don't know what are you talking about... Sorry.

If you take your first experience: "My first crappy product made $8K a year in profit 0 advertisement" You might draw the conclusion that '0 advertising is no way to build a company'. In which case you'd have never bothered trying the others.

Adwords and SEO are extremely valuable and profitable when done well. Bear in mind that Bingo Card Creation is a relatively small niche. There are people doing finance related things with adwords who make millions.

That's not really true. You would be encouraged if you made 8K off a crappy product with no advertising. You would think that making money is easy and that you could do much better with a less crappy product. True or not, that's how I'd read that statement and feel about that situation (which is actually pretty similar to my actual situation).

OP "This is all mildly interesting and nice, but honestly $30K projected for this year is nothing."

I was suggesting that if the OP applied his own logic...

Personally I think making any amount of money is cool and something to build on.

He made $30k at 5 hours a week. That works out to $120/hr. Not bad.

Last year only, he worked on the project for four years and I don't think he made that amount in the first year.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact