An Interview with Steve Plunkett: SEO Experimentation Done Right
Steve Plunkett is a well known for his SEO insights. You can follow him on Twitter or view his online resume at LinkedIn. And now enjoy an interview with Steve:
How long have you been in SEO and how did you start?
I got on the “Internet” in 1994.. (at 2600kbps) I was doing mac tech support for AOL and also moderating a chat room and bulletin postings (kinda sounds like Social Media a bit, no?), and if we count making sure my bulletin postings ranked first.. if you want to count ranking first for channels on IRC, then 1995. Otherwise, Brian was starting webcrawler and Jerry was starting Yahoo! Directory in 1995 while i was a webmaster/sys admin in the Univeristy of Texas System and I was experimenting with spam controls for webcrawler (I accidentally got all the first 30 listings for about 10 terms), and would modify my website and they would recompile to see if the filters worked.. then I started up a web design component to the ISP I was working at in late 1995 and ranking websites kinda just was part of our service. Back then charging $150/month for hosting a website was still kind of expensive.. We just bundled SEO in with hosting.. but since we designed the websites we kinda just did SEO from the first step.
Do you believe SEO is evolving and getting “more white” so to say (no paid links, no spamming, etc)? Do you think Google will ever be able to win the war with Black hat SEO?
That depends.. the only links I have ever “bought” were Yahoo! directory once they started charging for it.. and free website listings I would get from buying ads in print publications or by being a member of this community organization or that professional organization. Because I was working at an ISP/Web Dev., and kind of an “anti-spam” editor @ yahoo while I was an higher up category editor for ODP/DMOZ I hated spam. So I never really did anything other than follow the “Golden Rule of SEO” – provide unique content specifically relevant to the search query. I’ve always been very vocal about not using links and following Google’s implied rules to the letter, it’s not only my client’s website, it’s my career for the past 14 years.
Back in 96 I remember the big thing was domain shadowing.. you would buy a domain and point it to a subdirectory of a website so a website with 7 subdirectories might be listed under 7 domains.. I never really did this for SEO, but I sure hated it when I would find the same site submitted to Yahoo! or DMOZ like 14 times (each URL could be listed in 2 categories), and have to hunt them down and remove them.
Honestly I think there are many things that are a bit black hat in how Google utilizes them to rank websites, but if you understand the basic premise of Google you can do good SEO. If you want to serve up completely unique and relevant content edited for humans’ search queries, then you should do well. I think SEO is evolving in the terms that there are more people doing it. Hence more people, more tactics, more qualified population per query.
You seem to Twitter a lot. What are you using the tool for?
Networking, i’ve met some cool people like @karlrove is following me now… @mchammer is a die hard raiders fan (Raiders 4 Life!~)
Twitter/FaceBook/Myspace has replaced Yahoo! IM – it’s also a good barometer of people, those that consistently push out negativity on twitter are people I don’t want to work with.. nor follow, etc.. I like the shiny happy people.. It’s similar to moderating an online chat room. I also use it for experimentation collaboration.
Are there any other social networking tools you use more or less often?
Myspace is good for the entertainment and fashion industry. FaceBook is a good local social networking site. LinkedIn is the premier business networking website. I stumble a bit.. I’m actually doing a lot of experimentation with RSS feeds and music, and different types of searching patterns on different music blogging websites. It’s part of my job to test everything so I have tried almost everything, ranking for podcasts, (HINT: It’s totally like ranking bulletins on AOL). Video search, Image Search, book search are all different search types that I deal with for varied clients.
I know you are quite active at SEOmoz. What other SEO communitites do you frequent and why?
I was active with SEOmoz and I would credit my professional membership there with a lot of my contacts on the web.
SEOMoz was the first SEO community I really got into because at first it was nice and clean. I met some people who didn’t really need to hear themselves speak, but wanted to learn and collaborate, which was what Iwould want in an SEO community.
I personally dont’ think there is a better website for people to read about SEO and i still read their blog from time to time.
You once mentioned you very often deal with SEO experiments. Could you list most important rules one should follow when running an SEO experiment?
My actually job title should be “SEO Scientist” instead of “Director of Internet Marketing”, every day I sit in my office and turn up the music and experiment all the time. I am running about 70+ experiments right now. Using blogs, social media websites and several other websites I have access to. (HINT: A good reason to donate some time to a non-profit) And of course just modifications on some Organic SEO clients I have, and documenting the progress after I have tested on non-client websites then pushed the tactics to client’s websites.
RULES:
a. NEVER do an un-tested experiment on a client’s website. (never done it, it’s just common sense)
b. Document ALL the factors prior to conducting the experiment, add additional factors at baseline, first measure, secondary measure and tertiary measurement points.
c. Always allow for variances among subject and properly validate any outlying factors and account for them.
d. Re-evaluate your hypotheses at each stage of the experiment.
e. ALWAYS allow for statistical regression analysis.
f. Perform the same experiment 3 times before coming to a conclusion.
G. Either re-test or modify experiment if additional variables introduce themselves prior to experiment completion.
b. Document ALL the factors prior to conducting the experiment, add additional factors at baseline, first measure, secondary measure and tertiary measurement points.
c. Always allow for variances among subject and properly validate any outlying factors and account for them.
d. Re-evaluate your hypotheses at each stage of the experiment.
e. ALWAYS allow for statistical regression analysis.
f. Perform the same experiment 3 times before coming to a conclusion.
G. Either re-test or modify experiment if additional variables introduce themselves prior to experiment completion.
I know you’ve been conducting your own experiment on some issue I described in one of the recent post on internal anchor text over-optimization. Could you share your experience?
Well, I read the experiment but never saw the website, so my mind went to about 200 variables. So i re-created the test (my version of it, because I didn’t have all the factors the first person did, I guess?) on 2 brand new websites where there was no domain age, inbound links, etc. I tried in 2 ways, one standard top nav, sitewide, modifying one anchor tag sitewide.
Version 1: this worked positively and I am still #1 for the term.
Version 2: I modified the top nav on 8 different sub-sections to include the original anchor but NOT any of the global navigation elements. (still testing)
What are most common mistakes people make when condusting an SEO experiment?
Version 2: I modified the top nav on 8 different sub-sections to include the original anchor but NOT any of the global navigation elements. (still testing)
What are most common mistakes people make when condusting an SEO experiment?
1. Coming to a pre-mature conclusion.
2. Incorrectly diagnosing all the variables in an experiment.
3. Removing one variable from one of two competing subject groups
4. Not testing in a vacuum (running proper controls)
2. Incorrectly diagnosing all the variables in an experiment.
3. Removing one variable from one of two competing subject groups
4. Not testing in a vacuum (running proper controls)
What are your plans for 2009? Any new projects in mind?
I started 8 different experiments right before I left the office for 2 weeks prior to NYE, the basics, title tags, ALT tags, anchor tags, etc. I did this to see what if any different weighting factors might be applied to on-site optimization as the year changed and also as a control group for newer sites i am launching this week. That experiment is on-going and by making slight modifications to certain variables, we have been able to bring some members of the control group to #1.
I think I am speaking @ IMSpringBreak (Apr. 2-4th, 2009) on IED (information environment design) – contact @davesnyder for more info about a good IM conference. It’s kind of like a spring break version of the ScarySeo conference in Ft. Lauderdale.
Read more: http://www.searchenginejournal.com/an-interview-with-steve-plunkett-seo-experimentation-done-right/8302/#ixzz0yqoUxOsA
