Posted by brianspatterson
This post was originally in YOUmoz, and was promoted to the main blog because it provides great value and interest to our community. The author's views are entirely his or her own and may not reflect the views of SEOmoz, Inc.
We recently completed an interesting reputation management project and I thought it'd be helpful to post our strategy and results to the SEOmoz community. My hope is that you'll read this and get some ideas, or even better, you'll point out some areas that we overlooked or things we can do to improve our approach.
The Client's Problem
Our client approached us with a problem that we are now seeing fairly often for many companies. As you began to type our client's brand name into Google Search, Google Suggest displayed our client's brand name + the word 'scam' as the second option, directly below their brand name. Talk about damaging your reputation!
We signed a confidentiality agreement, so I can't say specifically who the client is, but below is a screenshot from another large company, Direct Buy, whom we found experiencing a very similar issue.
Our client believed they were losing a lot of business due to this issue, particularly in the case of people who were ready to buy, but then went to do a Google Search to learn a bit more about the company before they plunked down their credit card. There is a great quote from Dave Naylor on this exact problem, "If Google Suggest's second result is 'scam', then people WILL click on it". These customers likely clicked the 'scam' recommendation and were scared off from purchasing from our client.
Our client is not a fraudulent organization in any way, and they offer real services and products to their customers, but the way they offer service is also complicated and in a volatile industry where no company is without its detractors. Even though they deliver their product as stated, and 99% of their clients love it, there seemed to be a small percentage that just weren't happy. Some of these unhappy customers took their gripes to the web. Additionally, there were also a high number of obvious cases where competitors were posting negative information about the company in order to damage their brand. We would not perform reputation management for any company that was a scam or participated in fraudulent or misleading activities, and after fully researching the business, we were 100% comfortable with helping them with their problem.
Because of all of this negative content about our client, when someone indeed clicked the '[Insert Brand Name] Scam' suggestion from Google Suggest, they were finding the first 2 pages of results filled with very negative 'flames' about the company. Some of this negative content was on personal blogs and others on complaint sites and forums. There were even a few positive reviews on blogs that were inundated with numerous scam accusations in the comment section, thus making a positive article turn very negative and harmful to the brand.
Our Research
Before diving right into what we did to change this, I want to talk a little bit about some research we did on this issue. We wanted to understand, as best as we could, how this problem came about. We hypothesized that very few people actually typed in 'brand name scam' initially, but maybe at some point it was just enough to get it to be a suggestion. Once it became a suggestion in Google Suggest, searcher's curiosity was piqued and so they clicked on it at a high rate. Google likely interpreted the large amount of clicks to mean that the phrase is a highly relevant suggestion, and as such moved it up to the top of the list of suggested terms.
Google's official statement on how Suggest works, from this blog post, is:
"As you type, Google's algorithm predicts and displays search queries based on other users' search activities. These searches are algorithmically determined based on a number of purely objective factors (including popularity of search terms) without human intervention. All of the predicted queries shown have been typed previously by other Google users. The autocomplete dataset is updated frequently to offer fresh and rising search queries. In addition, if you're signed in to your Google Account and have Web History enabled, you may see search queries from relevant searches that you've done in the past."
We think there is quite a bit more to it than this. We recently read of a case study where a brand new domain had acquired a 'scam suggestion' from Google Suggest. It was evident that nobody had searched for this domain, let alone searched for the domain with the word 'scam'. What the domain owner found was that two scraper sites had scraped content from his site, and those two scraper sites had the word 'scam' buried in the URL. Based on this incident, we think it is very possible that content and associated words in Google's index may also influence the suggestions.
This SEOmoz Q&A by Dr. Pete is also about this very topic, and Dr. Pete believes it is possible that Google Suggest is biased to serve up the 'scam' suggestion, among others.
We kicked around the idea of working to influence Google Suggest to force out the ‘scam’ suggestion, and may revisit it down the road, but we decided that the fastest way to take action would be to push the negative content out of the SERPs with positive content that the client had complete control over. This way, when someone searched the scam phrase, they'd have to dig deep into the SERPs to find anything negative about the brand.
I know that you may be thinking that pushing bad results out of the SERPs feels a little dirty. I felt this way at first, however, after fully researching various approaches and processes we now believe firmly that it is indeed a Google sanctioned method. Our belief is based on this blog post from the Official Google Blog on how to get rid of negative brand rankings in the SERPs. In it, it states:
Instead, you can try to reduce its visibility in the search results by proactively publishing useful, positive information about yourself or your business. If you can get stuff that you want people to see to outperform the stuff you don't want them to see, you'll be able to reduce the amount of harm that that negative or embarrassing content can do to your reputation.
Our Approach
We pitched the client, and subsequently implemented, a pretty ambitious plan. Our stated goal was to own 90% of the first two pages of Google results in 6 weeks. To control at least 18 positions, we knew we needed to focus on more than just 20 pieces of content. We decided that we would define 50 pieces of content, and as time went on, we'd determine which pieces of content Google was signaling that it liked (by slowly moving it up) and which it didn't. The content we focused on fell into two natural categories, Pre-Existing Content and New Content. The content for each of these categories was as follows:
Pre-Existing Content
- Subdomains on the client's website - The client had created two of these before we were brought in. They were subdomains setup that specifically addressed the false accusations.
- News articles - A benefit of the client being a big company is that they've already had plenty of mainstream press. We identified positive articles from Business Week, The Wall Street Journal, and other Industry publications to promote for the scam phrase. We found that, even if the article didn't contain the word 'scam', anchor text alone, linking to these strong domains, could get them to rank for the scam phrase.
- Wikis - It seems that most industries and niches have their own wiki's. Our client had a page in a niche wiki, so we simply added the word 'scam' into the wiki in a natural way. Doing this, plus a few links, helped it rank for the scam phrase.
- Blog Posts - There were a number of positive blog posts about the company already online. The problem was, as I mentioned previously, that the comment sections of many of them were overrun with very negative comments (we could tell most of the comments were anonymous and contained inaccurate and fake information, likely from competitors). So, we chose to only promote blog posts that had disabled comments. Even if a blog post had no comments, we didn't use it if comments were open because they could always turn negative.
- Youtube - The client had created a few Youtube videos disputing the mis-information being spread about their business. Since YouTube allows for full content moderation, we found videos to be a great source of positive content that can be controlled.
New Content
- Content on the client's website - When the client originally tried to tackle this problem themselves, they had created a few posts on their blog that were optimized for the brand name + scam keyword. Since an official brand site is the most likely site to rank for any query containing the brand name, this was a smart move.
- Posts on sites we own - We have a fairly large number of blogs that we run as part of our business. Some of these blogs focus on the same industry as the client, so we simply created posts optimized for the scam keyword. Since these domains are aged and trusted, we knew it wasn't going to be too difficult to get them to rank.
- Article Directories - Squidoo, HubPages, eZineArticles, Buzzle, InfoMarketers, Go Articles, and many more - We have nice, old accounts on many sites like these, so we added new articles optimized for our term to them.
- Mini Blogs - We setup a number of mini-blogs on WordPress, Blogger, Posterous, Tumblr, and a few other WordPress MU sites we identified that we felt we'd be able to create a blog on that could rank.
- New Sites We Created -We bought the .com, .net, and .org versions of the exact match domains for the search phrase (including the word 'scam', eg. brandnamescam.com). We also bought hyphenated versions of the domain as well. We then setup mini-sites on different c-class IP addresses.
As you can see from the lists, our targets included a diverse set of content. The key was that there had to be some sort of control over the page. Either comments had to be turned off (to keep a positive article from becoming negative by a bunch of negative comments) or we needed to have control over the page/comment moderation to ensure we could control the message.
The general content on these pages included customer testimonials, positive stories, general information about the company, satisfaction guarantees, debunked mis-information, and other stories that either didn't pertain to the scam issue at all, or they showed positive aspects about the company. Is this a perfect strategy? No, I don't think so. But we believed that having 2 pages or SERPs with little information about an actual 'scam' is probably enough for most searchers to abandon the topic.
Link Building
After we had our content targets identified and/or created, we started the link building process. One thing I absolutely loved about getting some of these articles ranked was that it took almost no work to get something on page 1. Some of the positive pre-existing articles that we wanted to get on page 1 were on sites like the New York Post, so it basically took 2 lower-quality links with the exact anchor text 'brand name scam' to get it on page 1. It made me (briefly) dream about how easy a job it must be to do SEO for a site like The Wall Street Journal; you can practically rank #1 for any low-competition search term you want!
Our primary link building strategy was built around using article directories. We wrote hundreds of unique, quality articles (no spinning or machine generation) and submitted them to article directories, web 2.0 sites, blogs, and other sites that accepted our content. We varied our anchor text, and spread out the links across sites, and over time, so that the link profile was fairly natural.
Interlinking
We also wanted to interlink our sites in a way where they would all benefit, while avoiding obvious signals of 'link farms' or 2 or 3 way link exchanges. What we came up with is represented in the graph below. We've replaced the actual sites with S1, S2, etc, but this is the exact interlinking pattern we used. Sites that needed more help received more links, while some of the stronger sites only needed one or two links pointed at them.
Social Engineering
I also wanted to talk about another tactic we used to take on some of the more stubborn sites that just wouldn't seem to move out of the SERPs. In our case, these stubborn listings were two personal blogs. We heavily researched these blogs to understand the psyche of the authors. We then determined two separate strategies to pursue that would help us with our goal. In short, for one blog we made an offer to buy it outright. We didn't explain our background or why we wanted it (that is irrelevant to the buy/sell process), we just simply made an offer and began dialogue with the owner. In the second case, we talked to the webmaster and during discussions realized that the owner was not interested in the traffic received from the article, so we were able to work out a deal to help move the content out of the SERPs. We treaded very lightly with these tactics for two reasons: (1) We wanted our work to be legal and ethical, and (2) we needed to be very careful that these site owners didn't just create a new blog post talking about how our client was trying to 'buy their silence'.
Execution & Results
The results from our project were near-perfect. We obtained nine of the top ten results on page one, and all ten results on page two. We think that if we had more than just six weeks to complete this, we would have been able to get all 20 of the top 20, but 19 out of 20 wasn’t bad and our client was ecstatic.
I'd love to know your thoughts on how we approached this and what you would do differently. Based on the success we've had, we are looking to expand our offerings in this area. I personally loved the challenge of this and the interesting aspects of the problem.
About the author: Brian Patterson is a Partner at MangoCo, a Search Engine Optimization Company in Virginia. You can follow Brian on the the Twitter @brianspatterson.
No comments:
Post a Comment