Benchmarking an SEM Campaign

Apr 10, 2007 • 3:10 pm | comments (0) by twitter | Filed Under Search Engine Strategies 2007 New York
 

Moderated by Allan Dick, who will be throwing his awesome dinner tonight.

Cam Balzer from Performics will lead off. He says they have a new discipline called “benchmarketing.” Since search is such a public arena, you have to pay attention to what competitors are doing. How are others responding to the dynamics of the marketplace, thus the mix between search and benchmarking. Goes over the search funnel and benchmarking. First ask “how visible am I in search results?” How effectively am I reaching searchers, and lastly what is the cost/ROI.

You can first “benchmarket” against yourself. In above stages use the following metrics: impressions, clicks and CTR, leads and $$, in order to derive CPC, total cost, and ROI. Since search is so fundamentally competitive, but also benchmarket against competitors using the following metrics: Coverage, share of voice, share of wallet and overall cost and ROI. Will now walk through several benchmarketing examples for Performics clients.

One benchmark example he showed: Looking for paid visibility of select advertisers on a sample of 38 insurance terms. They use AdGooroo to load up keywords they want to track. They (AdGooroo) go to all engines in multiple countries and perform searches repeatedly in order to gain insight on visibility. Shows a chart with Avg reach on horizontal axis and coverage % on vertical access. Looking at a bucket this size gives a good overview of this vertical, and allows for analysis and predictions. Some questions that can be answered include: how visible am I? What kws are showing up? What are my competitors’ bid strategies? What level of coverage should I maintain on high volume/high cost keywords.

Then shows another visibility benchmark from Hitwise. He used the visibility benchmark of “share of visits.” Showed the top 15 sites that received traffic from the search for “iPod.” Apple accounted for 54% of total visits, but this isn’t as much as they maybe should be getting, since the Apple sites to well both in Paid and Organic rankings. So what more could they be trying to increase reach? Question to ask is what portion of visits are generated by particular keywords.

Next benchmark is the traffic benchmark. They use Hitwise again here that follows click data to determine the monthly percentage of traffic from search to selected home improvement sites. Questions here include how much traffic is coming from search versus my competitors. Bear in mind that this is both paid and Natural search blended together, and they cannot separate the two datasets. They like to combine this data with the market share data in order to gain more insight. Even though some seem to be found well in search, they may not get the level of visits that some competitors. Looking at this allows one to decide if they could be more aggressive in their SEM.

Natural versus paid traffic. According to ComScore qSearch data, the total traffic from paid and organic varies from 5% paid/95% Organic all the way up to 50/50 and more tilted towards paid. The average across all categories is about 89% of search traffic is organic. Car rental actually gets 54% of its traffic from paid search. This is very useful data on an aggregate/category level. Considerations include weighing the proportion of natural versus paid against the absolute volume.

They “take their thought leadership role in the industry very seriously.” In January of 2004 they started tracking an index of 50 of their clients. He then discusses the performance benchmark of cost. The chart he shows varies dramatically on an avg CPC level, and also indicates a strong seasonal relationship to the cost. Looks at other metrics as well. He notes that the Performics 50 can be found at Doubleclick.com.

Next is Mike Moran with IBM. One thing that they found with search marketing campaign benchmarking is that is was difficult to benchmark the competition since they have so many different products and competitors. They found that no matter how important IBM was, in 2001, Google didn’t really feel they were that big of a deal. Less than 1% of visitors came from SE’s. So they needed to really focus on their own benchmarking, which is what he will focus on today.

Conversion rate is a nice tool, but they do not want to optimize on that, instead you want to increase your traffic. The nicest thing is if you can do both. Many people don’t want to talk about what they are actually selling, instead focusing on their rankings. Each site must be trying to get someone to do something. At IBM, they want people to download the whitepaper. They have found that 2% of these will end up buying a 50K consulting package.

The first step in organic SEO is to be in the index. Which search engines are critical? This can vary by country. How many pages are in the index versus how many pages should be there. The hard part for some companies is figuring out how many pages they actually have. In a lot of cases, the exact number doesn’t matter, but that the trend shows increased traffic. The second step is to choose the right keywords. Ask what percentage of the traffic should be yours. Third step is to examine what the landing pages have on them. Want to look at conversions, since they will help you find out what needs to be done. If you have good rankings but no conversions, there are a couple of possible reasons. Your web site might not be the strongest, or maybe you are not focusing on the right people. This needs to be done week by week, month by month, keyword by keyword, landing page by landing page. Focus on constantly trying changes and experimentation to get you where you need to be.

There are some good tools to help with rankings, as well as some that perform Content audits. Focus on the analytics packages. The fourth step is figure out how to find links to the site. For any competitive keywords, this can be the difference. Use tools to find, “score” links. Tools can also manage link campaigns. Use the competitors backlink data to gain insight. Suppose you do not have time for all these tools? All-in-one tools that he likes for small businesses include soloseo (hosted) and WebCEO (software). It s nice to have one dashboard instead of having to reenter metrics into every different tool.

“I can’t look at each page individually.” Build tools. They built their own tool to analyze for missing or duplicate titles, for example, as well as other factors. They also broke it down per division, and used “management by embarrassment” to color-code the particular site areas that needed more help. Even though many managers didn’t really understand what was wrong, they wanted to move from a red code to a yellow and eventually to a nice green. These metrics can be updated and should be to include the current factors that matter most, and omit issues that have been handled already.

“None of this works for personalized search.” He has been looking at a tool recently called SEMlogic, which looks at the underlying factors across competitors, Analyzes which ones matter for your keywords, and allows for decision making. Now, IBM has 2M pages indexed in Google versus 10K. In 2001, 95% of pages had titles, now up to 99%. 2001: no top 10 rankings, now over 3000. Briefly promotes his and Bill Hunt’s book, “Search Engine Marketing.”

Martin Laetsch, SEMDirector. Will talk about what he has learned about benchmarking “with a small little company like Intel.” When he talked to marketers, they often came up with “fluffy” marketing terms like reach and frequency. This is not enough. With Intel, one problem was that it was a very siloed organizational structure, which made things difficult. When they started with Intel, 19% of keywords were duplicated in at least 2 campaigns. Pentium alone was in multiple SEM campaigns, costing hundreds of thousands of dollars in wasteful spent.

Typical management challenges with information included: missing information, too much information, numerous sources, not valuable, difficult to get from other areas of the company. Yet competitor information is difficult to get. The use of reporting analytics for benchmarking is important. In order to get actionable information, comprehensive reports are required. The reporting and housing of this information needs to be scalable to other software. You should have consistent reporting, and know the audience for specific reports.

When dealing with thousands and tens of thousands of keywords, so many things to consider. Even as deep as the delicate balancing act between PR teams and sales. So: used a centralized organizational structure for all search program management. They set up best practices, then moved on to tactical SEO and SEM execution, followed by establishing the reporting and analytics software framework. Also, then line up search with other marketing efforts… The Intel Viiv press release resulted in 15,000% more Google searches! He ends up with two quotes: Andrew Lang “He uses statistics as a drunken man uses lamp posts – for support rather than illuminations.” Sherlock Holmes: “It is a capital mistake to theorize before one has data…”

What does this mean? Marketing is changing, and online is going to con tin ue to be a larger percentage of the overall mix. Measurement becomes that much more important. Must use solid data to make sensible decisions.

Note: This is live coverage of SES NYC 2007. Please forgive any spelling or grammatical errors that slipped by n the interest of speed.

Previous story: Podcast and Audio Search Optimization
 

Comments:

No comments.

blog comments powered by Disqus