Identify, Analyze, Act: SEM by the Numbers

Aug 19, 2008 • 8:37 pm | comments (1) by twitter Google+ | Filed Under Search Engine Strategies 2008 San Jose
 

Many companies find it difficult to use web analytics for more than reporting and ad hoc investigations. By defining requirements, roles, tasks, and benchmarks, an efficient process replaces one-off requests. This session covers practical workflows that you can quickly implement to see improved, consistent returns from your data. This sets a platform for experience-based learning that helps a company to set standards, anticipate a build-cycle or campaign refresh, and prioritize search marketing efforts. Moderator: Chris Boggs, Search Engine Watch Expert & Manager, SEO, Brulant, Inc., recently acquired by Rosetta Speakers: Craig Danuloff, Founder & President, Commerce360, Inc. Brian Cosgrove, Site-Side Analytics Engineer, AvenueA / Razorfish Heather Dougherty, Analyst, Hitwise Michael Stebbins, CEO & Founder, Market Motive Brett Crosby, Senior Manager, Google Analytics, Google

First up is Craig Danuloff.

Roadblocks and challenges are the topic we'll be discussing. Lots of challenges in paid search, based on data, organization, bad habits.

Three factors of evil - what your up against - when digging into analytics.

1- Invisible info - data we want that's not there. Lots of pieces that we can't quite get to. 2- Deception - things are not what they seem to be. 3- Unlimited powers and resources.

Invisibility - most important is the full spectrum of what's going on in search process. The start and the end of what your measuring is missing. The queries that match the text ads are vital. This info is not as available as it should be. Want to know every query paid for and that's difficult.

Profitability is also difficult to measure. Done on an ROI basis non on ROAS basis. In order to do that must enter margin information. Sometimes hard to come by. Lots of times positive ROAS does not correlate with positive ROI. Want to enter margin for each good enter, and take that against true cost to make decision.

Deception: Can you trust what you see? Lots of keywords, and campaigns - what comes back in reports are averages - rollups. If you haven't carefully constructed grouping - without looking inside. Brand terms vs. non brand terms are a great example. Need to segregate.

Accuracy: It would be nice to know if the sample is statistically significant - tools don't take this into account.

Unlimited power and resources - massive amount of data and a time frame that's ever rolling. Really need to watch where you put energy and effort. The tool set does not give you direction on where to make decisions. Also, always making lots of changes with keywords, creatives, bids. Competitors are doing things. Often make changes without recording them. Tool sets don't provide this today. We need to watch our tests, otherwise we are just guessing day after day. Sometimes need to be disciplined and not make 50 changes a day.

What can you do to deal with these issues? To deal with missing information - demand to get the info on the query and ROI. Adwords for example does not tell us which keywords were connect to queries. Omniture has a tool that helps with this. ROI and margins - same thing. All a matter of awareness. Understand that averages and aggregates must be segregated into homogeneous piles in some level - so that they are meaningful. Low performers, high performers need to be segregated. The rollup must be a narrow component of info to be useful.

Lastly - better math we can apply. Take the data and put it into Excel. Make records of changes. When you go back and see which change was positive.

One tiny tip is to organize keywords in groups with exact match, phrase match, and broad match separately.

Next up is Brian Cosgrove of Avenue A.

Started out as an inhouse SEO and has an engineering background.

First topic is implementation. Poll: How many are running paid and organic. Almost everyone. Who has a base feed? A local feed? Videos? All these things are on the same page and any analytics package will record this as organic if not configured properly. It's going to bucket them together. Need to know ahead of time to make sure that the system can separate that stuff out. Especially things like paid inclusion.

Filtering: Many do not take into account that a lot of internal traffic comes in for many reasons. Important to filter. Look at reports and look for things you don't understand. Very useful for SEO campaigns. To get good statistics, need to take action on your site.

Data driven organization - people within an organization will request different reports. Many are not using this to drive business decisions. Comes down to roles. Analysts are good at deciphering the data. Come up with insights to feed to developers, PR teams, specialists (PPC/SEO) and lots of other individuals in the organization. One role is missing - the operations person. The project manager. Someone needs to have the foresight to get developers together, content writers, and line it up ahead of time with a concrete list of recommendations to act upon.

Case study: A company changed their site every six weeks. Spent first four weeks going through data. Next two weeks were dedicated to handing off the business requirements to hand off to other team members. When that's handed off, the project manager takes over. The analyst switches gears and move into another mode - maybe landing paid optimization or other areas not addressed. The analyst is putting together action lists all the time.

A few specific reports you should look into- landing pages for organic search. Looking at those pages is extremely important. How landing pages relate to keywords is fundamental in paid and organic. Often the page that ranks well for the word is not the best landing page. Sometimes you get lots of traffic to pages that are not bringing people to their objectives.

Conclusion: -Implement platform correctly -Identify actions you can take -Coordinate with other resources -Separate the analysis cycles -Staff people to manage changes

Heather Dougherty from Hitwise is up next.

Using CI to identify and maximize SEM Opportunities.

Identify and maximize SEM Opportunities. Identify trends. Shows slide of a retail brand. Shows Christmas and "back to school" patterns. Be able to plan ahead for trends instead of reacting.

Look at the breakdown between paid and organic within specific industries. Learn competitiveness and pricing. These ratios help you get a benchmark of whats going on. Knowing what competitor is doing is critical. Look at which search engines are sending traffic to competitors. Learn about their strategies. Look at ratio of paid vs. organic traffic. Identify branded terms and volume of traffic that comes from them.

Compare where people are searching to where they are clicking. Look for lower cost keywords. Improve keyword list efficiency. Prices are going up. Identify who is doing well in sponsored listings and learn from their copy. Look at how competitors are getting organic traffic. Is it the content? Maybe you can partner with competitors for ad buys, licensing, etc.

Determine user intent. Purchase or news? When looking at intent, can learn about what users are thinking about their brand. Case study: American Airlines. What are people typing in addition to the keyword? There was a major spike when all the planes were grounded. Queries showed that some searchers suspecting there was a conspiracy, and this was vital for reputation management.

Integrate search findings across the organization. Search is not just about SEO and PPC. Take advantage of findings to identify other marketing opportunities such as affiliate partners.

Next up is Michael Stebbins of Market Motive.

Will share own process for web analytics and how they shape paid campaigns.

What's in your data? Bounce rate, average time on site, and pageviews are good foundations for measuring basic interest. But make sure you are collecting conversion rate, cost of visitor, and revenue per visitor. Crucial. Cost per visitor is more difficult to measure. If revenue comes from off the site, its no excuse not to measure it. Need to put the data in analyics, even though it will never be accurate.

The Grim Reaper Approach: Tactical question - which 10% of my ads are not performing? Possible answers: Campaigns with low engagement, low ROI, low conversion need to be cut. People often cling to ads hoping they will perform. Better to create new ads. Can sort by ROI and eliminate candidates for deletion.

Check commercial intent tool at adlab.msn.com. Put in 2 terms and tells what the likelihood is for purchase. Adwords keyword tool is great too. Now shows search volume. Check forecast and demographics. Read Bryan Eisenberg on how to tune message to the user.

Placement - Google Ad Planner- free. Tool tells us sites that have the traffic we are looking for. Next you test. Create 3 copy of your ad. Don't want to cut a performing ad in half. A neat trick. Set ads to perform evenly.

Takeaways - Collect, Question, Cut, Create, Place, and Test.

View slides at http://www.marketivemotive.com/ses

Next is Brett Crosby from Google Analytics, formerly of Urchin.

Plugs the new book - "Always Be Testing" by Bryan Eisenberg.

More and more businesses and people within companies are demanding reports and analytics. The back end guys found themselves all of the sudden pumping reports. With GA, analytics jumped into the mainstream. The audience changed. No longer a small group within the company, but the entire company. Launched a new interface to put the data in context. As you use the tool, you get smarter. There is still data for the pros, but also for everyone. Allows people to get feet wet and just touch the surface.

Formalizing the process - getting the right data to the right people. Set up goals and funnels (for both ecommerce and non-ecommerce). Many people he encounters do not set up goals. Build customized dashboards, and customize email reports for different job functions (webmasters, executives, SEOs).

Goals and funnels - set up in the admin interface under "edit settings". Even without ecommerce, can set the goal value. Say a lead is worth $1,000 - plug that data in. Define the funnel steps. Can tell us goal value per visit. With ecommerce, the data is really rich. .

Custom dashboards - every report has an icon to add it to dashboard. Get the most important metrics in your face.

Custom email reports - get the data out to the guys that need it. Can run schedules. Customized data to the right people at the right time. Do this for each major role, and everyone is happy.

Contributed by Avi Wilensky of Promediacorp.

Previous story: Search Advertising 101
 

Comments:

Craig Danuloff

08/21/2008 08:26 pm

Thanks for the write-up of my presentation. I've posted the slides and the text of what I should have said (but maybe did not) at http://www.clickequations.com/blog/2008/08/ses-san-jose-08/

blog comments powered by Disqus