Insights in the inhouse keywordtool of T-Online

Jens, Andreas. We are happy to be able to interview you. Would you like to shortly introduce yourself?
Jens: I will try to keep it short; I’m 35 years of age and have been working in finance before I graduated in Information Science, which I studied at the university of Darmstadt. I started working for T-online during my graduation and after that I started managing the T-Online portal’s search engine. I got the opportunity to build the SEO-Team. Besides all that I also write for my blog Inhouse-SEO.

Andreas: I will try to keep it short as well. I am 25 years old and working at Deutsche Telekom for almost two years now. Before, I used to work at Lufthansa Systems and a small marketing agency. Like Jens, I studied Information Science in Darmstadt. We got to know each other during that time!

We met during SMX Munich. There you mentioned you were developing a Keyword tool. Can you tell us something about it?
Jens: the tool contains all the search queries that have been entered into the T-Online search engine. Daily the number of requests go into millions. For each search query we know which results were clicked. We know what the click ratio is and if these would go to organic listings or advertised listings. Of course we are able to tell about trends and are able to provide spelling corrections on entered search queries.

Andreas: with our newest features we will be able to tell how organic listings compare to clicks on advertised listings. We are able to tell which domains are often preferred per certain keywords and the coolest feature is the typing mistakes module.

Keyword analysis is often underestimated and mentioned together with SEO. Attention goes out to Pagerank sculpting and other technical methods for optimization. A good keyword tool however, is rare to be found. How important do you think keywords are for SEO and how can your keyword tool help?
Jens: Keywords are essential. It is the base for both SEO and SEM campaigns. But in the end it is always about anticipating to our users need; their desire and motivation. Here are information click ratio will definitely help. By this it is possible to detect navigational queries. Very popular, but often difficult to optimise for. If I am not the goal of interest, at the best I can generate traffic but no conversions. It is good to know if the user has commercial intent. The CTR on ads for “DSL test” is lower than “DSL”. It is is clear that in the first search phrase the information aspect is more important than information about DSL contracts. However we know a lot of these type of searches where it’s not really clear what the user intent is. By looking at trends and signals we are able to anticipate.

Andreas: There is not really much I can add to Jens his statement here. We can show in several tests how important it is to adept the content in the speech of the user. This doesn’t only This doesn’t only affect SEO positively but also increases the usability a lot.

For example we generate a 30% higher CTR with changing one keyword and place a keyword that where more often used in the same measuring period. Because of that I believe that keyword analysis will be more important for content optimization in the future. Our tool helps in achieving those opportunities.

Rankings based on keywords in Google are getting more difficult to compare and watch. This is why we don’t use ranking anymore as a KPI in our reports. More important to us is traffic, conversion rate, search density and cluster building on a keyword level. How do you rate your SEO campaigns, where is your focus?
Jens: I think about it the same. The positions don’t really matter. People from management don’t want to struggle with huge keyword lists. We aggregate visibility by clustering keywords and their clicks. Based on the cluster we are able to tell how much more revenue is possible when increasing the visibility of this cluster. The forecast will be most exactly how better the data in the system is that you get to do webanalysis.

Andreas: I believe traffic, conversion and search intensity are the most important KPIs. Ranking positions on certain keywords are so volatile it would be imperfect to add these to reports. It is more important to tell something about the visibility in the search engines for a specific search. Also I have to add that ranking position don’t always tell you something about the amount of traffic  you can expect. Often it is not the first but the second or third result that is being clicked by the user. In the end it is the traffic that counts and on page the conversion rate.

The search engine of T-Online is after Google and Yahoo the most important one. If I am correct; google is using a Google Custom Engine. By this you get search data you can use. How does this data compare to data you see in Adwords or the keyword database of Stefan Fischerländer?
Jens: Stefan Fischerländer is getting most of his data from MetaGer. But the market share of MetaGer is quite small and its user base very niche oriented. Andreas can tell more about this. In the opposite to Google we act similar in the filed of end costumers. Because of the integration on T-online.de we have access to a very wide spectrum of users. At business level I don’t see us comparable. But also in this case, Andreas can tell more about data quality of Google than I can.

Andreas: Yes the data that Stefan is offering is very specific for a special group of people. You cannot use this data for general use. We get most of our data from Germany’s widest known General Interest Portal t-online.de. Compared to trends from Google we can say that our trend curves look about the same as those of Google. The advantage of our tool is that we have more concrete  measurements to our knowledge, which we can use. But then again the question rises how this relates to traffic. How much traffic can I anticipate with a keyword if my content is already found in the SERP.

By combining data you get new values for keywords and positions. CTR based on position. Conversion based on position. Search intention on keyword level as well as conversion rate on cluster level. How do you use this data with new campaigns?
Jens: CTR in the SERPS we use both for adsense and organic results. We can only measure conversion rate on our own websites. For us its most important to get the best results for the users intent. Only navigational queries are a bit tricky and get us a bit frustrated. Nothing is worse than spending a lot of time and to conclude that your effort was worthless.
Also interesting is to search for reverse. Just look at all keyword that where used to click to some defined url’s. Sure with the search volume for each.

Andreas: Yes, this is one of the most interesting features of the toolset. Even the feature will help to do competitor analysis. First time you see with which keyword the competitors gets most of there relevant traffic.
 
Are you planning to release your anonymous data for 3rd level parties? How do deal general with privacy related issues within the T-Online Search Engine and the website?
Jens: privacy is really important to us. That’s why we don’t store any IP addresses in our database. Also we use no hash codes or similar. We also don’t store any data about a whole search session. If we would have had these parts of the users identity could be exposed. We still have in mind the old fall with the AOL-data. Surely some of the information get’s  lost but we can be sure that no search request will be assign to a person.

Andreas: We have been thinking about it to make the data public to 3rd parties. We do not know yet if this will occur and how it will look like in the future.

Which keyword was in March 2008 the most valuable, based on conversion, count and revenue?
Jens: I cannot give you an answer about that

Andreas: Me neither.

Those were our questions! Thank you very much!

More interesting information about keyword analysis you will find on the Inhouse-SEO blog.