How to Compose a Semantic Kernel
It is not enough just to compose the semantic kernel. It must be regularly updated…something may need to be added and/or something may need to be removed. It is a hot topic for commercial websites. For instance, equipment models become obsolete very quickly, new products appear, and all these changes must be reflected in the semantic kernel.
Thus, here is a step-by-step manual on semantic kernel composition.
1. Collecting queries from all possible sources:
a) Keyword statistics: Yandex.Wordstat, Google AdWords Keyword Planner
b) Keyword databases, related software (Pastukhov database, keybooster, liveinternet and similar, Key Collector, etc.). I know true Pastukhov database fans, who only use it while composing semantic kernel, not taking into account other possibilities described below
с) Entering keywords exported from Google Analytics and other web analytics tools. It’s possible if you already have your website. By the way, I gather topics for new articles from the statistics.
2. Specifying regional attribute and query frequency; estimating seasonal popularity; evaluating links budget by keywords (adwords.google.com will help us here).
3. Clustering gathered keywords (queries).
4. Creating landing pages for selected keywords.
5. Not forgetting to make sure that users go to the landing page by the query you considered to be the most relevant. In other words, correlating queries with landing pages.
I’m often asked this question:
How many and which keywords are needed for a website’s main page, and how many and which ones are needed for single pages?
Depending on your web resource subject matter, numbers may vary.
Let’s think: Can your website main page “sell”? If the answer is “no”, then put emphasis on the inner pages of your site. I promote the main page of my site by 15 queries.
For single pages I choose 2-3 mid-frequent queries.