Ionsearch "Killer Keyword Research Panel" Slides

"Killer Keyword Research" - or More - How to Improve Relevance?

In addition to these annotated slides below, there's also a blog post on my experience of talking at ionsearch here: Relevance Analysis, Keyword Research and Searchbots.

  1. 1-4. Keyword Research is actually Relevance Analysis (i.e. what content should our organisations be relevant to?) - which in turn is actually digital market research - it's actually finding out what our customers actually want.
  2. 5. In 2007, Udi Manber, VP of Google Engineering went on the record and stated that "20 to 25% of the queries we see today we have never seen before".
  3. 6. In March 2011, this was updated - Google are now seeing more repeated search queries, but nonetheless, they're still not seeing 1 in 5 search queries.
  4. 7. This is worth thinking about - this means that 20% of everything that gets typed into Google every day has never been seen before.
  5. 9. Yes, really.
  6. 11. What this monumental spread of search queries means, is that if you could plot individual search queries on an x-axis and the no. of times the query was typed in on the y-axis, you would have an absolutely massively long flat-ish curve - a very long tail that would stretch for miles.
  7. 12. To illustrate this, we have around 10 real-world search queries all of which are just for "pearl wedding jewellery", and around 8 which are just searches for "cheap laptops in the uk". There will be many many more such queries.
  8. 13. So, the key question is - how can organisations cost-effectively collect as much of this long-tail traffic as possible? If you target just 20 or 30 queries what about the 100, 1000, 10000 that you are not targeting?
  9. 16. The answer is: (a) To collect together a large sample of relevant search queries that you should be winning (b) Understand the things - the concepts - the objects behind the queries i.e. understand what things people are searching for (c) Understand how these things relate to each other (d) Structure your website, app, whatever digital thing you're building around this customer demand
  10. 19. Here, we have, 8 different search queries all to do with "outdoor clothing". They're all different queries, but actually all searching for the same thing: "outdoor clothing".
  11. 20. On this slide, we see three separate groups of search queries: one a group of queries for "Ski Jackets"; one a group of queries for "Mens Ski Jackets"; and one a group of queries for "Ladies and Womens Ski Jackets". These serve as an illustration of the idea of understanding the "things behind the search queries".
  12. 21. So, we can take the four clusters of search queries, clusters of search demand, and list them. But what more can we do with them?
  13. 22. Answer: turn them into a information architecture, a website structure (or an app) that is entirely based around the things that people search for i.e. based around what people want
  14. 25. Each point in the structure is a landing page that specifically targets a set of search queries. This means that you now have a structure that also directly matches the search queries and matches what people want.
  15. 28. Relevance Analysis (which is actually digital market research) can go right to the heart of your organisation. I did some relevance analysis (examining over 100,000 search queries) for a particular organisation that has offline shops, advertises on TV, and a website - and that has a mission statement that runs through the organisation like lettering in a stick of rock. None of the concepts, words and phrases in their mission statement appeared in any of the search demand analysis that I carried out. Essentially, no-one wanted what was at the heart of their organisation.
  16. 30. Relevance Analysis can also demand significant change of an organisation. Organisations, when presented with the findings from relevance analysis can require senior management to carry the required changes through.
  17. 32. Given the importance of relevance analysis, we need to look at what sources of search queries are available.
  18. 35. Experian Hitwise currently provides the best data in the UK, but, it is expensive. However, if you can afford it, it is worth it.
  19. 39. Google Adwords Keyword Tool is free, but focused on AdWords rather than providing just pure raw search queries - and - can be quite time consuming to extract search queries from.
  20. 45. Looking ahead to the future a little: we're already seeing an increasing percentage of "Not Provided" values appearing in Google Analytics (Experian Hitwise UK reckons its around the 5% mark) - meaning that we're beginning to lose sense of what we are already relevant to. Firefox is apparently experimenting with making people use SSL/HTTPS by default. Imagine if all browsers moved to this setting - we would completely lose any sense of what we are relevant to. And this would impact Experian Hitwise too - leaving us with just Google Adwords Tools as a way of identifying what we should be relevant to
  21. 46. Given this is the case, shouldn't we examine other possible ways of identifying what we should be relevant to? http://www.bloomreach.com/ is an example of a company doing exactly this. A company definitely to watch I think.
  22. 49. Remember: Keyword Research actually is Relevance Analysis which in turn is actually Digital Market Research