Understanding information retrieval for greater revenue potential

WHEN IT COMES to the topic of keyword research, there is plenty of advice and written guidance, tools, tactics and resources available. In my opinion, much of this information is flawed, not insofar as questioning the accuracy of the data, but more from the perspective of motive and methodology. It is my opinion (and regular finding) that with poorly conducted keyword research there is exclusive reliance on the quantitative method, with little-to-no understanding of the science of information retrieval. In essence, too much emphasis is placed on finding keywords in isolation (chosen because they may be relevant, drive suitable volume and are competitive to a level we're willing to engage in) without any thought as to the qualitative interplay between keyword 'query types', and how such interplay drives revenue.

What are query types?

Information retrieval (IR) is the division of computer sciences that focuses on the retrieval of information from large bodies of content, including the web. In particular, when it comes to the retrieval of information from the web, search engine data and computer scientists studying such data - it refers to four query types into which keywords (or search queries) could be classified.

Informational: queries that are broad in nature and corresponding results-sets. We cannot infer a great deal about user intent; for example, 'poker' may be a query I would use if I wanted to learn how to play poker.

Navigational: queries that set a defined target ('Amazon' or 'Sky News'). Again, we cannot always infer too much about user intent here, however, we do know where they intend to achieve their objective.

Transactional: queries that reflect an action-driven user-intent, for example, 'rent holiday home', or 'no deposit bingo sign-up'.

Connective: queries that seek to identify specific information about the nature of indexed data. However, 'link:[sitename]' *site:[sitename]' are not commonly used by most search engine users.

Query types and the purchase funnel

The concept of a 'purchase funnel' is a commonly used marketing model which is not web-specific, but looks at the stages of a consumer purchase decision.

Whilst variations and expanded models exist, a commonly used purchase funnel would encompass the following stages:

Figure 1: Purchase Funnel


Common sense alone tells us that different query types are analogous to different stages of the purchase funnel as shown in Figure 1. Informational query types such as 'bingo' or 'bingo offers' may be considered relevant for the awareness-to-refinement stages; navigational queries '[merchant name] bingo', '[merchant name] bingo sign-up offer' may fit in any of the refinement-to-purchase stages, and our transactional term types 'play [merchant name] bingo', 'join online bingo website', fit primarily at the end of the decision stage and firmly in the purchase phase.

In addition to common sense, click/ conversion-attribution studies support this; as does our own attribution modelling and advanced data analysis at theMediaFlow.

Which is all why probability distribution and the oft-mentioned SEO wisdom of targeting a range of short and long-tail keywords, lacks efficacy as a keyword strategy.

Why probability distribution (volume and competition) is a flawed basis for keyword research

Our use of the terms 'long-tail' and 'short-tail' keywords derives from probability distribution which, if envisaged, may be imagined as a comet shape. Short-tail keywords are (probabilistically) fewer in number, though volume (and competition) is high, whereas long-tail keywords stretch to an indeterminate length, with almost infinite variations on query length and make-up (though volume and competition decrease). Choosing to research and identify keyword targets that sit across the short to long-tail spectrum, purely because of volume and competition considerations is fundamentally flawed in two ways:

1. Understanding data in this way has volume and competition as its objective.

I'm not sure about you, but my primary objective when practising SEO is revenue. Sure, volume and competition levels have a contributory impact to revenue but my choice should not be led by volume and competition considerations.

2. Understanding data in this way has a 'single visit' mentality.

Targeting short-tail terms because they are high volume, plus long-tail terms because they are low volume (yet less competitive) does not take into account the relationship between the term types. Long-tail queries of a transactional nature do not tend to magically materialise in the mind of a customer. I need to research 'online bingo' and 'online bingo site offers' before I am aware of the existence of'[merchant name] bingo double deposit special offer'.

Tips for affiliates seeking to apply query-type keyword research to drive increased revenue

1. Analyse, classify and understand the interplay between keyword types currently driving search visits to your site.
2. In looking at your own site, 'navigational' query types will be those terms that include your brand name (and misspellings).
3. Implement conversion attribution tracking so you can see the effect of different query types on ultimate conversion activity.
4. Use Google Analytics advanced segments to monitor visitor behaviour, by query type.
5. Analyse your site content. Does it 'speak to' the different purchase decision stages and where do the query types fit in those stages?
6. Track and monitor the position and volume of visits on informational query types. See how increased volume and visits on information query terms leads to increased revenue/conversions on navigational query terms.
7. Use the data collected above and this understanding of the interplay between query types to re-work your keyword strategy.

Of course, do not lose sight of keyword volume and relevancy considerations, which can be sanity-checked well enough using Google AdWords tool (exact match) data, cross referenced against terms of a similar nature that already drive traffic to your site, using existing Google Webmaster Tools data.
printer friendly LAN_NEWS_24email to someone
  Login to rate

You must be logged in to make comments on this site - please log in, or if you are not registered click here to signup