Determining Keyword Relevance
SEO is not an exact science, but when dealing with large sets of data, even the most eagle eyed among us will struggle to process everything perfectly. This presents a big problem at one of the most crucial stages of the SEO process – keyword research. Even a site with just a handful of target keywords can throw up some confusing data, and trying to work with a list of several hundred phrases can be almost impossible.
My personal preference for doing keyword research is to use a permutation tool, throw all the data into Adwords and then tick all the keywords with traffic above zero that are actually relevant to the particular client’s products or services (adwords often throws up some weird suggestions). The result is a list of keywords that often stretches into triple figures. I have recently been trying to work on a formula that helps process this data, and I think I’ve finally done it.
I’m calling it the Keyword Value Index (or KVI) and the formula is as follows:
KVI = Keyword-Relevance3 x Search-Volume0.5
That’s Keyword-Relevance cubed, multiplied by the square root of the Search-Volume, divided by the SEO-Competition to the power of 0.3 (which is somewhere a between the cube and 4th root). For those who aren’t mathematically minded don’t worry, the exact formula to put into Excel is given below.
Now, I know I’m not strictly first off the mark on this. Before I set about doing this myself I spent a lot of time trying to use a formula that someone else had created, but what I found was that everything available missed out one of these three variables used in the formula.
I believe there are three things to take into account when doing keyword research:
- How many people are looking for this?
- How hard will it be to rank organically for it?
- Will it create a conversion?
Only by considering all three of these can you efficiently prioritise which keywords are worth targeting, but the human brain will struggle to understand two datasets together, let alone three.
So, how do you make the formula work?
This requires you to make a subjective judgement about the keyword and quantify it as a number. Using the same example as my post last week about searcher intent, the keyword ‘types of apple’ is information based, the keyword ‘buy granny smiths apples’ is sales based, this is something every SEO should be familiar with. The KVI formula uses a 1-5 ranking system, based on the following criteria:
- Typically one word informational searches or searches that are not directly relevant to your products or services.
- More detailed informational searches or relevant phrases with unclear intent
- Information about the products and services you offer
- Specific product searches or a detailed service query, showing purchase intent
- Search for your brand name or long tail search for your service area
This is the easy one, you pull the data directly out of adwords. Make sure you use the local data if you’re only interested in local traffic – the usual stuff. The only thing to note is that the formula won’t work if the search traffic isn’t put in, or if it’s marked as zero. We all know that the adwords tool doesn’t display keyword data for a lot of long tail phrases when the search volume drops below a certain amount, even though our traffic will show otherwise. For these keywords I just usually put the traffic in as 5 searches per month, just a nominal figure to make the sums work. Don’t assume they have no searches at all, it’s rarely the case and you could miss out on valuable phrases.
The adwords competition metric is pretty much useless for organic SEO, as the data is almost always skewed towards either end of the 0-1 scale, and it doesn’t give an indication of how many other websites are competing in the organic listings. For this metric I use the allintitle operator, as shown below:
This is quite a popular metric for keyword research. Unlike the broad match search which shows all the pages that list the keywords anywhere in the page or the “phrase match” operator using inverted commas which shows pages with the exact phrase anywhere in the page, the allintitle operator does as it says, it shows all the pages which have all your search words in the title of the page. If you think how page titles are written, this operator gives a good indication of the number of pages which are competing for organic placement for that particular term.
The only downside with this is that there’s no easy route, you have to manually input the keywords one by one. In theory there should be a way to use Google’s API to manually populate the results – if anyone manages to do this then please let me know!
Running The Formula
Once you have the three variables it’s time to put the formula into action and calculate the KVI for each of your keywords. Assuming most people will do this in a spreadsheet, the excel operator to use is =(A2^3)*(B2^0.5)/(C2^0.3) where A is the relevance score, B is the number of searches and C is the number of allintitle results.
You then just need to rank the results from highest to lowest KVI and you have your keywords ranked in priority order. The actual number is not really relevant, what’s important is getting them in order, and it’s certainly not true to say that a keyword with KVI of 100 is twice as important as a keyword with KVI of 50. It’s just important to prioritise your resources on those higher up the list.