How to use Wordtracker, Google Adwords and web analytics to identify the best keywords for your site by Sally Kavanagh, 14 December 2007

We may all know that thorough keyword research underpins the whole foundation on which all online traffic generation is based. But how do we separate the wheat from the chaff – the keywords that deliver quality not just quantity, customers not just visitors?

We may all know that thorough keyword research underpins the whole foundation on which all online traffic generation is based. But how do we separate the wheat from the chaff – the keywords that deliver quality not just quantity, customers not just visitors?

My answer is to use a combination of Wordtracker, Google Adwords and web analytics.

Let me take you through the process I use. I recently took on a new client who sells standard and personalized calendars online, so I’ll use this as my example.

A few definitions

• Keyword – either a single word or phrase that is used in a search.
• Customers – any visitor who fulfils the purpose of my website. For an ecommerce site this will be sales. For a site whose main purpose is to build a subscription list, then it will be arrival at the thank-you page following a successful submission of the sign-up form. Different sites will have different ways of defining a customer, and in some cases may define customers in more than one way, for example signed up to the newsletter or bought a product.
• Most valued visitor - any visitor that fits the profile of those likely to convert into customers, and these are the ones I am targeting. Suffice it to say, you cannot define your most valued traffic unless you have defined what it is you want them to do!
• I want conversions, not rankings. A number one ranking for a keyword that doesn’t deliver customers may be good for the ego, but doesn’t help the bank balance. Even worse is buying pay per click traffic for keywords that don’t convert - this feeds the ego and depletes the bank balance.

The initial trawl

At the start of my keyword research, I always like to spread my net wide and then hone the results down to the best keywords for my site.

Wordtracker’s Keyword Universe is always my first port of call. Starting with the very generic term ‘calendars’ provides a wealth of ideas, everything from ‘cat calendars’ to ‘promotional calendars’ to ‘nude calendars’. Obviously this market is very wide so I will make an initial decision about which part of the market I am interested in. I am going to keep to the mainstream family and business market and ignore everything else. Even so I already have a sizeable list of possible keywords, but which are the ones that will deliver customers?

To keep things simple, I have selected just 12 keywords from the list that Wordtracker has generated. These are shown in Fig 1, a screen shot from Wordtracker’s Keyword Universe. ‘personalized calendars’ is an American spelling so I shall use ‘personalised calendars’ when working in the UK market.

Fig 1 – Wordtracker Keyword Universe results selected from a search on ‘calendars’

Wordtracker provides a count of searches used for each keyword contained within its database and this gives me my first indication of quantity. Now I need to refine the data on the popularity of each Wordtracker keyword in my particular marketing environment.

Testing with Adwords

Google Adwords is perfect for testing keywords for real, on the exact same audience that contains all my potential customers. I even get some traffic and sales while I am doing my research!

I set up an Adwords campaign using the keywords derived from Wordtracker. My purpose is to generate information on my keywords, not drive traffic to the site. Traffic is a by-product, albeit a very welcome one! As with any Adwords campaign, I make the title and description fit the keyword, but optimizing the ads for the best clickthrough rates is not crucial so I do not labor over it at this stage.

I aim as far as possible to bid so that each keyword is at the same position. The results will be more accurate when comparing data for different keywords that Google displays at the same position on the page. This is not always easy but bear in mind that once the position drops below about 8, then the number of times the ad is displayed will be greatly compromised and comparisons become increasingly invalid. Ideally I aim at about position five - this helps to keep the budget down (remember I am buying research data not traffic), while maintaining reasonable impression levels.

The number of impressions each keyword generates is an extremely valuable number. It is the best indication we can get of the traffic to a particular keyword, in our own current trading environment.

Fig 2 – Adwords data for one day period for our 12 keywords

Fig 2 shows the data for our 12 keywords. Ideally the average positions should be more consistent but even so it is clear that ‘2008 calendars’ is the most popular term by far with 3455 impressions. ‘personalised calendars’ is next (1530 impressions) with ‘create a calendar’ (1237 impressions) not far behind.

So now that we have a very good indication of where the quantity of traffic is, let’s have a look at the quality.

Enter web analytics

How visitors to a website behave is fascinating and provides extremely valuable insight. I use ClickTracks to analyze my web stats, but whichever solution you prefer should be able to give you the same information - though perhaps in a different format.

Which keyword searches convert into sales?

Different web analytics solutions show this in different ways but ClickTracks does it graphically by labeling different groups of visitors in different colors. I label all those visitors that made a purchase, so that I can then see which keywords they used to find the site. I’ve exported the analyzed data to Excel and taken just the keyword information.

Keyword	Total Visitors	Sales	Conversion Rate
2008 calendars	570	5	0.09%
printable calendars	39	1	2.56%
printable 2008 calendar	175	2	1.14%
blank calendars	169	5	2.96%
create a calendar	842	240	28.50%
promotional calendars	82	3	3.66%
personalised calendars	1980	23	1.16%
dog calendars	199	134	67.34%
pocket calendars	95	2	2.11%
cat calendars	120	40	30.00%
garden walk calendars	51	0	0%

Fig 3 – Web stats showing visitors and sales over a three week period

For the 12 keywords we are looking at, Fig 3 shows the total number of visitors to the site and the number of sales they delivered. The data shows all traffic - not just that from the Adwords campaign. This doesn’t matter. We are using Adwords Impressions to measure the popularity of different keywords and our web analytics data shows how those keywords convert.

‘personalized calendars’ may have produced the most traffic (1980 visitors), but it did not produce the most sales, with a conversion rate of only 1.16%. ‘Create a calendar’ on the other hand produced the most sales (240 sales) and ‘dog calendars’ produced the highest conversion rate at 67.34%.

Going back to my point about the need for both quality and quantity of visitors, I need to look at both the actual level of sales and the conversion rate. There is no point putting all my effort into a keyword that converts every time (ie has a conversion rate of 100%) if only one visitor a month uses it! A balance must be struck between quality and quantity.

So the data from our very limited initial keyword list, gives us the following information:

Outcome	Keyword
Generated maximum sales	create a calendar
Had best conversion rate	dog calendars
Generated greatest traffic volume	personalised calendars
Had high traffic generation potential (Wordtracker data)	2008 calendars

An Adwords campaign is already in place so now I go back and use it not for research, but to deliver traffic. I now know which keywords are likely to convert into customers so I can now fix my bids based on knowledge not guesswork. I don’t want to waste my Adwords budget on buying visitors that do not convert into customers. For example I will bid higher on ‘dog calendars’ than ‘create a calendar’ because I know there is a higher chance of visitors turning into purchasers.

Natural listings

Always keen to save my clients’ money, my long term strategy is to develop free (aka natural or organic) listings for these best keywords. When deciding on which are the best keywords to use, the argument is rather different for natural listings than it is for pay per click traffic. The cost of developing natural listings is a one-off figure, ie the cost of the time to do the necessary work, and is independent of the amount of traffic produced. So here it is the absolute number of customers a keyword will deliver that is important, more or less irrespective of the total number of visitors, since we are not paying for them individually. So in the calendar site example, I would start by working on developing high rankings for ‘create a calendar’ as this is likely to bring in the most revenue. In other words the conversion rate is less important when traffic is free! Absolute numbers are what counts.

Site performance

‘personalised calendars’ was a popular search term. The Adwords campaign showed this and it was borne out by the web stats. Why then did it not produce sales? The site sells personalized calendars so I must now look at the site itself and do a usability study and see why it is performing so badly for this product.

Potential traffic streams

The Wordtracker data showed that the keyword ‘2008 calendars’ had the potential to bring in the most traffic.
Although generic terms do not usually have high conversion rates, working on a good ranking for ‘2008 calendars’ could be a sensible strategy, though not a priority. It is certainly relevant to the site’s content and if the volume of traffic is high enough, could well deliver a reasonable number of sales. Also, the overall level of traffic is one of the parameters used by search engines to determine rankings, so developing traffic from such a generic keyword would do the site no harm at all.

Just when you think you have finished

Developing long-term good traffic streams is a bit like painting the Forth Road Bridge - just when you think you have got to the end you have to start all over again! Get good rankings for the best keyword, then start on the second best etc. Sometimes, I just go back to the data I generated from the Adwords campaign and the web analytics work. Sometimes I prefer to go right back to Wordtracker and look at a different range of keywords. With the calendar example, I initially decided to keep with the business and mainstream areas of the market. Since calendars are a rather seasonal product, going back and looking at say ‘perpetual calendars’ might be a very good idea for overcoming the seasonality of this market.

Last word

If keyword research is the foundation on which all traffic generation is based, then Wordtracker is the foundation upon which keyword research is based, and it provides invaluable information on the overall use of keywords across different engines and time. I hope I have shown that the results Wordtracker delivers become infinitely more powerful in delivering quality traffic in quantity when combined with information that relates to your own site.

Caveat – the data presented has been selected to explain and clarify different points and is not a complete picture of any particular site.

For some more information on strategies for SEO have a look at this SEO strategy article.