Google Conversion Room Blog Tips on tracking and improving conversions online

Don't Rely on Conversion Rate Alone

Wednesday 16 September 2009 | 18:15

Most website owners focus a lot on conversion rate. This is understandable. They want to see how many visits they can convert into paying customers and try to improve this ratio over time. After all that is what this blog is all about. However focusing on just conversion rate can be a dangerous thing to do. This is especially true when it comes to evaluating your sources of traffic.

Let's pretend we are an online marketer for a travel company, 'Watersports Holidays'. The strategy we have adopted is to review our Google Analytics data and invest more money in the sources that convert well, while taking money away from or optimising those that convert poorly. On the surface it sounds like a sensible strategy. So we log in to our Google Analytics account and we start reviewing our traffic sources. We are going to focus on our AdWords keywords but really this can apply to any traffic source.

In the report Traffic Sources > Keywords > Goal tab/ecommerce tab paid keywords we find two contrasting keywords. "Windsurfing holiday Fuerteventura" is doing great, it has a conversion rate of 1.4%, twice the average conversion rate for this site. It is one of our top performers so we decide to up it cost-per-click in order to push it higher in the ad auction and get more traffic to it. On the other hand we have "Watersports holidays Europe" which has a conversion rate of 0% for this period. It seems clear to us that this keyword is a poor performer and we immediately delete it.

We continue this process over time and we find that although the conversion rates of the remaining keywords are relatively high they are generating less revenue and overall sales are down. What have we done wrong?

Let's step back and put ourselves in the shoes of one of our customers. We are considering a holiday. We are going with two friends and want to go somewhere sunny that is great for watersports. So we do a search on the keyword "Windsurfing holiday Fuerteventura", go on to the Watersports Holidays website and buy immediately, right? Of course not, almost nobody is this impulsive in the real world. We start by researching a few possible destinations, we pick a destination and at that point find the Watersports Holidays site. We show our friends the site, decide on our package and make sure our dates are available. The next day at lunch we go on to the site and buy the holiday package for the three of us. At this point we've been to the site a number of times but Analytics attributes the conversion to our most recent source. If we judged our other keywords by their conversion rate alone we'd be removing keywords that albeit don't often contribute to the final conversion, are important in sparking initial user interest.

The second danger in looking at conversion rate alone is that we focus on high conversion rate but low volume keywords. Often our most important keywords have a conversion rate that is OK or worse when compared to the average but deliver a large portion of the revenue. When we focus on rates (or percentages) at the expense of the actual volume numbers, we sometimes miss the bigger picture.

So what is the solution? We need to focus on more than just conversion rate when evaluating our traffic sources. When we see a keyword or source that has a low conversion rate we need to do three things.

Firstly and most importantly check the margin and revenue metrics for this keyword. Is it making a positive contribution towards revenue? If so, by how much? Keywords that bring in revenue and make us a profit should be valued whatever the conversion rate (and all the more so if they generate high volumes of conversions).

Secondly we need to evaluate based on multiple metrics rather than conversion rate alone. If we take our "Watersports holidays Europe" keyword let us take a look at how it performed for two other key metrics; Bounce Rate and Average Time on Site.

When we evaluate our report we can see that 'Watersports holidays Europe' actually performed pretty well in terms of both bounce rate and time on site. It has a Bounce Rate considerably lower than average and a Average Time on Site double that of the average. Given this and the nature of the keyword itself (it is quite general and more likely to find prospects who later come back to buy) it would make sense to retain this keyword despite the low conversion rate.

Thirdly if we find our keyword has a low conversion rate and also underperforms in terms of other metrics such as time on site and bounce rate we need to look at optimising this keyword. Is it relevant to our products/services? If so is it matched with a relevant ad landing page? The more we preserve a consistency from keyword, through ad to landing page the more likely we are to pull users deeper in to our site.

In conclusion, there's no doubting conversion rate is a useful metric to understand the proportion of visitors who become buyers / leads on a site. However, it should not be used in isolation.

If you have feedback on this post or any other posts please drop us an

Posted by Brian O'Sullivan, Google Analytics Team.


ClickEquations said...

Good point. What's really needed is the ability to attribute some of the revenue from that final click/purchase back to earlier keywords in the purchase cycle - this gives a more honest value of keywords which aren't 'closers'. In ClickEquations we show you the revenue (and assist count) for every keyword clicked before the final conversion. Details here:

conversion_guru said...

This is something most digital marketers are aware of but because clients are often groomed to focus on conversion simply assuring them that "Watersports holidays Europe" will contribute to an eventual sale. Yes it makes sense but clients want accountablity and conversion.

Subhash Surampudi said...

Good post.
I will only add another point to what has been said:
Look at additional searches in the same session. I think, that more than anything tells us about the effectiveness and relevancy of the ad. In fact, I wonder why that isn't a direct metric on GA :)

- Subhash