How we use Google Analytics and Website Optimizer
On October 21st, 2009 by Duane Jackson
If you’re building a business that uses a website to sell a product/service or to collect information from visitors then you’ve probably found out that getting people to your website is just the first step in a long process.
Once they’re on your site you need to make sure they do what you want. That might be buying a widget, subscribing to your mailing list or – as in our case – registering to try the software.
For us there’s a further step – converting the trialist to a paying customer, but we’ll leave the detailed discussion of that for another time.
So once you have some traffic coming to your website the next thing you need to do is find out how many of them are doing what you want them to do – “converting” – and then work on improving that conversion ratio.
Goals and Funnels
Initially we started using a free service from Google called Google Analytics. By putting a small piece of javascript on all of your web pages you can get some useful insights into where your website visitors come from and what they do once on your site. You can then set up numerous “Goals” and measure how many people complete those goals. A goal being the desired action such as buying or registering.
You can also measure effectiveness of what Google call a “funnel”. A funnel is usually a goal that consists of a number of steps/pages.
So for us we measured how many people go to our registration page and then how may of those actually complete the registration page. It turned out that of everyone that visited our registration page, only 45% of them actually went on to complete it. So over half of everyone that looked at our registration page sailed off into the sunset never to be seen again.
We’ve managed to gradually improve that to almost 70% by trying a few different things.
- advertising a free e-book on marketing if they completed the page
- stressing that we wont spam them or pass on their email address to third-parties
- splitting the form in two so the first page doesn’t look so daunting
The free e-book never made much of a difference. It was 2) and especially 3) that made the big difference.
Feeding the Funnel – Measuring effectiveness of different headlines
With the registration funnel now converting at a respectable level, the next thing we had to do was improve on the number of website visitors that went into that funnel – that is, actually go to the registration page.
At first we changed the headline and/or other elements on our homepage and watched Google Analytics to see what happens to the overall Goal conversion rates after the change. The problem with this approach is that there are often too many other factors that interfere with the experiment.
What we really needed to do was a split-test. In a split test you test two pages side-by-side at the same time. Half of your website visitors see one headline and the other half see another headline. You can then measure them against each other without other factors being involved.
Google Analytics has nothing to help here but a different free tool from Google does, enter Website Optimizer.
There are two options. A simple A/B test and a more complicated Mulitvariate experiment. We went with the former.
We created two versions of our home page with just one difference – the headline. Our original said “Easy to use, jargon free Accounting Software” and our alternative page said “The easier to use alternative to Sage and Quickbooks”. Website Optimiser then led us through setting it all up.
All you need to do is paste some javascript code into the two pages and on your conversion page (for us, that’s the page you see once you’ve completed the registration). Google takes care of the rest.
Now a large dose of patience is needed. You need a high volume of traffic and conversions to get a meaningful result.
In the above graph you have time on the x-axis and conversion ratio on the y-axis. It shows that over time the new page (orange) consistently outperforms the original (blue) by an average of 20%.
Interestingly this also demonstrates what I was just saying about external factors. There was a general upwards trend throughout the month on both pages. If we’d just switched from one page to the other and looked at the Goals in Google Analytics we’d have concluded that there was a larger improvement than there actually was.
So we’ve now dumped the original headline and adopted the alternative headline instead. We’ll now start experimenting with other aspects of the page, or maybe other headlines to see just how high we can push the conversion ratios.
Simultaneously we’ll continue working on improving the conversions of trialists into paying customers, a ratio we’ve improved by 100% already so far this year.
Kicking myself
I’m really pleased we’ve found the time and tools to do this. What really irks me is that we didn’t do this ages ago. I could sit down and calculate what our revenues and customer numbers would look like if we improved conversions like this years ago – but I’m scared to.
Every day that you’re not actively working on improving your conversion ratios is a day of lost opportunities.
![[Post to Twitter]](http://blog.kashflow.com/wp-content/plugins/tweet-this/icons/tt-twitter-micro3.png){kind=icon}
![[Post to Delicious]](http://blog.kashflow.com/wp-content/plugins/tweet-this/icons/tt-delicious-micro3.png){kind=icon}
![[Post to Digg]](http://blog.kashflow.com/wp-content/plugins/tweet-this/icons/tt-digg-micro3.png){kind=icon}
![[Post to StumbleUpon]](http://blog.kashflow.com/wp-content/plugins/tweet-this/icons/tt-su-micro3.png){kind=icon}
Tags: conversions, google analytics, tools, website optimizer
October 21st, 2009 at 6:10 pm
[...] Duane Jackson, CEO of SaaS accounting provider Kashflow writes up his experience of using Google Analytics and Website Optimizer to fine-tune his site to increase conversion: [...]
October 22nd, 2009 at 7:42 am
Website Optimiser is a great tool – but what this does really point to is the importance of so-called “split testing” – where two (at least) versions of something are tested against each other.
The “something” might be an entire webpage, a layout, a way of providing a form for completion – or a block of sales copy – almost anything, but only by comparative testing (an arduous but worthwhile process) can we achieve better results.
I keep hearing storys recently about eCommerce shop owners who have changed the look and feel of their entire website and suddenly a successful site has nose-dived.. cautionary tales at best!
Great stuff though and keep it up – would be very interested to hear a little bit about what you conciously do to help convert trial users into paid users (beyond the product itself)!
October 22nd, 2009 at 9:06 am
I’ve been considering using the website optimizer for some time – but never got round to it – now you’ve convinced me.
I’d be interested to see – although I would not expect you to publish – if the increased trial sign ups also read to increased end sales.
We have found in the past that increasing the number of trial sign ups can simply drain your sales resources if those signing up are not really interested in your product.
I would guess in this case that by putting the “big brand” names at the top you have told people exactly what you do – which should prevent time wasters.
I notice you also sometimes use Google ad words. We did do some optimization around this (or rather Google did) and managed to increase our conversion (both goal and real) and 1/2 our spend.
Have a look here for a rough guide.
http://www.fusionaccounts.com/Blogs/tabid/100/EntryId/53/What-has-Google-ever-done-for-you.aspx
Thanks again for the advice!
October 22nd, 2009 at 9:17 am
Hi Stuart,
There’s always the danger that if you pump low-quality into the trial pipeline then you’ll decrease your trial-to-paying ratio.
But that’s not what we’re seeing. Ratio of trial-to-paying is gradually increasing as we improve the processes we have in place and also as the product is continually improving (and therefore of use to a wider audience). There’s still a lot more we can and will do on those fronts to further improve on the conversion rates.
Is there a human/time cost for you guys to convert trialists to paying? For us there isn’t much involved (besides support, who would only be involved if the trialist is actively using and interested int he software) as we don’t pro-actively contact triallists. There’s just way too many of them every day and too low a monthly fee to make it practical or economical.
We do very well from organic traffic from Google.
We work with a *very* good external firm to manage our adwords. We stopped it for a while as the cost of a customer acquisition was too high because the conversion rates of vistor->trial->paying were too low.
We’ve recently re-started it again because now we have improved conversion rates through all the different stages all of a sudden it’s very cost effective.
October 22nd, 2009 at 10:49 am
Duane,
No – we don’t actively try to convert trialists. They get a couple of automated emails telling them how to contact us and showing them how to get started
We find there are 3 main types of “trialist”
1) Sign up, enter 100 transactions in the first day and use forever.
2) Sign up – contact us for some advice on functionality or to do a data import – usually Sage but often price list/stock imports
3) Sign up and do nothing.
We did, for a period, try and find out why some people had signed up for a trial and done absolutely nothing. We wanted to know why? Were we missing something?
This was a lot of effort for very little reward. Those who we did manage to contact seemed to have little recollection or reason for having signed up. Most simply ignored us.
So now, like yourselves, we concentrate on not only getting more traffic, but also the right kind of traffic.
Stuart
October 22nd, 2009 at 10:51 am
Interesting. We did a test a few months back where we phoned EVERY contactable trialist for that month with a “how are you getting on, anything we can help with?” call.
Positive responses from just about everyone, but no marked increase in conversions.
October 22nd, 2009 at 11:23 am
All sounds familiar! Stuart’s comment at the end is spot on – the right kind of traffic is essential.
By using Google analytics and webmaster tools, plus cute SEO work you can drive more targeted visitors to your site. (And this of course applies to any online business.)
Visitor numbers may well suffer, but the effect is beneficial as the trial-to-pay ratio necessarily increases as the propensity to purchase is greater with these visitors.
I think Duane has a good point too about contacting the trialists. Gets positive feedback – perhaps to avoid a “sell” to use the software – but we’ve found the same result, or lack of result.
October 22nd, 2009 at 11:34 am
Getting the right traffic is essential – 10000 irrelevant hits do nothing but waste resources (maybe one or two would convert if you were lucky OR really, really good).
Essentially it can in some cases boil down to an admittedly complex ROI equation – what amount of effort or investment is going into obtaining and dealing with each “trial user” (nearly did trailer there!) – and what is the associated return if they sign up.
The critical thing is to get the balance right, so you maximise conversion but without having to put in so much work that it’s not worth doing!
Re: Adwords, people often go wrong by simply chucking cash at it – its essential to tweak, adjust and re-work until you can prove that running an advert is profitable – then scale up/out appropriately. As Duane notes – once you get it right it can be very efficient!
November 16th, 2009 at 7:08 pm
We stopped using Adwords a while ago, prices got out of hand. To advertise a 100% free service on certain keywords it would cost $6/click, i just can’t believe this is true… while the whole process is not transparent.
February 5th, 2010 at 1:50 am
@Duane,
A couple of points on converting trial users you could consider.
Offer a reduced price for 1st year if they commit within 30 days, stressing the initial 60 days remains free.
Send a mini survey after 45 days of the trial period, with simple tickbox answerable questions asking if they will be signing-up soon, or possible reasons why they won’t
February 5th, 2010 at 12:00 pm
Good suggestion. I particularly like the second one. Watch this space!
February 5th, 2010 at 1:26 pm
Dang! as a current trial user I preferred the 1st one ;-)