CRO, or Conversion Rate Optimisation, is simply about optimising your site so that users will perform a conversion action.
This is easier said than done though, as there are a range of things which can be changed on your site to help improve your conversions. Unfortunately, many aspects of CRO are hit and miss, but you need to find out what works badly to get to the good stuff!
The basic outline of performing CRO has a few steps:
- Decide what you want as a conversion.
- Create a baseline statistic.
- Make alterations.
- Monitor results.
After this you can then make educated decisions on what performs well and what does not. This can either be integrated into your current site or help with future developments. It is worth noting that what works now may not in the future though. A site optimised from 2005 wouldn’t convert very well today.
What Is A Conversion On Your Site?
This might seem obvious to sites such as ecommerce, but if you are lead driven or informational, it isn’t as clear cut. Even if you are an ecommerce site, you might be looking into conversions from a blog post or non-product page rather than transaction numbers.
For informational sites this can be interactions with videos, scrolls down the page, clicks on accordions or more information, etc.
Lead based sites will probably revolve around contact form submissions, further information, file downloads, etc.
Whichever one it is, you can look big or small for CRO. Micro conversions for retention and awareness are just as valid for CRO as direct transactions.
Creating A Baseline
Once you’ve decided what a conversion is, you need that baseline statistic to see where you can improve. Ideally this will be something you’re already tracking and have been for some time to gain historic metrics. If not, don’t worry, but try to keep in mind seasonality and other factors which could alter your baseline figures.
Traditionally we’d say to go for 12 months’ worth of data to get a good range and remove potential seasonal fluctuations, but with 2020 being as it was, this might not have much weight.
Again, these pandemic caveats need to be accounted for, so pre-2020 data is ideal.
This is the main aspect of the activity and will probably require development time to make changes, unless you are using a landing page builder type service.
A great way to find out how users interact with your site and what they find annoying is through anonymous polling. This gives you a great place to start with your CRO work. Trying to get information from users who don’t convert is just as important as those who do, so try to keep this in mind.
It can also be difficult to see issues with your site if you are used to it, so try and get as many people’s opinions about what is and isn’t working. A good spread of technical ability is also good, as someone in digital marketing can probably work with a less friendly interface than your average user.
Here are a few examples of areas to look into:
This can often be overlooked in terms of CRO in favour of style and branding, but it is the first impression you give. Clear links and purpose are the best routes for CRO. This will undoubtedly conflict with design, but let the figures speak for themselves!
These should be easy to refine for users and not be too overwhelming. Filters not working as intended, overwhelming categorisation and other issues can cause users to miss products or become frustrated and leave. Tracking interactions with these elements is relatively easy in GTM, and frustration clicks can be easily seen.
As above, broken elements are often seen on product pages. Duplicate information, lack of specifications, erratic stock levels – all can cause issues with conversions.
If you have modifiers within one URL it can lead to less reliable data, so this should also be considered. E.G: if a t-shirt is on one URL, but it has different sizing and colours with some in stock and others out of stock, it is harder to discern the issues compared to individual product URLs.
Abandoned carts and checkouts are frustrating to marketers, as this is the final hurdle. There are multiple points of potential failure on these pages – logins, address fields, payment options, payment gateways, etc. It’s hard to optimize for all at once on this front, but you can still get anonymous data on how far down a form users get or what stage in the checkout they drop off on.
Service pricing pages are a mix of emotional and information-based factors. From changing your price points lower or higher (yes, sometimes a higher price point encourages more trust and conversions!), or choosing a different colour for your “buy now” button, it can feel like a bit of a black art.
Posting on your blog does more than keep a stream of fresh content, it also gives answers to questions and engenders trust. At a basic level you should be linking back to products and services you talk about, but you can also experiment with different CTAs. These could be links in body text, links in sidebars, forms underneath content or other methods. Make sure that you aren’t just building content solely for ranking purposes and put it to good use.
Keep your initial guess as to what will happen as just that – a guess. Rely on the data that you obtain but try and be critical of it to find holes, rather than enforcing what you want to believe.
Unless you are working on a larger site it can be hard to get meaningful amounts of data, which is probably where a lot of people trip up. Try and leave your tests running as long as possible to get the data to make decisions on and discuss it with your wider team. The data won’t lie, so you need to be objective on what you are told and make your permanent adjustments afterwards.
Third Party Tools
There are numerous third-party tools to help with tracking, heatmaps and recording – including the new Microsoft Clarity script. This is basically a stripped down version of Crazy Egg or other heatmap software, but is definitely worth a try.
With a combination of an analytics platform, a tag manager system, variable A/B testing and some good results, you should have everything you need. The go-to is Google Analytics with Google Tag Manager to get the data you want, but there are numerous alternatives out there.
I’ve mentioned a few caveats around historic data and timing your tests, but these are worth reiterating. Compare times which are like for like and take into account other offers, sales and promotions which could affect your data sets.
Also don’t take anecdotal evidence to heart. There are numerous studies out there, but you won’t know what is going on in the background. Your site is unique and will have different interactions, client bases and other factors which will make even comparisons within the same industry unreliable.
This is a lot to complete, and many companies need guidance with set up, implementation and interpretation – and we can help.