Personalisation goes a long way in outbound marketing, but is not yet applied on a large scale on websites. Even though, by combining data points, you can approach your customer in a much more specific way.
Within three years, I expect all websites with high-conversion aims and/or an e-commerce theme to be 70% more dynamic. We are increasingly gaining insights into what type of user is going to the website, and which data points he has encountered. As a result, from the first visit, you can assess what a prospect is intending to do and more effectively target him on the road to conversion.
A search provides information which allows you to reveal more dynamic content, such as our customer Vacanceselect. Imagine that a person types ‘campsites in Tuscany with a pool’ into the search engine. They will then see a page with the header ‘camp sites in Tuscany with a pool’ and a corresponding photo and text. If the customer was looking for ‘camp sites in Tuscany in the woods’, he would see the same page but the emphasis would be on the content regarding the nearby woodland. The new Artificial Intelligence from Google is used to link the right images to the search term; this realises accuracy of 93%.
You can go even further with personalising the website, if you link information from the CRM about the visitor and from the CMS about click behaviour, with offline visits to stores. This can be done via a loyalty programme or by identifying payments with a hash that also works online. The use of beacons is just getting off the ground; it allows you to ask for visitor emails and match them with your database. You can then create specific audiences on the basis of the profile information in your marketing automation tool. Your own and third party cookies via your DMP complete the picture. Combining all of this information enables you to make the majority of your website dynamic. There are fixed elements, of course, such as contact information and that is why we keep it to 70% dynamic content.
Once you have the necessary data, you must make strategic decisions about what product information you will show. Is a person in the orientation or decision phase? Harder data, such as demographic characteristics and the product categories that are clicked, is useful here too.
We are not outlining a barely comprehensible, futuristic scenario. It is a question of connecting sources to one another. What stops companies getting to work with on-site dynamic content? These are the arguments we often hear and the answers we provide:
The fact that personalisation is good for conversion has already been proven. That is why it is applied on a large scale to email marketing. There are no figures available for dynamic websites but is may be assumed that the personalisation effect would be comparable and that there won’t be any alienation.
Technology has been developed for this purpose. Also, you maintain quality by comparing the performance of your dynamic site with a test group that only sees standard pages. You need good content, and this is often underestimated. The more customer journeys that take place, the more content you need, in the form of products sheets, but also good visuals.
If you work with a legacy system that does not offer the option of a dynamic website, it is a question of budget, not technology. The technology is available. Investing in a DMP, given the price, is not an option for everyone, but a reasonable CRM system will have a click-tracking tool. The option of dynamic content is also available within more and more CSM’s and commerce platforms, but this works on the basis of what the customer has clicked, without considering other data sources. This may mean you overlook opportunities for hyper-personalisation.
This is a valid argument. But you have to shoot, if you want to score (J. Cruyff). There are many commercial sites with predictive recommendations, such as bol.com, Wehkamp and all other large e-commerce players. This, however, is a limited use of dynamic content that is mainly based on the clicked product category. You won’t achieve 70% dynamic content in this way, but you can conduct initial measurements to see if personalisation works, by comparing the relevant conversion with a control group that has seen the standard website.
In principle, you can already personalise the site on the basis of your own data. The broader the customer profile, the better, but third party data is not a hard requirement. You can start small, e.g. by showing personalised recommendations, and then expand dynamic content on the basis of success.
For companies that are able to conquer their nerves, the first step on the way to a dynamic website is to see how many data points you have and how much personalisation you can add to the site. In this quest, it is also important to chart customer journeys and identify which target groups demonstrate which behaviours. You can then place a layer, in which you process the logic, across all data points, so that the data works together. This requires a simple data storage method, otherwise it won’t work. The location at which the data is stored is also very important; it must be easy to read out and visualise data, and then create corresponding reports. This is how you can search for patterns that you recognise yourself or which are analysed by AI. AI must, of course, be ‘fed’ and trained by people; it is not an easy way out but, as a result of its self-learning capacity, will save time in the long run.
Companies that make genuine efforts with regard to dynamic site content, will have to face certain obstacles. It goes without saying that the data must be of a very high quality. It is certainly not the intention that you address the wrong person on the site. If people share a computer, for example, profiles could easily get muddled. An intelligent algorithm, however, can distinguish between these users. And if there is any doubt, you can show generic content blocks.
The algorithm must also be carefully tested before converting the site to dynamic content. Errors are unavoidable with AI because the algorithm is ‘fed’ by people. This is a phase that it must run through.
Another issue is the impact of dynamic content on your website’s loading time. I think this would be ‘low’ as you can address more scripts in real-time. Sending the search to a page and returning the content takes a millisecond so has no impact on the site’s loading time. I also don’t expect it to have a detrimental effect on Google ranking. The content is the same, it is only the visitor that is tackled differently. Google also tends to assess a good user experience as positive. Furthermore, it is unlikely that dynamic content on websites will have a positive impact on SEO, as is now the case with AMP sites.