our views and our knowledge in analytics and other releveant topics

our blogs

Adobe Target Recommendations: does it offer more personalised suggestions than a concierge?


Marketers and consumers are constantly talking about the need for “personalization” – making sure people, in any context, have an experience that suits their needs, desires and preferences. This isn’t a new thing, but expectations have definitely heightened!

The pinnacle of personalized service, may make your mind wander to a hotel concierge – it is their job to predict guests’ needs and ensure they recommend them things that will match their mood at the time. Last month, Time Out had a “things you only know if you’re a…” interview with a top London hotel concierge, he talks of a secret “Golden Keys” society that concierges can apply to be a part of. But wait, to do so, you have to work as a concierge for 5 years, have two member recommendations, and complete an interview. After completing this gauntlet, what’s on offer to them is a world of mediocre: deals on top attractions and the most popular shows. In fact, their “personalized” recommendations run on popularity and price – not necessarily that individual or well recommended.


Enter a whole range of digital products and brands that now pave the way in making people feel like they are being well recommended to.

At Station10, we’ve been spending some time exploring the possibilities Adobe’s Target Recommendations can offer. And whether they can offer a more personalized experience than our “top London concierge”.

The below “how-to” guide assumes Adobe Target has been implemented, and as a business you are clear on what your conversion goals are. Our 5 step process will guide you through the practical set-up of Target Recommendations, ensuring you make the most out of the algorithm(s) for your audiences.

  1. First things first: telling Target the types of web pages on which you want your recommendations to appear. Do this by setting up the page configuration for your Recommendation: create a template rule for your overall website URL and then add page delivery specifics using additional domain, path, query or mbox parameters. These page delivery specifics are crucial for ensuring Recommendations will be served on the correct parts of your website.
  2. Red-herring alert: as you create and configure your Recommendation page(s), Target will give you an option on where you’d like the recommendation to be displayed. So far, so sensible, but it’s not strictly true. Wherever you choose to “place” your recommendation will be over-written by the code you create and select at step 4. The only pre-select option to worry about at this stage is whether you want your recommendation to replace what’s already on the page, or just be inserted above or below other page elements.
  3. Getting your Recommendation “Target” right: here is where you actually build out your recommendation algorithms and make sure they are published to the correct audience(s).
    1. Audience: if you already have audiences created in Adobe Marketing Cloud they can be applied here. Alternatively, audience scripts and profiles can be created in Target. The latter option is particularly useful if the URL profile can be matched to part of the product criteria, in order to limit the product recommendations returned.
    2. Algorithm Criteria: here is where you create the logic that decides what criteria needs to be met for a product to be recommended. Their “off the shelf” algorithms are to be expected: top viewed products, people who viewed this viewed that, people who bought this bought that and items with similar attributes. However, they can be tailored using filtering and weighting rules to give them more specificity for your customers. For example, matching specific entity or product information, or attempting to up-sell with pricing rules.
    3. Control Group(s) and Splitting: like any good testing tool, Adobe Recommendations allows you to specify whether you’d like to split the recommendation criteria shown to audiences, or put in a control group. Highly recommended if you’re looking to work out which Recommendation algorithm will work best for your audience.
  4. Design specifics: Target Recommendations will now prompt you to insert a design for your Recommendation. Create this script using Velocity language and ensure the specifications given match the rules you want your recommendation to follow (e.g. number of recommendations returned). It will need to speak to your site’s CMS components and this will tell Target where your Recommendations should be published on the page(s).
  5. Setting your Recommendation goal measurement: the final step is to help the reporting in Target know what you’re looking to measure. Choose from conversion, orders, page views or any other suitable analytics metric and specify the desired action. This measurement setting is for you to quickly view a performance report in Target, Recommendations links to Adobe Analytics where you will be able to look at additional success metrics.

Your recommendation(s) will then be ready to go live, and this is when you can start getting clever with refining what will be shown to your audience(s). Specifically testing variants of the same algorithm: playing with product classifications, attribute matching and pricing structures – to see which recommendation feels the most personalized for your audiences and gives you the conversion you’re looking for.  And that’s when things start to get fun.

What’s interesting is that the Recommendations part of Adobe Target is often overlooked; most people think of it as just a straight optimisation tool, and often companies will have a separate recommendations engine.  But this adds a significant string to the product’s bow.  It’s often regarded as difficult to set up and configure.  And whilst this isn’t as simple as just flicking a switch, as some people might want or even expect, it’s akin to the sort of set-up one would have to do on similar tools.


However, if you do find the details above a bit too involved, there is an alternative; get in touch with Station10, and we can help you get the most out of your recommendations engine.