In today's retail world, AB testing and website optimisation is a vital part of ensuring traffic coming to an online retail business converts. Now, there's finally a way to scientifically experiment on your product data for product ad visibility.
Time has progressed, and online has become the typical consumer’s first touchpoint. Instead of going to the brand or retailer's website, its common practice for consumers to go directly to offsite ecommerce channels (Amazon, Facebook, eBay) to both research and purchase.
This means that in the same way retailers AB test and optimise their own websites, testing and experimentation is the key to offsitesuccess and driving performance through these channels.
In the previous article of this three-part series, we talked about the five core channels from which you can generate multiple streams of revenue:
- Paid search
- Comparison shopping sites
- Paid Social
Using Product Listing Ads on Google, Amazon Ads or Dynamic Product Ads on Facebook not only generates more revenue, but also increases brand awareness and customer acquisition as your products are exposed to large audiences.
Many of these channels (Amazon being a key player), are where consumers begin their product searches.
None of this means anything unless your product data is complete, optimised and being managed continually, which was the subject of our first article.
If you're still fixing and optimising your product data, take a look at our earlier article before continuing this one.
Taking your AB testing offsite
Onsite experimentation is widely practiced in order to drive clicks and conversions on retailers’ websites and is considered an integral part of ecommerce.
Entrepreneur, Jared Goetz, who managed to grow his online store from $0 to $2 million in two months said: “You don’t know what people will respond to until you try a lot of things. Don’t be afraid to think outside the box.”
This is good advice, given that consumer behaviour is continually changing and although he isn’t talking about offsite channels, the same logic applies.
Whereas the optimisation of product data can do a lot for you in terms of performance, perfect product data cannot be achieved without an in-depth knowledge of:
1. What terms and phrases to use in titles based on consumer search habits
2. Which images drive clicks and conversions
3. Which categories are most used for specific products
The first can be somewhat achieved with some SEO research, however, this can be an arduous task when you have thousands of products.
Product content experimentation is the third step to product visibility and is how retailers can be sure that these three questions are being answered.
Furthermore, experimentation is a great way of ensuring your offsite campaigns are driving maximum revenue and customer acquisition without increasing your ad spend.
Do I need to experiment with product data?
Sophisticated offsite experimentation is just as essential as website testing, driving peak performance on every single ecommerce touchpoint.
Gaining that scientific, evidence-backed insight is invaluable when making informed decisions and ensure you square up to the competition, or even get ahead.
Whether you’re performing well, poorly or just not as well as you think you should be, every retailer, big or small, can up their gain by experimenting on product data attributes.
Intelligent Reach’s Experimentation Module is an invaluable tool for digital marketing and ecommerce teams. In the following section, we will explain what elements you can experiment with and the common challenges facing retailers that experimentation can help with.
What to experiment with and what problems do retailers face?
Every channel has specific requirements for your product data, including titles, images and categories.
Conforming to these requirements will prevent disapprovals and increase visibility, however, you could be driving even better performance by knowing exactly where and what your ideal customers are searching for as well as in which categories.
The title is the single most important element of your product data as this is what will primarily increase visibility and relevancy. Retailers often struggle with what to include in the title and in which order to list the different attributes.
There are many options to consider, and the differences withith the channel you’re listing into makes it even more confusing.
Dilemmas retailers face:
- Should you include the brand name?
- What colour would you search for? Is it best to be simple or more specific?
- If there are multiple names for something, which do you go for? E.g. sneaker or trainer, flip flop or thong.
- Which attributes are most important to consumers? Should I include size?
You’d think the image would be easier, but since you can only include one with your product ad, it’s important to not only pick the one that will drive the highest clicks, but also give an accurate representation of the image at hand.
It’s no good to deceive the consumer with an image that doesn’t properly represent the actual product because this will damage brand trust.
Dilemmas retailers face:
- Which angle to show of the product. For example, a close-up of a dress or have the whole model in the frame?
- Should I include a cut out image or a model image
- Which image should I use on which channel?
- Will the chosen image be approved?
When consumers are searching for products, some will go directly to their desired category.
It’s important to firstly, choose the correct category – but this will most likely be obvious – and secondly, find out if there are multiple categories within which you can list, if there are, you’ll need to discover which ones your ideal customers are going to most.
Dilemmas retailers face:
- Are there multiple categories within which you can list your product?
- Is there a category for your product on one channel, but not on another?
Types of experiments
The most common method of testing, especially with onsite content. This is done by measuring two groups (a control group and a modified group) of product content alongside each other.
Multi-variate testing (MVT)
This is a method of testing whereby multiple variables are modified and tested at once, with the goal being to determine which variation performs best.
Before and after
A Before and After Test, like the first, examines two groups of data but separately; first, unchanged for a period of time, then modified for another.
As it stands, Intelligent Reach is proud to say that it is the only SaaS platform of its kind offering sophisticated experimentation for offsite product data.
If you’d like to find out more about how we experiment with product attributes or even about the subjects mentioned in the previous two articles, please get in touch below. We’d love to discuss your requirements.