We are often asked to help our (prospective) clients understand the difference between various types of guided selling apps. How does one compare to the other?
To make everyone's life easier (after all, we're in the business of making lives easier), we've decided to write down our own thoughts on the most common type of guided selling app: a filtering assistant.
In this article, we're breaking down the pros and cons of a filtering assistant using a best-in-class example. For this we'll use the filtering assistant of Coolblue, the 2nd largest e-commerce player in The Netherlands, with more than €1.4 billion (yes, with a b) revenue in 2021. We'll consider their filtering assistant for televisions — the company's largest category, and a product that every consumer knows.
So basically, we're stacking the deck against ourselves: it's one of biggest e-commerce players, with virtually unlimited resources, their biggest category and a well known product. This guided selling app should work perfectly.
How it works
Each television comes with technical specifications, also known as product data, provided by the producer. These specs are presented as filters on the lister page of each product category:
To help customers figure out which televisions are relevant for them, Coolblue has created a filtering assistant that present choices regarding the product (specifications) in a conversational (question-by-question) manner:
In the example below, users are asked a question on the desired screen size of their television, and given three answers to choose from: Small, Medium-sized, and Large. Each answer is further elaborated on through some textual and visual content:
The answers relate directly to the available product data, meaning that each answer will automatically select the corresponding product filter on the lister page. Once every question has been answered, the 'product advice' is complete: the lister page has now been filtered and shows only the television that match the selected criteria:
In short, a filtering assistant asks the customer several questions that relate directly to the available product specifications, which are applied as filters on a lister page.
Let's break down this approach. What works and what doesn't?
Pro: it's easy
Asking a few quick questions is a pragmatic way to help your customers find their right product. It beats offering no assistance at all by a large margin.
Pro: It's scalable (*)
With a filtering assistant, you're relying on your product data as the go-to source for assisting your customers. The big advantage is (or should be) that your product database already exists. You've invested a lot in building it, so it only makes sense to put it to good use.
So, in theory, relying on your product data should work and scale. As new products enter your shop, their product data is uploaded to the database, where the filtering assistant can immediately access it and take it into consideration for its product advice.
There is, however, a major caveat. Having access to complete, correct and consistent product data is an ongoing concern for most retailers. If your product data is not perfect, your filtering assistant will not be able to properly assist your customers.
Con: It's about products, not people
Product specifications are the technical, objective characteristics of a product. For televisons, these are a.o. the amount of colors, the refresh rate, color depth, sharpness, weight and the year the product has been introduced.
There's just one problem. As a layman, such (technical) specifications don't mean much to you. A television may partially appeal to someone's imagination (and higher/more is better, right? 🤷♂️), but what about, say, an ironing machine? How many grams per minute of steam production would you like? Or, if you're looking for hiking boots, do you know the implied difference between a leather, synthetic, suede and nubuck finish?
You need to either be, or become, a product expert to understand these technical specifications and their implications before you can match them to your needs.
Another example. If you want to buy a new pair of headphones, you'll have to choose from specifications such as 'wireless' and 'noise-canceling'. As a non-expert, you'll just happen to have to know that both these specifications will make your time on the train a lot more enjoyable.
Product data tells you the what of a product. So whereas a filtering assistant might help you to present choices regarding the product specifications in a different (conversational) way, it's very limited in actually helping customers translate their intended use cases to these data points. So it's not about the why.
Con: it's binary
Good product advice is nuanced. It's usually a tradeoff between different functionalities, aesthetics and prices. Depending on the customer, some of those factors will carry more weight than others.
Filtering assistants lack nuance. That's because product data is binary: it's either true or false. If a customer is looking for a 'television' that cost 'under €1.500', has a '55 inch minimum' and has 'ambilight', only the products that fit this exact description will show up in a filtering assistant. However, if there's a perfect alternative that costs €1.501, it wouldn't show up, because it doesn't fit in the applied filters.
This type of binary data is also problematic if you want to check preferences instead of requirements. The color of a device, whether it's lightweight, or whether it's by a specific brand: these are all preferences. You may prefer Samsung television, but you would perhaps be willing to try a different brand that meets your other requirements. Filtering assistants lead users down a funnel in which such soft preferences aren't taken into account. So, even in the Coolblue filtering assistant, there's no way to mark options such as a soundbar or ambilight as 'nice to haves' or 'optional'.
Finally, if multiple products meets a certain criterium (e.g. comes with a soundbar), a filtering assistant cannot further differentiate between them. Every product that shows up has that specific capability. Then why does one of them cost €1.339 and the other €2.569? Supposedly (hopefully) there's a difference in the execution of this functionality. But these differences aren't apparent to the customer, and he'll need to further investigate what sets each product apart and which one fits his exact needs. That's not real product advice.
Con: you're not solving your customers needs
Imagine: you're in a store. You know, one of those built with bricks. You've asked a sales representative to help you find the right television. She asks you a bunch of questions:
- What are you going to use the television for?
- What size television are you looking for?
- Do you have a preference for a brand?
"Thanks," she says. "Give me a moment and I'll show you which TV's are just right for you." After a few minutes pass, she returns and takes you to the storage room where she presents you with 80 televisions. "These are all perfect for you!" When you ask which one to choose, she tells you that she sorted them by relevance, but that you can compare their specifications to make up your mind.
This is a ridiculous interaction, but still the best that 25 years of e-commerce has to offer.
Here we see the wrap-up of Coolblue's filtering assistant:
A whopping eighty televisions have been selected, which are apparently all perfect for "streaming movies", have "excellent image quality" and are "large" (>65 inches). The price difference is a mere €28.530 - or 2000% - between the cheapest and the most expensive one.
And these are the first few product results:
How does this help a potential customer? How can they make up their mind between these five, let alone the full eighty models? There's still a substantial amount of digging to do to understand which model suits their needs and wants best.
What's the alternative?
Using a filtering assistant virtually ensures that you're still not solving the number 1 problem that your customers have: choice overload.
Customers will buy from you once they're convinced that a product will work perfectly for them. Hence, your goal should be to cut the clutter: reduce information, get rid of product overload, and cure anxiety and stress. A filtering assistant doesn't achieve this to nearly the extent that is needed.
That's because a filtering assistant is based on the needs of a webshop: "We have a bunch of data readily available, and if our customers would just select the right specs, we can conveniently point them into the right direction, and sell more".
A true guided selling app is based on the needs of a customer. So:
- Ask them questions about their intended use and goals. Such questions are actually easy, and fun, to answer.
- Create a nuanced experience that differs between 'must-haves' and 'nice-to-haves' and circumvents the pitfalls of product databases.
- Offer guidance on which products attributes are a 'great' fit versus simply 'good' (e.g. 5% over budget isn't great, but probably still good).
- Show a limited set of results, and tell a customer how and why those products are a good match for them. Doing so actually solves the pain and the problem a user has.
You expect all of the above in a store, so why not an online store?
We've built Aiden to help build guided selling apps that deliver — for the customer, and for the business. And it works. Check out our case studies that show how our approach to guided selling results in conversion uptakes of up to 500%.