Personalisation is a great way to target different consumer segments or even individual shoppers with tailored content that improves the user experience and increases the likelihood of a conversion.
It's a topic we've looked at in more detail in blog posts about the ROI of personalisation and four tips for improving website personalisation.
Online fashion retailer Stylistpick used personalisation to increase conversions by 33% among one customer segment.
To find out how it achieved these results and more about its personalisation strategy, I spoke to Stylistpick's Ian Sutherland...
1. Have you developed personas for different types of shoppers? If so, how many do you have and how does it help you to improve conversions?
Personalisation sits at the heart of Stylistpick's proposition and we are always tweaking and improving our work in this area.
Fundamental to the strategy is our Style Quiz that encourages customers to discover their own personal style. We know women love quizzes in fashion magazines, and our quiz is both entertaining and engaging, so it's real customer benefit.
Behind the scenes, our algorithm analyses the output of the quiz and enables us to create personas that we then use to deliver personalised showrooms of products each month.
Today, our approach to personalisation is developing beyond the showroom concept. We now use a personalisation platform to identify customer segments from patterns of interaction with the site over thousands of visits.
So, rather than just developing personas ourselves, the technology creates an endless number of user segments based on how our customers behave on the site.
We can therefore target women based on their actual interaction with the site and we are seeing some great early results from this approach.
2. Are different types of data (e.g. demographic information, browser history, on-site behaviour) more valuable for determining what type of content should be served?
Any type of data can potentially be used in order to personalise content delivery, but what makes the difference is sample size - the more instances of one data type you have then the more accurate you can be in your personalisation assumptions.
It's why big data works - by collecting vast pools of data about millions of interactions we can draw really valid assumptions about behaviour based on a large, reliable pool of information.
The main data silos we're leveraging for personalisation are around user behaviour on the site, because that's what gives the clearest messages about a user's intentions at any given time.
3. Some types of personalisation are more obvious than others, e.g. product recommendations based on purchase history. What personalisation techniques do you use on-site that wouldn't be obvious to the user?
Product recommendations are just part of our approach – it is possible to make assumptions based on data about what a person has done historically but there is so much more scope for personalisation.
The sort of model we are moving towards is much more subtle - for example we will soon be able to tell when a customer is showing signs of confusion.
Say our customer doesn't understand our shipping policy – very soon we will be able to dynamically serve her some prominent messaging about the topic.
She won't know that she is being targeted or why, from her perspective we're supplying an answer to a question she didn't even know she had.
4. How do you capture personal information from users? For example, do you force all new customers to create an account?
Personal information can be useful, but if you just look at that then you're limiting yourself to understanding people who have registered, who are largely people who have already purchased from you.
That's ignoring the whole universe of people who visit your site but don't purchase - that's a much bigger audience and one that you can learn an awful lot from.
5. What systems do you use to collect data for on-site personalisation? Do you use Google Analytics?
Other than our in house developed Style Quiz - our primary tool for on site personalisation is the QuBit Platform.
They provide us with the whole package for analysis of customer behaviour, from data collection and analysis through to powering dynamic personalisation on the site and even testing campaigns to ensure that they're based on valid assumptions.
We've been using this as a managed service to data, with QuBit doing much of the legwork, but their new self service functionality is going to let us bring a lot of this in-house, from analysing the data right through to creating, testing and rolling out campaigns.
6. Can you give us any examples or case studies for a personalisation initiative that had a major impact on conversions?
We had a great example recently where we identified that people who visited more than one category page without putting something in their basket were highly susceptible to discount offers. We also identified that Chrome users were the most price sensitive of any browser.
We therefore built a campaign that offered a targeted 25% discount to Chrome users who visited more than one category page without adding something to their basket and we saw a 33% increase in conversions amongst this segment.
What's doubly powerful is that this not only generated additional, profitable sales but it also meant that we could stop offering blanket discounts, so we increased margins on a whole lot of other transactions.
7. Do you use social data to inform your on-site personalisation? If not, is this something you plan to do in future?
Social data is definitely something that will feed in in the future, but for the moment it feels like people's direct interactions with out site are far more powerful in giving us a direct indication of their intentions.
8. Do you think personalisation is particularly effective in the fashion industry, or should it be used across all industries?
All industries could definitely benefit but fashion feels particularly relevant.
A lot of fashion purchases are driven by the heart as well as the head so being able to gently influence people's attitudes using an understanding of their intentions can make a real difference.
That might not be so powerful if you're selling a commodity product where the decision is going to be made on much simpler criteria such as price and availability.
9. Which team mainly deals with personalisation? Tech, marketing or someone else?
Traditionally, something like personalisation has been a cross-departmental initiative, involving marketing and technology working together really closely.
This has often made it a long and complex process as information and requests have to be fed backwards and forwards and slotted into development queues and the like.
Using the QuBit platform, and I believe this will particularly be the case with their self-service functionality, then it becomes something that can be managed almost entirely from the marketing department.
This means that we can be a lot more nimble and fast moving, which is really important as reacting quickly to changes in the data can make a huge difference.
10. Is it easy to justify spend for personalisation considering that the ROI is normally easier to measure than for other digital channels?
The short answer is yes - if you've got a system in place that gives you the analytics that can easily calculate the real benefits of your campaigns then it's easy to justify the spend.
As I mentioned in the example above, one simple campaign generated a 33% uplift - that's a huge difference and the sort of thing that most site owners would kill for. With that sort of result then it's an easy decision.
No hay comentarios:
Publicar un comentario