-Last of the day! Time for drinks! -Data analyst -3 years working there -Working on personalisation -Today talk on how we use data to be relevant in online marketing -How focussing on the customer helps instead of focusing on the channels
-I assume you know -A website on .nl and .be -We had a good year and a good quarter
-We have a mission -We want to connect with the customer -Made our own translation / goal
-Personalized means customer focused -Personalisation is not about matrix factorization etc. its about helping the customer and thus creating business value
-Omnichannel is not only combining store and online -All touchpoints -Also look at your online marketing channels -We all know that often more than one channel is used
-Most of you know this report -It is from analytics -It shows which channels customers use after each other before converting -Shows interactions between channels -So channels help each other
-Use the information to have a consistent dialog on all channels
-But somehow when we use online marketing tools these still work in silos with their own data
-Every channel uses its own subset of the data, own model, own owner -It applies its own logic and selection -Focus is on channels! Where is the customer? -Causes problems for the customer
-Get an email for Michael Kors bags -I get a display ad for dresses (Ralph) -On the site I get living items -Just because every focus is on the channel -Insane for the customers -So focus on the customer
-Focus on the guests (single customer view) -We know a lot of the customer -Use the data on all channels -Respect privacy -Try to understand the customer journey
-Customers go through different phases before buying -Different channels fit better with certain phases -Channels should try to tell a consistent story across the journey -There is a system involved, which we build ourselves -But you need a system to build this.
-To be able to target everyone consistently: -have one source of truth (data) -have one place to build models from -Snowplow is a open-source analytics platform that you can setup yourself on a cloudprovider of your choice
-Might look daunting, but stepwise approach -Determine your data requirements only when you have a proper business case for a big data project
-Lets start with the early part oif the journey -The customer is still looking -So innspiration is important, but inspiration that fits with the customer -Lets start with a simple example: the newsletter -We need old data
How do we find your interest Lets start simple with a newsletter
-We start with a data problem -No Filippa K jeans -New brands get out of stock so collaborative filtering is hard -So we need to look at other ways to understand the needs of the customer -Look at attributes of the products
-Cluster analysis based on passed bahavior -People get assigned to a cluster based on their browsing and buying behavior -We can do add other segments to this, such as the discount sensibility, category preferences, etc.
-Walk through it -Sent people the mail they might want -Another advantage is we can easily create the content, its manageble -These simple attributes helped a lot
-So to build this, what’s the plan?
-Easy! -Walkthrough -Next step is ofcourse to automate this further
Use the data in a machine learning model Might say, sent out no mail. Respect that! Focus on customer not the channel But email is just one channel and moves people to the site Be consistent!
-We send an email to get people to the site! -There is no reply to buy functionality we have available -Email cannot function on it own! -There can be time between the two -I can move from email to site, wait and return tomorrow or tonight!
-Sort products based on your profile attributes -Technically not difficult (might depend on your site)
- Add the site, maybe using an API endpoint
-To extend and respons faster on the site, realtime might be needed
-We can inspire the customer with newsletter and on the site -Display
-So lets move to the other phases -These are more triggerbased -What have they done
-Wait -Most familiar of all the triggers -Customer has shown interest in a product -Abandon basket mails are really good functioning
-After people have actively searched -Use their profile data to extend the content -What wedo here we can also apply to retargeting on other sites
-Most important question is: what should you show the customer if they browsed alot
-But don’t annoy people. Be relevant!
-Customer journey can extend to the stores as well! -Important to note -Omnichannel focus -Need to tie together
1 miljoen privilege members So we know what they buy ion store Easy to integrate
-Combine email and retargeting for a consistent story
-But is it needed? -Does this customer need it? -We can decide based on customer variables
-This strengthens the interaction between channels. -Lowers our costs, increases satisfaction, and is consistent for the customer -But we can also use the profile data of the customer
-If this behavior is shown, most retargeting will focus on the ring -But there are more patterns in here -Match with the customer! -Technically you just need something to send out the ads with, like a DMP)
-These were just some examples -Can be extended to more channels
-These were just some examples
-Which channel when and for who with what message? -Combine more -Attribution can be taken into account
-Please stop doing this! -Ignore channels -Just a way to get in touch with your customer
And please put technolgy even later!
De Bijenkorf Niels Reijmer
The Shared Value Of Data:
• 7 flagship stores, 100.000+ M2 shopping space, 25 mln visitors
• Online platform in NL & BE, 55 mln visitors
It’s our ambition to be the most
inspiring, surprising & creative
To create the ultimate
shopping experience for our
11Get into the journey
Use data from the previous journey
Abri / tv
App push Trigger ad
Active looking Deciding Consuming