Services
Shopify Plus
About
Work
Blog
Contact

Episode 3: Creating a Personalised Experience for Ecommerce Customers w/ Murat from Segmentify

6th December 2018by Ben Froedge
Personalization is the ultimate marketing. Can you give each customer a unique experience based on their preferences and behavior? Why, yes, yes you can.

You'll learn:

  • How personalization can put machine learning tech to work to increase revenues for your store.
  • Why you should be careful in picking which products to recommend to visitors.
  • What is the difference between segmentation and personalization?
  • Why buying patterns change, over time, even for the same individual.
  • What should you change to make for a personalized experience?
  • Why personalized communications matter.
  • Can personalization actually make customers lives better while increasing sales?

Resources

Join the conversation

Keep listening & help others find the show

What else have we been up to?

Make more money with your Shopify store.

Transcript

Miz Trujillo 0:00
Hello and welcome back to another episode of Make Merchants Money, the podcast by Underwaterpistol where we discuss everything Shopify and Ecommerce related. I'm your host Miz Trujillo and on today's episode we will be focusing on "Creating a personalised experience for ecommerce customers". I'm really excited to chat to our guest today, we have Murat Soysal on the show co-founder of Segmentify. Segmentify is a great ecommerce tool which is featured in Forbes "25 Machine Learning Startups To Watch In 2018". Thanks so much for joining me today Murat.

Murat Soysal 0:35
Thank you very much. It is my pleasure to be with you, I'm quite excited for our podcast today.

Miz Trujillo 0:42
We're really happy to have you on the show. So Segmentify helps ecommerce store owners provide a personalised shopping experience for their customers right? Could you tell us a bit more about what that entails and the services you provide?

Murat Soysal 0:54
Exactly, exactly, it's funny to record this podcast with one of the high seasons of ecommerce, we have Black Friday, some have started, others are early adopters, for others it's on the edge. So what we try to do for ecommerce is to understand the visitor and customer behavior for ecommerce companies and let them use this data for increasing their revenue; basically. I mean it's one of the very hyped words, as you mentioned we were ranked by Forbes as one of the 25 companies for this year to be watched for machine learning. Yes, we heavily use machine learning to provide a personalised customer journey for each individual, even if they are nearly missed, even if they are not logged into our website. Understand their needs and also understand the belongings in ecommerce websites, such as the products or the campaigns, ongoing discounts. And make an immediate matching between these two, the visitors; customers, and the products; campaigns, we have. So that is all we want, right? I mean to sell to the customers we need to identify them, what they need and just show them what we have and make this in very real time.

Miz Trujillo 2:17
Amazing. Yeah that's absolutely right, selling to the customer I'm sure is the reason many people will be listening to the show. One thing I wanted to touch on was the importance of product recommendations; which I know is very key to what you provide. Why are these so vital and how should they be used when it comes to upselling, which I'm sure is something our listeners will be interested in?

Murat Soysal 2:38
Yeah, let's start with some data, I mean any ecommerce website might sell a couple of products, maybe hundreds or thousands, in some of our customers cases they have millions of products, but, think about a visitor, how many different products do they get engaged during a session? Is it 10, 20, 25? The average is around there, that is what we see from our analysis, from our 150 customers now. So an ecommerce website lands, he or she gets engaged; what engaged means is they understand the product, the brand, the price, the picture and make a very quick decision to go further on this product or not, so this is what we call engagement. So the average number is around 20 to 25. So think about the case you are selling 1000 products, or couple of thousand or a couple of hundreds, which 20 products should I prompt to this single individual visitor? In the homepage? In the product page? In the category page? This is a decision making process actually. Among a set of products which 20 will we pick? So what we call personalised product recommendations is to pick these 20 products intelligently, by using this machine learning technology which helps us to understand the current visitor. So if he or she is a discount seeker? or a Supergrammer? or is there an upsell opportunity for this customer? On top of it, what is the size, brand or dress sense of this specific customer? That is what we try to identify with product recommendations, but what is actually a little underestimated is what we're actually selling now; I mean based on the season, based on our product catalog, what we are really selling now is also identified. And this information is combined in real time to find out the best matching personalised product selections to each individual and pop it up in various parts of the page, in the widgets, in the pop ups or sometimes in the emails, or sometimes in the push notifications. So any channel you can use with the customer to pop them up the related products, or the related promotions you have, is actually what we understand from the person as product recommendations. So from upsell perspective if you identify there is an upsell opportunity with this current visitors; if it is an existing customer we love it, because we already have lots of data about them, what they checked previously, what was the average order value, even where we send out the purchased products. So we know the location, time, what kind of device they use, so we love the previous historical data yes? But even if we don't have the previous data we can understand their current needs and our current position and make the best match to actually make an upsell. Because, we generally talk about revenue right? But we little bit forget about the profit, but still, even if we call it ecommerce this is still commerce, we do this business to earn our life, so we need profits there. So if you have an opportunity to make an upsell in midst of not only revenue but also in midst off profit, let's use it, why not? So more clearly, if I have two products to recommend to Murat himself, with the same price. but my profit margin is is much higher in product "A"... I should definitely highlight that product "A" to be sold to Murat. Which at the end of the day Murat will be happy as my customer, but as the owner of the business, I will be happy since I will generate more revenue actually.

Miz Trujillo 6:34
Definitely, definitely, the value in that is so huge and you touched on quite a few things that I did want to cover in our conversation today and we'll go into, but something that I think is key to the discussion we're having is the terminology used when talking about some of these things. So maybe if we're looking at segmentation and personalisation, what are the differences between the two?

Murat Soysal 7:02
Perfect question, perfect question for me in two fold actually, the terminology since I used to be an ex academic, terminology is something I think crucial for success. But secondly since I'm the co-founder of Segmentify which uses segmentation for personalisation. So I would love to talk a little bit on top of the differences between segmentation and personalisation. So very basically, segmentation is: Like goes with like. This is generally based on the demographic information, that data we have about the customers. But in modern marketing technologies we use the behavioral data to make the segmentation as well. Very simple definition, when you enter an offline fashion store you will see that all the clothing for men in a side, for women in another side and for the kids is another side, they're not all mixed up. So this is basic segmentation. So if you're looking for something to buy for a men, you go to the left, for women, you go to the right or for the kids, you go direct to the store. So this is actually a very basic definition of segmentation. If I might make it a little bit complicated, what we understand from segmentation is to put the customer behavior data on top of this demographic data. Demographic means, you know, the age, the culture, the location, we can use all this data, but on top of it, let's put a little bit about the behavioral data. You and me might be the same age, in the same gender, from a similar level of family foundation, but still our expectations might be different. So what you could do on top of this demographic segmentation, you could put a little bit of your behavior data, which makes you understand who are the discount seekers, who love discounts? Who are Impulsive Instagram customers? So they arrive just check off products and generally buy always the new arrivals for high prices. The customers who seem to be returning, the customers who are sleepers. So this might be the behavioral segmentation. So what about personalisation on top of it? Yes both men go to the man stand in the offline store; but, I like black, but you like blue. I'm a size of, let's say, 10 in my shoes but you're a 9. I don't like to spend too much money on shirts so that's why I generally look for the medium price level shirts, but you really love to spend a lot on your shirt selection. So, now this is personalisation. On top of what we have from the demographic, we as specific visitors have specific choices. So when I call personalisation, it's not just saying Dear Murat, so it's not just changing the name field in an email or in the website when you call the customer, but where you can personalise your webpage, you should actually. That is what I understand from personalisation. So these two terminologies actually do not race each other, they actually complement each other. You can use segmentation before doing personalisation or they can both work individually. But just to finalize this description from my perspective, personalisation is actually taking actions on top of; or along with, segmentation.

Miz Trujillo 10:49
Perfect, because the segmentation side of things basically can help a store owner better understand how a group of like-minded consumers act is that correct yes?

Murat Soysal 11:00
Exactly.

Miz Trujillo 11:01
Once they have that information is that where targeting would come into play? How how should this data to be incorporated into marketing campaigns?

Murat Soysal 11:08
Perfect, perfect, this is actually good for me, this is how we designed Segmentify to work. Everything in Segmentify; let me tell you a little bit from my perspective on top of it. It actually starts with the segmentation and on top we conduct the personalisation. So let's say in today's customer centric marketing world we all talk about the customer, customer, customer, behavior analysis, customer journeys, customer experience, but little bit we identified that we underestimate our products that we are selling, or the campaigns we have provided,or the promotion we have provided. So segmentation and personalisation should be working together by starting doing a basic level of segmentation, this will help you first segment your customers or visitors; so use any data you can collect from them, and also segment your products. So how do they really work together? With segmentation you will identify that the current visitor "X" is an upsell opportunity, is an upsell opportunity in my website. So you can define this either based on the historic data, or definitely whenever they check a couple of products in your website you can identify their is an upsell opportunity on top of this, but with the personalisation, what you should do actually, you should highlight the related products or campaigns that matches this current visitor. So in this specific case, if visitor "A" let's say is not an upsell opportunity but is a discount seeker and lands on product "one" in my webpage, so this product "one" from the segmentation has multiple alternative products, let's say five, so you can highlight these five products to this customer "a", but since you identify him or her as a discount seeker from your segmentation let's now use the personalisation on top. Among five alternatives of product "one", let's pick the ones that best matches the size selections of this customer if you are a size enabled ecommerce, the color selections of this visitor, the brand selections of this visitor if you are a multi brand. But especially since that visitor is a discount seeker, use his or her price preferences, so among those five alternatives pick up the best matches on top of these preferences. So there is a ranking that should be going on among your segmentation to make the personalisation. At the end of the day you should be recommending products that are really picked personalised to this individual visitor.

Miz Trujillo 14:09
Yes, that's what's going to have the most value and make the most impact. So from our listeners perspective, I think it's clear the value that this can add, implementing segmentation and personalisation. Ideally, they want to begin tailoring the visitors experience as quickly as possible, but in order to do this they need data from from the visitor. So what's the best way of collecting this data and making this possible?

Murat Soysal 14:30
So if you asked me for using personalisation and segmentation, yes, the availability of data is crucial, the more data you have, you will make more precise decisions, more personalised solutions to your customers and visitors. It should all start with you start with your website and the initial first visit of a visitor. So let's say I'm visiting the Underwaterpistol website today, this is my first visit or you and I just clicked a Facebook advertisement that says that there's new arrivals available in the store now. So I am coming from London using my Mac computers and just clicked the Facebook add. Even this one single click, I tell a lot actually about myself, not to be personalised, but still to be segmented. First of all, I am using a Mac computer says that; okay guys, for some articles of my life, I will be happy to pay the best available price in the market. So I will pay if I like the article. Second one, combined with the Facebook advertisement example, that new arrivals are most probably in the top price. So I still says that; hey guys, I am a follower, generally I want to be a trendsetter, so please keep this in mind when you get engaged with me in your website. And then starting from this point, my location, my browser type and the more I spend time on the webpage, the more I check products, then I will say about my selections of brand and price and category in your website, as I said if it has available sizes. So the more I add products to the baskets, now I'm little bit moving the funnel, so I give more strong signals to you. So this is the starting point of actually collecting the data. So if you are able to collect this data in real time, and accordingly, so if I am a recurring visitor of you, I am an existing customer and if I used to buy large shirts from your website, and I land now and check three different extra large products, which means unfortunately I put on some weight. So please use this brand new data in real time, and change your ongoing marketing activities for me, So help me actually find what I need much more easier. Well, as a customer, I will pay money to you, I will buy products from your site. So that should be the part, but still, we have some customers who has very strong offline presence as well. So again, little bit of data customers tend to buy single products on the web, but they tend to buy more products at a time on offline stores, what we do we use that offline store data as well to build up our machine learning activities to fill these algorithms and learn more about the cross sell opportunities, because as the owner of the ecommerce website, treating generally what might be the best matching bundles or cross sell opportunities for each individual product, but users know better than us which might be a complimentary of the other product. So if you could aggregate any offline data, it is also welcome, to speed up any machine learning algorithm. So whatever data you can collect is actually precious, but you should start from collecting your own data from the online store in real time. This is critical.

Miz Trujillo 18:30
So for people jumping into this; maybe at an early stage, when this data starts coming in it might seem a bit overwhelming, what would you recommend in terms of analytics that they should be focusing on?

Murat Soysal 18:42
Ah, this is a basic funnel for me. So from the analytics perspective, I will start always like this, little bit joining the customer journey. So I always try for my customers, especially for the ones in Shopify Plus, that they should be engaging with their customers just for branding first. So yes, as I said, this is commerce, we need to generate revenue, but we need loyal customers as well. So we need to first increase the customer loyalty, discuss with the engagement, even before the loyalty, which I mean customers or visitors should understand that they are not in ANY website that is selling a specific product which he just googled. This is very harmful for all of us, because visitors tend to be a little bit price sensitive, search a product, just click a couple of links from Google, they will not understand in which webpage they are. This is real awful for all our efforts to build up this company, the ecommerce website, run it. You want them to understand, no, you are not ANYWHERE, you are in MY store, in brand "A" store. So to build this up, we need the customers or the visitors to visit three product pages in a session. So let's focus on how we could do it. So from an analytics perspective, if we cannot achieve this, we need to identify why. Where in the funnel they're actually leaving the website. So if they lend to a single product and if there's always loss from that, so it is saying that we are not able to grab the attention of the customer, so let's try to avoid this quick leaving from our website as that loss is a problem. So if you asked me from the analytics perspective, funnel by funnel, starting from the upper side of the funnel, we need to improve one by one.

Miz Trujillo 20:42
Excellent, thank you so much for that Murat, I think that's something that is very key to the reason we're having this conversation is conversion optimization. That's what people are really interested in and it's also a significant challenge for ecommerce websites. How can personalisation help? In terms of what kind of impact should listeners expect to see in conversion rates if they apply this properly?

Murat Soysal 21:06
This is important, what you should expect from your personalisation efforts in terms of increasing your revenue, average order values, decreasing the bounce rates, that is all fancy words or sentences we all follow that we want to achieve. Again, I will go back to data, what we see with our current customer base. So we have, as I said, 150 customers from 10 different countries, these are scaling from very small vertical shops going to marketplace giants. So if you employ machine learning to build that personalisation, more specifically the personalised product recommendations, we see an average of 10% revenue should be coming from these personalised product recommendation widgets just on the side. So what I mean, you landed to my website, I recommend four products to you, you click product "A" from the widget I recommend, you add product "A" and if you buy that product "A" in the same session; which means in 30 minutes after you clicking it, the overall revenue coming from those recommended products should be around 10%. So we are happy to be that clear because it is really easy to track this conversion. You collect product "A" from a widget, add it to basket and make a purchase. So then I talk to customers, some customer candidates or in some meetups and people unfortunately little bit underestimate the performance of these widgets. But on the other hand just think about your own product page, you sacrifice around 30% of your page to these widgets and sometimes they even do not count or track how much revenue these widgets are generating. It's actually a really big missed opportunity because that 30% of the page is the crucial thing that help us to engage with the customer, keep them on the site and increase the revenue. So once you start tracking the revenue coming from these widgets it should sum up with an average value of 10%. We see some customers they generate 30% of their revenue from these widgets, which I am not happy actually with this one, because that is most probably that the website is designed in a wrong way in means of the user experience, so customers can not scroll or browse the website easily from the categories so that is why they usually tend to use the product recommendations to browse the webpage, which I mean from my perspective it is 30%, I should be happy, but I generally I'm not happy with this one. So around 10 to 15% is ok for the onsite widgets. But personalisation is not being held on the website only, as I said you have the emails and push notifications to be sent to customers in different steps of the funnel, so this should sum up to 5% addition from the email marketing just with the personalised emails I'm talking about not mass emails and 2.5% which the web Push Notifications should be expected actually. But as I said these are very average general numbers, it differs from vertical to vertical but it is always good to have an indicator that you should compare right? If I'm doing right or bad.

Miz Trujillo 24:44
Yeah definitely I mean those stats really highlight the importance of implementing this into a strategy and we talked about Push Notifications on the show recently, but something else that you mentioned was the email campaigns so I wanted to touch on the benefits of personalizing email marketing campaigns. Why is this so important and how should listeners be going about this?
Murat Soysal 25:07
Oh yeah, I just listened the awesome podcast for the Push Notifications with Push Owl so they do an excellent job as well on top of this. And from the email perspective let's start with the usual suspects right, abandoned carts, we are all doing this; almost, and even the customers are also now aware of this so they want to check abandoning their cart if they really want to buy and try to understand if you are giving discounts or not. So it is very well known now. How I start this conversation with all my customers, okay we do abandon carts, we all do, I always ask them do you know which products you have are top abandoned? This is a very good question. We need to little bit chase activities here. Doing automation is good, sending automated abandoned cart emails is ok, but on top of it we need to dig, why the top abandoned product is actually abandoned? Because we spend too much money getting these visitors to our website. They'd like product "A". They even add it to the basket and they leave it. Why? There should be a reason, either our competitors are selling this more cheaper than us so they have a huge discount on top of it, do we want to compete with them with the pricing or not? Still fine, but we need to know. Or maybe that specific product page is missing some information, maybe the details of the product is not able to convince the customers to make the last step in the funnel and purchase it, or maybe the images are not that good. So you need to identify this to take action, so that is why I wanted to highlight the emails first, but on top what we should do using personalisation with the abandoned carts for example, you need to decide when to send for each individual, what data to send other than the products abandoned in the cart and to whom we should send. I mean should we send abandoned cart emails to any single customer that leaves the product in the cart or not? So if a customer really abuses this, so leaves the cart abandoned then comes immediately and if they repeat this multiple times should we still continue to send abandoned cart emails to them or not? This is actually what I understand from the personalised abandoned carts or personalised emails. But coming from the abandoned cart to the on top, if you asked me what should be in the personalised emails, my favourites are; Price Drops, so if I check a product and if my ecommerce company sends me any emails if there is really a price drop in this product personally I am quite happy and also we see that the positive responses coming from these campaigns are quite high from the visitors. And also; In Stock Again, so if you are working with tight stock numbers customers tend to land into many out of stock products, or if you have sizes in your products some sizes of your products might be out of stock but if you are able to get them in stock again and inform your customers immediately who checked that size in and out of stock stage, we can email, 'Hi Murat, come back, the shirt you just checked is now in stock we even saved one for you for the remaining 1 hour, 6 hours', any way you can set, that will be the real engage personalised emails right?

Miz Trujillo 26:29
Absolutely yeah that's that's amazing, not only on the email front but I was also going to mention abandoned carts and you covered that perfectly, it's such a huge worry for people. so seeing how personalised emails can be used to reduce this is absolutely great. Whilst on the subject of sending out emails I wonder if we can touch on email preference centers? They allow people to change their own subscription details, opt in and out of newsletters and even if set up correctly enable people to edit any custom field information that sellers have stored for them. And I often see them underused really, or undervalued. How would you recommend sellers make the most of these? What options should they ideally be including for customers?

Murat Soysal 26:29
Again good question, before the options even, when they are selecting the email service provider, they should really take this into account how much your email service provider lets you manage these preferences, how much customization you will be able to do. Because as I said previously, if you're able to visit all these steps and send different emails to different persons or individuals at different steps of the funnel now you will need email preferences. So in general, basic ones, first you need to understand your customer behavior. Do they really not want to get an email after 10 or 9 at night? I'm not sure about this for example, because since we have 10 different countries expanded with Segmentify, different cultures actually respond differently to this and by changing mobile activity in our web pages I'm not really sure if you should not send emails to your customers after 10. I am a dad, my kid goes to bed at around half 8 or 9, I can only find time to check my personal email other than my business email for engaging with my ecommerce websites after 10 actually, at night. So during that dual screen time, when I'm watching something from Netflix, I still check my emails if there is really an ongoing campaign that I can follow. So this should be actually set pragmatically, maybe regionally if you're sending products to multi regions, or maybe individually so customers, all customers should be able to define their preferences as well. That is what we are now facing with integrity to Email Service Providers. As an individual visitor I want to set my preferences as well. Because if you're talking now in personalisation words, if we have a chance to make 7 billion personalised campaigns which covers all the world actually, that will be awesome. So from my perspective, when to get emails, which emails to get are quite crucial. So really, should my customers be able to set if they want to get Price Drop or In Stock emails or not? Or abandoned cart email or not? So even they should be able to set this. So I would love to work with those Email Service Providers who let us personalise any parameter that is in the email itself.

Miz Trujillo 32:27
Amazing. Thank you Murat. It seems ridiculous really for people to be ignoring personalisation, especially after the things we've been discussing on the show today. Before we start maybe wrapping things up, are there any other key components to optimizing a customer's journey you would highlight for our listeners?

Murat Soysal 32:45
Yeah, I mean, I think we should set division also in the discussion. So I try to highlight some data to make it more emphasized in midst of the cruciality of personalisation, but where it should go up to? So let's talk a little bit, very small about division. So if you really want to understand where personalisation can go, I love an example. So when you have your first baby, you buy a baby carrier for your new born. So you buy it from one of your landmarks. So your landmark should send you an email in six months of your kid and highlighting that, hey, you should buy your car seat now because your son or your daughter is going nine months in three months. so let's start checking these car seats. So that should be actually in the customer journey, that personalisation should help me. That should be division we should implement because, you know, it is always three times more expensive to acquire new customers. So from your existing customers, you should be their landmark. That is what I want to be building to all my customers or my company segment for as well to achieve this. If you asked me, could you do this perfectly? It's never perfect. It never happens, perfect, but this should be division we should be engaging with all of our customers.

Miz Trujillo 34:20
That's such a great example Murat. Thank you. Maybe if you had to leave our listeners with one thing from what we discussed today, what would it be?

Murat Soysal 34:31
The most important thing I think is the return on investment in personalisation. Because I throw some numbers there, 10% of any contribution from on site, 5% from email, 2.5% from web Push Notifications. These are all very easy to track. So return on investment in personalisation is actually very easy. So you should avoid all the noise for personalisation. And we all need to, I mean both the customers and the providers should focus on the return on investment we provide to ecommerce companies from personalisation, that I would always love to highlight actually.

Miz Trujillo 35:14
Absolutely the return on investment is the reasons we have these discussions at the end of the day isn't it?

Murat Soysal 35:19
Exactly, exactly.

Miz Trujillo 35:21
Well thank you so much for being on the show Murat, it's been absolutely great talking to you.

Murat Soysal 35:25
Thank you very much for inviting me and I really enjoyed the discussion. Thank you very much for the chance.

Miz Trujillo 35:32
Excellent, I'd like to thank our listeners too. You can find links to both Segmentify and Underwaterpistol in the description below. And if you need any help with Shopify and ecommerce, you can book a free consultation with Underwaterpistol via the link below too. Please subscribe to the podcast to keep up to date with all the upcoming shows and you can get involved in the discussion and receive the latest news by joining our Make Merchants Money Facebook group. Thanks for listening and thanks again Murat.

Murat Soysal 35:58
Thank you very much. Bye bye.