Latest / Inside Commerce: Ecommerce Strategy, CX and Technology Podcast / EP283: Digital Revolution - How VIU Eyewear Rebuilt Its Ecommerce Stack To Improve CX Through Smarter Automation & Connectivity, With CTO Elian Kool
Transcript
- 0:00We've built an MVP because we had a good idea how it should
- 0:03look like at the end. So we've, we've had like 3 days
- 0:05to have actually the support from the commerce layer people
- 0:08to remote this or to ask answer questions.
- 0:10But we've been able in those three days actually to build up
- 0:12like a first store where we could sell a few things that we
- 0:15could return something where we where we where we've been able
- 0:18to have to use the order management in the way we
- 0:20intended it. You listen to Inside Commerce, a
- 0:25weekly ecom strategy, technology and CX focus podcast hosted by
- 0:28experienced digital consultants and re platform specialist James
- 0:31Gedacle Rogers. Get insights into ecom strategy,
- 0:33leading brands, hear from industry thought leaders and
- 0:35leading tech vendors with practical advice to help you
- 0:38improve your online CX. This episode is sponsored by
- 0:43Shopify. Another week, another fantastic
- 0:46guest and interesting discussion.
- 0:48We're talking about digital revolution.
- 0:50Have you rebuilt it's ecom stack to improve customer experience
- 0:53through smarter automation and connectivity?
- 0:55Sounds like a mouthful, but it's actually really important
- 0:57because we know lots of businesses that go through
- 0:59digital transformation to varying degrees onto different
- 1:02tech stacks and that brings different challenges.
- 1:04And hearing how people have approached it, what they've
- 1:06learnt and giving you practical advice is really useful.
- 1:09So I'm really delighted to welcome Eliane Cole, who's the
- 1:11CTO at Via Eyewear. Hey Elena, lovely to have you on
- 1:15the podcast. Great to be here.
- 1:17Thanks a lot for the limitation, James.
- 1:19Yeah, I've really enjoyed chatting to you and our little
- 1:22preamble to get set up on this and hearing a bit about the
- 1:24brand story. So before we get into the
- 1:26talking about that digital revolution piece and the tech
- 1:28piece, let's get an introduction to you.
- 1:30Have you? Let's go about, you know, the
- 1:32brand story and what brought you to lead the team.
- 1:36Let's get started with views of views started actually back in
- 1:382013. So it's already quite a while
- 1:41back, but it was a very simple idea back then to change the
- 1:44eyewear industry by cutting out sort of all the unnecessary
- 1:47intermediaries and like we offering high quality glasses
- 1:50directly to our customers. So we will work with boutique
- 1:54frame manufacturers both in in South Tyrol in Italy and in
- 1:58Japan, as well as with some of the top tier lens manufacturers
- 2:02in Switzerland and Germany to really create outstanding
- 2:05quality at very fair prices. So initially it really started
- 2:09as an online business. It was the, the initial, very
- 2:12initial idea on that. But it has grown into a true
- 2:14Omni channel retail company where today we have around 50
- 2:17stores and across the dock region or probably to say it in
- 2:21short, I wear it's very often overpriced because there's a
- 2:24very expensive distribution and and we're here to change that
- 2:28and make premium glasses more accessible.
- 2:31Yeah, I love that because this, this is the thing I sometimes
- 2:34have struggled with as a consumer is I'll find a brand I
- 2:37like and maybe I can buy into its values.
- 2:39It's got some like ethical element to it.
- 2:41It's got some, you know, it's using really good materials, but
- 2:43then the price point is, is crazy high and being able to
- 2:47deliver the quality at the price point is quite, it's quite a
- 2:50compelling proposition. And and that really also
- 2:54resembles with our customers. And that's what you see now like
- 2:56from the past 10 years, we've always did the designs
- 2:59ourselves. We usually the high quality
- 3:02frame manufacturers in those places where they, where they
- 3:05are reaching from them making sure that we get our cool
- 3:09designs in a high quality combined with the industry
- 3:13standard high quality lenses. That's really what our customers
- 3:16are are asking for and why they they keep coming back and back
- 3:19to us. Yeah, it was dangerous for me
- 3:22talking to you because as soon as I did, I went on to the site
- 3:24and I looked at the sunglasses because now we're starting to
- 3:26get some sunshine finally in the UK.
- 3:28It's like, oh, looking at the gambit and looking at these very
- 3:31I was like, Oh my word, this is this is going to be a fun
- 3:33challenge for me pre summer to pick out the the best one.
- 3:37Yeah, no, indeed, sunny season is always a strong season for
- 3:41us. Yeah, well, people get happier
- 3:43and they feel better about spending money.
- 3:44So let's get into the tech story because I, I'm fascinated by
- 3:49this. I love hearing how people go on
- 3:50change and when they have to make radical changes.
- 3:53And you looked when you came in and you looked at consolidation
- 3:57and simplification and you said that you focused on the
- 3:59principle that nobody does anything twice.
- 4:02And I think everyone listening can just resonates where you
- 4:05might do one one thing in one system like somebody does some
- 4:08product work in an ERP, then they redo the product work in
- 4:10ecom or somewhere. And they might have to set up a
- 4:13discount once in a system and then they have to then reset it
- 4:16up in a different one in order to get it launched.
- 4:19So with various principles, what were the key issues that you saw
- 4:22needed to be fixed? Oh, as I joined then, I mean
- 4:25Dick, I would say views suffered many of those issues that quite
- 4:28a few first start-ups in the age of like 8 to 10 years, they,
- 4:32they typically get into that because some processes were just
- 4:35set up, I would say in a very start up way.
- 4:37So if you needed something, you build it in a way that somehow
- 4:40works, but you actually never adopt that to a real grown up
- 4:44stage where you actually already are for quite a while.
- 4:47So there was, I would say there was just no clear master data
- 4:51management like and I would say the very basic 1 is just product
- 4:54data. But also just as one example, if
- 4:57you offer an eye test in the store, it typically sets the
- 5:01availabilities for that. When are people able to to do an
- 5:04eye test in a specific tool? But actually in reality you
- 5:07already know that because you have in retail, you have a shift
- 5:10planning tool. Somebody actually know exactly
- 5:12who will be working when we have the HR tool where we know
- 5:15exactly who is able actually optician who is able to do eye
- 5:19tests. So you have all that data.
- 5:21But nevertheless, because it's easier you just create up an
- 5:25additional system and ask them to to to enter the same data
- 5:29twice, which obviously others quite error prone.
- 5:32And that is really was one of the main principles.
- 5:34We want to make sure that everything is is only on one
- 5:37place, whether it's store data or whether it's opening times,
- 5:40whether it's product data, whether it's availabilities of
- 5:43employees, etcetera, etcetera, that there is nothing is
- 5:46emptered twice in the system. Yeah, I, I really like that
- 5:50because a lot of the time when people plan change, it's planned
- 5:54on one part of the tech stack. And therefore it's not thinking
- 5:57about the data needs across the business.
- 6:00And therefore how you can consolidate centrally and then
- 6:03use modern tech, whether that's APIs or file transfer, whatever
- 6:06the mechanisms need to be to ensure that you have an owner of
- 6:10each bit of data and that owner is going to control via a
- 6:12centralised system. So I like that vision.
- 6:16It was, it was definitely quite a hard change because like in
- 6:19the past, just to illustrate that I'm not sure whether it
- 6:22makes sense to celebrate too much, but like just to create a
- 6:25product actually was quite hard because some the, the major
- 6:29product data was owned by the product team, whereas marketing
- 6:32was setting up the the product in the content management system
- 6:36more than in the shop system. They were adding the images in a
- 6:40different FTP server, just naming them in the same way how
- 6:43the product number was was looking like.
- 6:45And whether that actually came together or not was to some
- 6:49extent a bit like by chance. And really centralizing that has
- 6:52on one side a super big advantage because you really
- 6:56have to validation and everything once.
- 6:57But on the other side, it's also people have to get used to the
- 7:00fact that once they're changing something, it actually has an
- 7:03impact everywhere. So for example, they're
- 7:05launching a new product and they forget to set to the proper
- 7:09start date or the start date is moved to the back.
- 7:11Well, that has just the effect directly.
- 7:13The product will be visible on the website, but also available
- 7:16in the stores to sell. So with with all the good things
- 7:19attached to it, that you also have to realise that what you're
- 7:22doing actually has a real immediate impact.
- 7:26Yeah. Do you know another example I've
- 7:28I've come across very well. I've known about it for a long
- 7:30time as an example of this but I've seen again in a new project
- 7:33and sometimes you have to accept you can't do it all at once but
- 7:37you don't lose track of the vision and the requirements and
- 7:39you have them on post launch. Things is stockist data for a
- 7:42retailer that sells via other retailers and they want to show
- 7:46where to buy on the line. You often find that a sales team
- 7:50have the data somewhere and even with the team, then take the
- 7:52data and re key it. But there's no link.
- 7:56But you have modern tech to if you said that right, you whether
- 7:59it's webhooks or whatever it is or triggered events that go
- 8:02we've updated here, push that now into the database over here
- 8:07so that you don't have to rely on that manual process of, you
- 8:10know, Hey, hey, Joe in sales, has anything changed?
- 8:12Oh, great, I'll update it in the back end.
- 8:15Admin of the econ as well. And it all stock is data.
- 8:18Always a mess. And, and I mean, that's like
- 8:20actually an excellent example. I mean, like we had so many
- 8:23manual processes which even couldn't be done by the store
- 8:26teams themselves. Essentially, they wrote a ticket
- 8:28and the operations team did something and stock was a
- 8:30constant issue. And that so it was actually like
- 8:33we constantly have to do inventories and to make sure
- 8:35what, what do we actually have? At the same time, it was
- 8:38actually, it was never really sure.
- 8:39So what they've always done, especially if something was on
- 8:42low stock, then they just called the operations hub, which also
- 8:45was the warehouse back then and asked Katie, you still have from
- 8:48that. So there was they, they, I
- 8:49essentially had a constant phone line from the stores to the
- 8:51operation centre to figure out what is actually the reality.
- 8:55And that is really what we've what we've changed entirely with
- 8:57the approach. Yeah, interesting.
- 8:59So let's talk about the, that consolidation of process and
- 9:02tech, because having seen this in various sized businesses, it
- 9:06can be really confusing sometimes to know where you
- 9:08start in what order. Because like I remember you
- 9:11saying to me, like lots of stuff have been patched together.
- 9:14You get all these dependencies where you, it's like, I call it
- 9:16whack a mole. You, you change something over
- 9:18here and then something goes wrong over there.
- 9:20Like, well, we didn't know that was going to.
- 9:22So how did how did you approach it to identify what to change
- 9:25and what to prioritize so you didn't create massive
- 9:27disruption? So I mean, as I looked at it and
- 9:31as I then sort of presented my findings afterwards also to the
- 9:34board, I used the pictures of a Chinese survey, like, you know,
- 9:36where they built the house and then they attach a balcony and
- 9:39they add an additional story and as something in addition.
- 9:43So that the whole thing was indeed like overdose 10 years
- 9:46being patched together. So and then again, like from
- 9:50from a tech perspective, what we learn and in school, university,
- 9:53it's you actually, you do all those things step by step that
- 9:56have a Big Bang approaches. That's way too, way too
- 9:59dangerous. If you if you do many things at
- 10:01the same time, there's way too many things that will break.
- 10:04There will be interference in the in the business.
- 10:06So if actually took exactly that step because going down the
- 10:11route and saying, hey, we're, we're taking now first the
- 10:13product data management and then this part and then this part,
- 10:16that would have been a journey of seven years, which the
- 10:18company has just wouldn't have been able to afford almost sort
- 10:22of on the good side, because we knew we, we want to change so
- 10:25many things. We've actually took the approach
- 10:27to say we're building a new company in the company almost.
- 10:31So we've, it wasn't the only tech project, but the authors
- 10:35have redone all the processes. So at the same time we've
- 10:38outsourced the warehouse, we changed the entire way how we do
- 10:43fulfilment processes. We changed the entire definition
- 10:46of how an SKU looks like. So before we just had article
- 10:50numbers, which also were the same between Sony's and frames.
- 10:53So it was actually quite hard to know what you actually have.
- 10:55But the we've, we've did all that change at the same time,
- 11:00which then essentially LED sort of to the way we're, we're not
- 11:04changing sort of system by system, but we're changing store
- 11:07by store. That means that if we for now
- 11:11April last year, we started with the first pilot store and the
- 11:14first pilot store then did sort of was spooked out of the old
- 11:17inventory within inventory new system.
- 11:19And from that moment on that store was fully being served by
- 11:24the new system. So that means both like on the
- 11:26sales side, but also back to the apparent to the ERP system where
- 11:29all the finances we managed. So it was indeed, it was like 2
- 11:33entire systems that were set up. And the the migration path was
- 11:37essentially store by store that we moved over.
- 11:40And either a store was still in the old world, which doesn't
- 11:43meant that they had stock of the old world, or it was entirely in
- 11:47the new world, which then also was served by the stock and all
- 11:50the processes of the new world. What do Mattel, Victoria,
- 11:55Beckham, Sennheiser and Glossier all have in common?
- 12:00They power their businesses with Shopify.
- 12:04Shopify is the most innovative and scaled commerce platform on
- 12:07the planet. It also happens to have the best
- 12:11converting check out on the planet Shopify Powering the
- 12:16future of enterprise commerce. Find out more at
- 12:19shopify.com/enterprise. Yeah, I'd, I really, I'm a
- 12:24massive fan of the don't try Big Bang it.
- 12:27If there is a fundamental reason why you have to, then you have
- 12:30to plan around it and accept there's going to be risk.
- 12:32But the I, I'm working with a pet gamer at the moment.
- 12:36I really like their approach as well, which is their number one
- 12:38priority is to, is to migrate from the old econ platform to
- 12:41the new one and not get it by adding loads of other stuff and
- 12:44get that right. But as part of that fix known
- 12:48issues. And the product data model was
- 12:49one of the key ones of let's get the product data model set up
- 12:52correctly around, you know, SKUs, bundles, product tags,
- 12:56hierarchies as part of that process.
- 12:59So we're in a cleaner state, but don't try and complicate it by
- 13:02adding all these other like integrations and things that
- 13:04aren't critical to trade. Yeah.
- 13:07I mean, and there's some things that we thought that then only
- 13:09done a bit later on like you know, analytic side etcetera
- 13:11came in and only again on the end of the year.
- 13:14But sort of the if we migrated in a store and onto the new
- 13:18world, it really contained them that the entire new process,
- 13:23everything came from a new warehouse, but it was very much
- 13:26risk reduced because it was only affecting one single store.
- 13:29So also the load that was happening from that single store
- 13:32was was quite overseeable. So we could really use that
- 13:35source then to fine tune everything, make sure that all
- 13:37the processes work fine and all the shipments work fine.
- 13:39The software itself is working. Then we've essentially went on
- 13:44to to the next countries or making sure that from each
- 13:47country we have to have one store up and running.
- 13:49And then we've been sort of done managing to roll out relatively
- 13:53quickly at the end. I love that.
- 13:55It's really good. I think this is really
- 13:56interesting for people listening to hear how people approach
- 13:59phase in and there's difference. But I love that store by store
- 14:02process because you you, you get your process right and you learn
- 14:06from it. You talk any issues before you
- 14:09start rolling everything else into it.
- 14:11And that's actually has been super, super important because
- 14:14in reality, the the orders, obviously on day one, we had the
- 14:18first exception from something that actually should never
- 14:20happen, but it just happens. And we have so complex orders as
- 14:25with all the changes that happened on the orders, et
- 14:27cetera. So there's almost only
- 14:29exception. Exception is the default we've
- 14:30always been saying so so so we really have to work with that
- 14:35and and you have to get real data of it and you have to
- 14:38figure out how are people actually using it for example,
- 14:40the whole goods entry if built in a Bay.
- 14:42We haven't been really happy and we said, Hey, we actually don't
- 14:45know right now what's the best way.
- 14:47So let's ship it to the first store and then just let's just
- 14:51watch them how they're doing it. And in the moment we we've seen
- 14:54what they've actually been doing with the new tool.
- 14:56It was immediately crystal clear.
- 14:58What do we need to do? OK, then let's change that now
- 15:00quickly. And as only one store and we've,
- 15:02we've been working quite closely with them.
- 15:04We've had to been able to make changes from today to tomorrow
- 15:07and really making sure that we get it, that we get it right for
- 15:11for all the rest of the stores. Yeah, it's nice.
- 15:14The, I think the E Comm industry is, has been guilty of talking
- 15:17about agility as a generic term and it doesn't always mean
- 15:20anything. But that to me is true agility.
- 15:22It's it's not just about sprints and development release.
- 15:26Agility is around process, it's around ways of working.
- 15:29It's about learning quickly and adapting quickly before you've
- 15:33built something massive and gone oh shit we've got to unpick this
- 15:36now. I think that really has been one
- 15:38of the of the main learning. So we you, you can't especially
- 15:41like where, where orders are that complex.
- 15:44Obviously if you if you just have something you sell over the
- 15:46shelf and you, you probably can return it and that's it.
- 15:49That's probably relatively straightforward.
- 15:50But all the other complex processes around processes
- 15:53around that is really makes more sense to to come up with a
- 15:57minimum viable product, figure out how to actually use it, and
- 16:00then extend and in those areas where it makes sense.
- 16:03Yeah, excellent. And let's get into that, that
- 16:05tech journey in where you end up in the core econ platform
- 16:07because as you'd said, it was quite convoluted tech journey of
- 16:10like lots of different things being bolted on and various tech
- 16:13stacks being considered. And you ended up with Commerce
- 16:16Slayer. And a big shout out to Seth
- 16:18Bendenagle, who's a long term friend of the podcast who
- 16:20introduced us and said this would, you know, it'd be really
- 16:23interesting brand story. But yeah, how did you get to
- 16:26Commerce Slayer? Because there's so many
- 16:27platforms and options in the market.
- 16:30So one decision that we've took from the very beginning on is
- 16:34that we really want and need to be in control of the interface.
- 16:38So on the on the screen that the stores are seeing our business
- 16:43and I mean probably many businesses are saying that, but
- 16:46we have a few very, very specific demands like the whole
- 16:49ivary stuff, we have lenses, we exchange them, we upgrade them,
- 16:53we downgrade them. So it would have been impossible
- 16:56to implement that based on normal POS because was typically
- 16:59our patients have they have been one sales system and one system
- 17:03where they order lenses that's totally or two totally separate
- 17:07systems. So in our evaluation that we've
- 17:09had is we had I would say two key criterias.
- 17:13The first one was flexibility. So we looked for a headless
- 17:16commerce systems and they are essentially all built for online
- 17:21sales. But our reset whatever we want
- 17:23to build as long as we can can cater for our stores, we will be
- 17:27able to do it online. We're not so worried about that.
- 17:30Our online business is not that difficult.
- 17:32So if we get it right with the stores, we will get it right
- 17:35for, for, for the online business.
- 17:37So we really were looking for a headless commerce solution which
- 17:41was flexible enough that we could use them for the stores.
- 17:44And that was the real critical part.
- 17:46So how how many things does as a solution assume that it's just
- 17:50the way it is? And many of them they were so
- 17:52online driven, it actually was quite hard even to imagine how
- 17:55do you put the store in there. And the second part was the
- 18:00order management. Order management, many of the of
- 18:03the commerce suites, they say order management is a different
- 18:07type of of beast. We don't do that.
- 18:10You need to go to a system XY set and do that there.
- 18:13But in our case, because we have so many returns, we have
- 18:16exchanges, etcetera, the order management is such an integral
- 18:20part to what we're doing. And I had the very much a
- 18:24feeling and actually like in retrospect, I would say even
- 18:26more, it was one of the most important decisions.
- 18:29As he said, the order management really needs to be very close to
- 18:32the sales part and it needs to belong to a sale.
- 18:35If we have a different system for order management, that would
- 18:37make it so much more complicated for us to build that whole
- 18:41application that we actually wouldn't know whether we would
- 18:44succeed in that, especially in the time.
- 18:46So I would say flexibility from a headless API perspective or to
- 18:50really build the the whole store experience and the order
- 18:54management side to be really neatly integrated into the into
- 18:59the core commerce part. That was sort of our two key
- 19:02criteria. And with those, there's actually
- 19:03not that many left. And this, this shows the
- 19:06importance of it. And I, I speak to some of your
- 19:08business about this is you don't start off by saying, I want to
- 19:12be on platform X. You start off by going, what are
- 19:14my business critical risks of if I get this, what is my project?
- 19:18Is it a selling online project? Isn't it an Omni channel
- 19:20project? Is it a custom experience
- 19:22project? And therefore, what are the
- 19:23critical dependencies? Because that you're right, you
- 19:26filter down to like maximum of two or three instantly by
- 19:29getting those questions answered properly.
- 19:32Yeah. I mean, we obviously have done
- 19:34that obviously was we didn't just believe that because we
- 19:36knew that you're building something that they probably
- 19:38haven't built before with a customer in the way how, how,
- 19:41how we plan to do it. So the first thing was that
- 19:43we've done was just to build an MVP on that because we had a
- 19:46good idea how it should look like at the end.
- 19:48So we've we've took like a week or so actually let's guess we
- 19:51had like 3 days maybe I had actually support from the
- 19:55commerce layer people to remote this or to ask answer questions.
- 19:58We've been able in those three days actually to build up some
- 20:00kind of like a first store. But we could sell a few things.
- 20:03So you could return something where we, where we would, where
- 20:07we've been able sort of to use the order management part in the
- 20:10way how we intended it. Because we, we really, because
- 20:12we have to assemble the, the, the, the glasses at the end as
- 20:15well as essentially creating a shipment per glass and
- 20:18everything needs to come together.
- 20:20And that's like one piece that needs to be worked on.
- 20:23So we've really been like within the first 3 days able to, to
- 20:25show that we are able with the basic model to, to do that and a
- 20:30few bit, a few smaller adaptations that actually went
- 20:32in relatively smoothly. Excellent.
- 20:35So let's prove a bit more on that because one of the areas we
- 20:38talked about is corrections order changes, they are
- 20:41challenging because in in most businesses it's basically right
- 20:44Council 1st order create new order.
- 20:47How are you using Commerce Lend to help you with that order
- 20:49management piece? So probably what does I mean, I
- 20:54would say the whole after sales area like around glasses, it's
- 20:57like the important part and we have such in in the after sales
- 21:00world world we have warranty where we say that is there's
- 21:03something broken relatively simple.
- 21:06We just reordered the same product essentially in commerce
- 21:08they we just recreate base in the same order.
- 21:11We recreate a new shipment which does like everything like how it
- 21:15is being transported also probably from, from the, from
- 21:19the, from the warehouse to the store to essentially reorder the
- 21:22same piece without any financial implications.
- 21:25That opposite there's returns, but what it makes really
- 21:28complicated is sort of, we call it the goodwill or intolerance
- 21:31exchange. So you, you buy a pair of
- 21:34glasses and after a week you're coming back and somehow it's,
- 21:37it's it, it doesn't fit exactly. So that can be a mistake in the
- 21:41prescription. It can be that you've even like
- 21:44with depending on them if you have a metal frame that you have
- 21:47an intolerance with the metal frame.
- 21:49So we are exchanging that actually relatively for a long
- 21:53time. So in that case also it's not
- 21:56only exchanging something, but it's also it could be the new
- 21:59product could be more expensive, it could be it could be cheaper
- 22:02than before. So what we essentially built
- 22:05based on that is it it logically does create a new order, but it
- 22:09is it is logically connected to the initial order, which also
- 22:12allows us then to do a payment. So in reality, it does create a
- 22:18return internally and return sort of the the value of that
- 22:23frame onto a gift card. And the new order is
- 22:26automatically being paid with that gift card.
- 22:27With that. We also have the real link for
- 22:29that. We also have the real payment
- 22:31date on top of it. For example, if the if the, if
- 22:33the, the exchange would be more expensive or because could other
- 22:37do an essentially a refund of the of the rest of the gift card
- 22:40if it if it became cheaper. So all those ways we've actually
- 22:44been able to find well to to model that within within the
- 22:47data model of commerce layer. Nice.
- 22:51I think anything it's just it's about simplification for the
- 22:53people handling it on a day-to-day, isn't it?
- 22:56And let's talk about some some other ways of like improving
- 22:58through automation because you, when we were discussing this
- 23:01podcast, you're talking about, they wanted to create efficiency
- 23:04they see. So can you share some examples
- 23:06where you're using the new text act to automate key processes or
- 23:09removing points of failure? One example you gave me is about
- 23:12the automation of lens ordering, which I hadn't.
- 23:14It would be so complicated. Yeah, that's exactly that.
- 23:18Initially that sounds like if you, if you hear it sounds very
- 23:21easy. Now in, in most applications,
- 23:23it's actually is the case that you have like a sales system
- 23:25where you say, OK, I want to have like a complex correction,
- 23:28I want to have this isn't that. And and then you have a totally
- 23:31separate system which is often provided by by a lens supplier
- 23:35where you actually order the lens and they're not really
- 23:37connected with each other. Now we're ordering from multiple
- 23:41lens suppliers and we said we actually really want to have
- 23:44that connected to make sure we are not doing any mistakes.
- 23:47So in the past it used to be the case that in the stores that
- 23:50they've only done sort of the, the initial, the sales piece and
- 23:55that everything went to our operation centre and the
- 23:58operation centre actually took that and then manually ordered
- 24:01the right lens for that based on various Google Sheets and stuff
- 24:06like that. And like initially, as we said,
- 24:08okay, that's the first thing we need to automate.
- 24:10And they were, they were like, no, no, you can't automate that,
- 24:12that that doesn't work. And it was like, So what you're
- 24:14telling me essentially is it's like you order lenses back out
- 24:18feeling. No, of course not.
- 24:19That's, that's there's science behind that.
- 24:22So, okay, for science behind that we can automate it.
- 24:25So, so we went behind that. I mean, we have at the moment
- 24:28around 1800 different combinations that we have what
- 24:33we order on lenses, but but broken down to the mineral of
- 24:38decisions that you have to take, which obviously every occasion
- 24:42easily can take, but then fully integrated into.
- 24:45And that's like the reason why we call it not a POS, but the PO
- 24:49E for point of everything system, which automatically
- 24:52based on the sales process, you choose the frame, you choose
- 24:55your prescription, you choose what type of lens you want to
- 24:59have and automatically calculates from which supplier
- 25:03we could order them. And adoptation can change, alter
- 25:07if he wants to. The lens supplier needs to give
- 25:09a reason. Sometimes it doesn't make sense.
- 25:11For example, if you have a son and a Noel prescription glasses,
- 25:14you want to have the same supplier.
- 25:16So you can influence it to some extent.
- 25:18But it's everything is sort of predefined like what is the,
- 25:21what's the default supplier we have for that combination,
- 25:23etcetera. So it's a relatively complex
- 25:26process behind that. I love that I I worked on a
- 25:29project a few years back, a jewellery company who did like
- 25:32custom engagement rings. And at face value, when you
- 25:36think about buying a ring, it's a simple, I'll pick my size, job
- 25:39done. But actually engagement is very
- 25:41similar to what you're saying, though there's complexity of
- 25:44logic around what combinations fit and how that drives price
- 25:48and price change. And I remember working with so
- 25:50Tom and Kay were the Magenta agency and an amazing and lead
- 25:54engineer architect called Jacob who what you did in converts
- 25:59basically spent the time to understand the logic and the
- 26:02data rules to map them out, to define, which then enabled the
- 26:06smart people in that business, including Yakub to create the
- 26:10programming that then calculate. And when you when you spend that
- 26:13time in technical discovery and spend a bit more money, a bit
- 26:15more time, a bit more effort, the savings you get later on are
- 26:19massive. Exactly.
- 26:20And just like the mistakes that you make with, with ordering
- 26:23from lenses, which you anyway have to pay, umm, is, is that's
- 26:26a message saving other like, for example, like depending on the,
- 26:30if you have like a large frame and depending on your view
- 26:32height and depending on the prescription data, it works or
- 26:35it doesn't work. So all those things that we
- 26:37actually able to calculate that upfront and figure out does that
- 26:40work or doesn't it work? Can we do that?
- 26:43What kind of do we need thin glasses for that or not
- 26:46depending on on prescription. So all those things to, to have
- 26:49like a real rule set behind that, which opposite the
- 26:51opticians in the verse of the they have have that.
- 26:54So they've done that or just all manually, which also means like
- 26:57between the countries, actually, it must not ever being done in
- 27:01exactly the same way. Because there's, there's always
- 27:03multiple truths and there's always like those things that we
- 27:06actually had them to agree on. OK, look, guys, now get together
- 27:09in, in Germany, we're doing it that way and Swiss will be doing
- 27:11it that way. Figure out what the truth should
- 27:14should be, because in the future, there's only one truth.
- 27:17Yeah, I I love that now it's a really good kind of there's a
- 27:20great standby to take away from this episode for people is it's
- 27:23it's challenging current ways of doing things because we all get
- 27:27into bad habits, not through bad reasons, but because pressure
- 27:30time, everything and to replicate bad process into a new
- 27:35new tech stack is just, it does not make sense.
- 27:38No. And that is actually spot on.
- 27:40And that's also we actually didn't even look at, we went
- 27:42that far. We didn't look at the old
- 27:44processes. We actually didn't care.
- 27:46We we just essentially wrote up, OK, what does the process need
- 27:49to do? OK, we need to be able to
- 27:51assemble glasses for this data like for some parts and for
- 27:54example, like if you have a shiny glass, a shiny frame, you
- 27:58you're able to met in it. And that obviously can be only
- 28:00done in the workshops where they actually have all the devices.
- 28:02So, so all those, those rules that they are behind, but in, in
- 28:07reality, we all know what's happening and we are all looking
- 28:10behind that. Yeah, and.
- 28:12And essentially sketching that process and especially the after
- 28:16sales process was a little big challenge obviously with all
- 28:18those, those variations. But then sketching up that
- 28:21process really from scratch and then figuring out how do we
- 28:24implement that. That was, was one of the most
- 28:26important things I would say that we've done.
- 28:29And really not seeing it as just a tech project.
- 28:31It was. It was fully, fully from the
- 28:33beginning on a tech and process project.
- 28:37Yeah, and this is where where engineering leads, not just
- 28:41developers, but the engineer leaders you've got the
- 28:43architectural vision and the strategic lens are worth their
- 28:46weight in golden projects because it's that.
- 28:49OK, how do we deliver this best? Because everyone can build a
- 28:52solution, but building the right solution takes thinking to to
- 28:56come out and iron out all the nuances.
- 28:59And one thing is capability and the other part is actually what
- 29:01power do you have? What can you do?
- 29:03And I would say that was really one of the big advantages.
- 29:05So we've we've, we've had sort of the full power over
- 29:08everything. And if we could change
- 29:10everything in the way how we thought it was best.
- 29:14Yeah, excellent. Let's come back to that after
- 29:16sales process because you said you made improvements the after
- 29:19sales process and that helped you like reduce back office
- 29:22costs. Can you talk us through what you
- 29:24did and what impact it had on the business?
- 29:27So in the past essentially that was every after sales process
- 29:30and that was writing a ticket. So no one actually.
- 29:34So it wasn't really being done too much in the system.
- 29:36But essentially everything had to do then with with manual work
- 29:40of the operations team. Even like with normal sales
- 29:42orders, more than 50% of all orders had a ticket attached to
- 29:47it, which means that that operations team manually had to
- 29:50do something with that. So what we've done essentially
- 29:55because we've, we, we were able to really bring all those
- 30:00processes fully into the application.
- 30:02And the principle we've applied from the very beginning on was
- 30:05it said, I don't want any functionality which stores are
- 30:09not able to do it them to do themselves.
- 30:11Which means we also had to build them in a very simple way
- 30:14because you don't want to train like a, a workforce of like 2-3
- 30:17hundred people on that. But it, it has to work for them.
- 30:21So I didn't want to have any functionality where they would
- 30:23have to go back to a support team and say, hey, can you do
- 30:26this? Isn't that for me?
- 30:27So the default answer always had to be no, you can do it
- 30:30yourselves. So and really bringing all those
- 30:33things that before they wrote a ticket, oh, OK, someone will do
- 30:36it for me. Suddenly they were able to do it
- 30:38for themselves, which obviously also from beginning on, it was
- 30:42like for them kind of a pro and a con.
- 30:44On one side, they could serve the customer directly.
- 30:46On the other side, they actually had to do it.
- 30:48And there was no one controlling what they've done.
- 30:50So there was a bit like less double checking on that.
- 30:53But he obviously checked many things from a technical
- 30:55perspective. So they, they barely could do
- 30:57any mistakes for that. But making sure that really all
- 31:00the functionality is part of a very simple interface, we've
- 31:05been able to hugely improve the, the, the cost on the whole
- 31:10operations process because we, we just, we don't need that many
- 31:14people anymore. And let's get to to a specific
- 31:19bit of technology because you talked about going headless was
- 31:23the right architectural approach.
- 31:25So you got to commerce layer and you've adopted a composable
- 31:28strategy to have like elements that fit together well to
- 31:31deliver the capabilities. And sanity, I know is a critical
- 31:33part. It's your source of truth for
- 31:35content. So how are you using sanity and
- 31:38like what it what? What changes this had on the
- 31:40business from a content management point of view?
- 31:43You know, if you if people here sanity it's they immediately
- 31:46think website CMS. So that's essentially what it's
- 31:49built for. But we're having a relatively
- 31:51small team. So one principle is also, we
- 31:54don't want to have a specific tool for each and every use
- 31:58case, but you want to use a tool if you use it for many, many use
- 32:01cases. So in the meantime, we're using
- 32:03sanity also the websites CMS. But our first use case is about
- 32:07actually to manage the whole product data and to manage our
- 32:10stores. So essentially, instead of using
- 32:13an, an, an off the shelf product data management tool, we've
- 32:16built that on top of sanity. I would say actually in, in in a
- 32:20very, in a relatively complex way.
- 32:22So we have, we have products, we have variance of it, we have
- 32:25sort of designs which can make up the varieties.
- 32:28We have all the prizes in there which come in directly from the
- 32:31from business central from the, the, the price mastery system.
- 32:34But everything sort of from the very beginning on, if the
- 32:37product product team is sort of creating their first prototypes
- 32:42until the, the very marketing material, everything is sort of
- 32:45fully managed based on sanity for that.
- 32:49Instead of once we activate insanity, the queues are pushed
- 32:52into commerce layer to be able to sell it and type sense, which
- 32:56we're using as a as a, as a catalogue system or search and
- 32:59everything which empowers both sort of the, the, the store
- 33:03application as well as the website.
- 33:06I'm interested so you've you've created a a component insanity
- 33:09with the logic to control the product data structure and
- 33:13pushing the I guess the data into it so that it then displays
- 33:16it how you need to in the front end.
- 33:19Exactly. So some various insanity, you
- 33:21really have sort of that, that, that product and and variation
- 33:25model and that is being flattened into ASQ model for the
- 33:29product, for the product catalog.
- 33:31Because in most we actually talk about singles queues, although
- 33:33they're obviously still being held together.
- 33:35But if we we want to sell it, it's anyway SQ.
- 33:38But if we, for example, want to generate a Google product data
- 33:42catalogue, we essentially it's also Q based.
- 33:45So everything that that runs after the data management is Q
- 33:48based. So we have actually do a
- 33:49translation if it's being activated insanity, but they'll
- 33:52translate that only into SKUs and those singles queues are
- 33:56then being held in type sense which we use and then sort of as
- 34:00a product catalogue to browse that.
- 34:02Yeah. OK.
- 34:02Interesting And composable is often associated with large
- 34:07engineering teams, rightly or wrongly, it is.
- 34:10Can you talk us through your team size structure for
- 34:13development? Yeah, that's indeed a question
- 34:16as we getting quite often. So as a choice, I would say
- 34:18there was quite an engineering team which is way too large for
- 34:21what what has to be done. So it's reduced that quite
- 34:24heavily. So essentially we had a team of
- 34:27two which were more like, so the product owner, business analyst
- 34:30type of people, three software engineers, a UX and a product
- 34:34developer, a designer myself and sort of with that team of seven
- 34:40rebuilt sort of Windows 18 months like everything from the
- 34:43core commerce, the product data management changed the whole
- 34:47transaction signature is what you need for POS systems in
- 34:49Germany and Austria, build the whole PowerPoint of everything
- 34:53application, the website, apartment booking, data
- 34:56warehouse, analytics website, outsource the warehouse change
- 35:00fulfillment processes. So I would say it was a
- 35:02relatively lean team for that for for what we accomplished
- 35:06within those 18 months. Yeah, that seems quite great.
- 35:09And that's you're not using any external development like
- 35:13resource in addition to that like agencies to back up around
- 35:16source parts to. It's all internally in your
- 35:17team. No, it was fully done
- 35:19internally. The only exception was for the
- 35:21ERP system. So we also migrated from
- 35:23innovation to to business central.
- 35:25That's why we used a consulting partner.
- 35:28Everything else including all the designs of the for the
- 35:30website and for the for the for the application were fully done
- 35:35in house, which also in that speed, I think it would have
- 35:39been quite difficult to integrate externals with that as
- 35:42well. Yes, yeah, that process wise
- 35:44that gets really challenging especially to manage like the
- 35:46development pipelines at the speed.
- 35:49And interestingly from I speak to a lot of I, I find it
- 35:52fascinating the the cadence of which engineering teams release
- 35:55because you go back a few years and it was you'd have a big
- 35:58release every two or four weeks. Then people started to go, well,
- 36:01we can release every week. And then actually some
- 36:03businesses who truly adopt this, this kind of more agile approach
- 36:07to development or potentially releasing every single day if
- 36:11not multiple times a day. How do you approach it?
- 36:13What is your kind of frequency? Or does it change depending on
- 36:16the business needs? We initially it was always
- 36:20multiple times a day. Obviously we had to have to fix.
- 36:23Oh, again, like an exception, we need to change something.
- 36:25OK, let's, let's ship it quickly.
- 36:27So we've like initially I would say we deployed probably even
- 36:30like 8 to time to 10 times a day to production.
- 36:34Today we're probably down to I would say pretty much daily.
- 36:37There's a few exceptions. If we have if having something
- 36:40on test that we don't want to deploy it, then then we probably
- 36:43don't deploy once for two days or so.
- 36:45But in general, we deploy at least once a day, often more
- 36:49than once a day. Yeah.
- 36:51And I think for I know a lot of senior econ people who are used
- 36:55to come in from legacy tech stack world, where actually then
- 36:59getting your head round weekly is a bizarre thing because still
- 37:03most of those primes, it's two week sprints, but daily a lot of
- 37:07people go, how can you possibly do that and manage the code and
- 37:09make it the developments accurate so you don't create all
- 37:12this QA overhead? What, what, what I guess what's
- 37:15your, what's your voice on that? How do you how do you do that
- 37:18without creating so much QA work that it slows you down?
- 37:23Yeah, we, we actually also never planned in sprints because there
- 37:26were more so many things that changed in priority.
- 37:29So actually managing sprints would have been at the end just
- 37:33an overhead. We we were working with a quite
- 37:36normal Kanban style and everyone had a lot of his top, top
- 37:40priorities. We shipped as quickly as
- 37:42possible, as crazy as possible making sure that we are testing
- 37:45everything at the same time. So other the the, the POS or the
- 37:48business analysts, they were also working on essentially on
- 37:51the QA side for that, although we had a dedicated QA initially,
- 37:56the complexity of those things is so high that it was almost
- 38:00impossible for them to actually create valid tests for that.
- 38:03Now in the big advantage also of a store application is that you
- 38:07have a whole test crew out there with your old employees and with
- 38:12in combination which we're able to ship other changes pretty
- 38:15quickly. We've also had quite a few
- 38:17things that OK, it's good enough.
- 38:19We've tested a bit a few HHH cases, but now for the real
- 38:22test, we'll ship that out. We'll say, hey, look here is the
- 38:25first version of that test that give us feedback.
- 38:28And we've done that quite constantly.
- 38:29And I would say initially they had to get used to it a bit, but
- 38:33that actually worked super well. And also the, the quality of, of
- 38:36feedback increased quite heavily.
- 38:38So we've, we've been working there quite dynamically with the
- 38:41people. Like I said, at the moment we
- 38:42rather have to make sure again goes a bit more into operations
- 38:46mode and that we're not getting all their feedback from
- 38:49everything what they would want to have on a daily basis
- 38:51anymore. I've found that really
- 38:54interesting. I love the fact that every
- 38:55business has a different approach, especially when people
- 38:57got internal engineering. I'm going to finish around like
- 39:00the customer experience theme because I know that's been a
- 39:02really important driver of this, this change and and the
- 39:06technology section piece, But you touched on it earlier about
- 39:09point of everything is like the North Star for the business.
- 39:11Can you give the listeners a bit of context?
- 39:13What does point of everything actually mean in the context of
- 39:16the customers? Absolutely.
- 39:18So customer, I mean, first of all, in terms for the employee
- 39:22or for, for the, for the, for the stuff in the store.
- 39:25And before they had essentially an order application, they used
- 39:28Presto shop to manage the orders.
- 39:30They used Navision for a day and closing.
- 39:32They had an another tool for the appointments, inventories and
- 39:35many other things were done by Google Sheets.
- 39:38And again, like as mentioned before, for 50% of the orders,
- 39:40they anyway had to write a ticket and ask someone else to
- 39:43do something. So for us it was really to bring
- 39:46all those things together. So whether it's sales, whether
- 39:49it's after sales, the inventory management appointment, booking,
- 39:53the campaign management day and closing, the analytics and the
- 39:57whole, the whole store, KPIs, etcetera, etcetera.
- 40:00So everything that the store somehow needs to do during an
- 40:04entire week that has to be that should be done via the
- 40:08application. And let's say we've we've
- 40:09succeeded actually quite, quite well with that.
- 40:14OK, no, I I think it's so important to have a concept that
- 40:18you can pin stuff around because that's, I guess that's the
- 40:21success project. Are we achieving our vision?
- 40:23Yeah. And just probably to give that
- 40:26like one concrete example what that means for the.
- 40:28So for example, for the appointment booking.
- 40:30So if you have like separate tools, it's separate tools in
- 40:33our case because we've also integrated, we're obviously
- 40:35using a separate system for in the back end for that, but we're
- 40:38using the same front end system. So in the sales system, like you
- 40:41have like a header where you see, oh, what is my my next
- 40:44upcoming appointment. So also while selling, you
- 40:47actually know, do I have an appointment in the next, you
- 40:49know, 10 minutes or so, you obviously tap that, that that
- 40:52appointment, you're directly logged in, assess that customer.
- 40:56The although that they were chastered by an appointment
- 40:58booking system, the customers already created in the commerce
- 41:02system. There's no data that needs to be
- 41:03transferred somehow. Everything is there.
- 41:05We know everything where the customer was coming from,
- 41:09Adaptation just has to add the prescription and everything is
- 41:12being added automatically to the order.
- 41:13So, so those things and really making those processes just way
- 41:17leaner, way faster, easier for for the the stuff in the stores
- 41:22that has been been proven to be super valuable.
- 41:25Yeah, amazing. And and helping them to better
- 41:27service customers. I'd love to know because you
- 41:31talked a lot about technology change, operational efficiency,
- 41:36being able to deliver a better overall set up to, to label
- 41:38staff, to deliver better service to customers, simplifying all of
- 41:42all of this stuff, which benefits the business, but also
- 41:44has a positive impact on customers.
- 41:46How, how are you? Like, how are you getting
- 41:48feedback loops to know what impact it's having with
- 41:51customers and how they're perceiving it and how it's
- 41:53benefiting them? Like, you know what, what do you
- 41:55see? Like the customer behaviour
- 41:56says, actually do you know what? We're going in the right
- 41:58direction. This is landing well with
- 42:00customers as well. And it's, it's, it's really a
- 42:03bit twofold because we we are essentially a drive to store
- 42:07business. So other what's for the website,
- 42:10the number one, Target isn't even to sell much more wireless
- 42:13via the website. What we want to drive the people
- 42:16to the stores. So one of the one of the things
- 42:19that we're looking at is like do essentially then obviously
- 42:22conversion rate in the stores. Are we sending the right
- 42:26customers to the stores? Do we get the right, the right
- 42:28customers there? But also online, I would say
- 42:32pretty much with the typical KPIs, making sure what are they
- 42:36doing. And in our case, because we have
- 42:39lots of prescription data, we actually know about the customer
- 42:43and whether he's coming, coming back.
- 42:44And if he's locked on, once we have a very good idea, we have a
- 42:47better, better, he actually will come back.
- 42:50But he ordered something later on.
- 42:52If he had, for example, an experience with, with exchanging
- 42:56a lens, is he coming back? Is he not coming back?
- 42:58So we have lots of data on the customer directly with that,
- 43:01that that helps us. But we obviously also using
- 43:04other tools to figure out sort of where are people dropping on
- 43:08the website, what doesn't work with them.
- 43:11We're we're constantly monitoring the website also for
- 43:14what happens. We have quite a lot of lots of
- 43:17historic data where we figure out or where we know sort of
- 43:20what have been the typical ratios in the past and
- 43:23monitoring quite extensively against that.
- 43:27Amazing. Interesting.
- 43:28Yeah, I guess there's so many data points on there where you
- 43:30can measure customer satisfaction, customer
- 43:32experience and impact. Ellie, a lot I've really, really
- 43:36enjoyed. I mean, I could carry on asking
- 43:38questions for hours, but I know you're a very busy man, so I
- 43:41should be very respectful that and not take up too much of your
- 43:43time. But I've really enjoyed that.
- 43:45Thank you so much. I love the brand as well.
- 43:47I love the story that you've shared and I know that listeners
- 43:50will get a lot of value out of it.
- 43:53And I do recommend anyone going and checking out the
- 43:54prescription sunglasses are beautiful.
- 43:56You'll get you'll get a hooked to browsing them as just as I
- 43:58have. So yeah, thanks for coming on
- 44:00the podcast, Selium. Thank you very much, was a real
- 44:03pleasure. Thanks a lot and looking forward
- 44:05for further episodes. Excellent.
- 44:07And to everyone listening, hope you enjoyed it.
- 44:09Do reach out any questions. We want to probe further.
- 44:12And I guess if anyone's out there who's been looking at
- 44:14headless composed war and considering some of that comma
- 44:17Slayer and wants to ask a few more questions about the
- 44:20practicalities of it, feel free to to connect.
- 44:23I'm sure Ellian's speaks very eloquently about the that whole
- 44:26change management process. It would be a real pleasure.
- 44:30And I mean, you've also been in combat quite a few sort of
- 44:33competitors from the landscape, but it's always been super
- 44:36interesting exchanges on that. I'm happy to exchange
- 44:38experiences on that. Cool.
- 44:41Thank you. You're listening to the Inside
- 44:44Commerce podcast, hosted by experienced consultants James
- 44:47Gert and Paul Rogers. Inside Commerce is your fast
- 44:50track to digital knowledge, featuring interviews with
- 44:52leading e-commerce retailers, industry thought leads and the
- 44:55C-Suite, key tech companies to explore the latest digital
- 44:58strategy, customer experience and tech.
- 45:00Discover what the latest trends really mean and get practical
- 45:03advice based on experience to help fine tune your ecom
- 45:06strategy and tech stack. Follow us on LinkedIn and reach
- 45:09out to James and Paul if you'd like to discuss an ecommerce
- 45:12project. We hope you enjoy the latest
- 45:14episode.