Latest / Inside Commerce: Ecommerce Strategy, CX and Technology Podcast / EP221: Optimising Your Product Catalogue: Building Data Models and Nailing CX & Efficiency Wins
Transcript
- 0:02Welcome to the Replatform Podcast sponsored by Patch reps
- 0:05hosted by me, James Get and Paul Rogers.
- 0:07Each week we bring interviews with industry thought leaders,
- 0:10all the CS with your tech companies and and retail
- 0:12e-commerce leaders. We also do like Chalking Heads
- 0:15episodes where me and Paul discuss like key strategic areas
- 0:18around e-commerce and that's today's episode's going to be
- 0:20about in return. All we ask is you to other
- 0:23people. We'd love you to get more people
- 0:24listening to the podcast for us and if it's your first time,
- 0:26give us a rating really helps boost visibility.
- 0:29So we got quite broad topic today.
- 0:32It's about product data and using like attributes, custom
- 0:36fields, tags, whatever the terminology might be developed
- 0:39on the platform, using to improve and in rigged product
- 0:42data to drive smarter automation and to help the business achieve
- 0:46better capabilities around product capital and management
- 0:48and trading. And we recently talked about
- 0:50fixing a particular issue related to product data with the
- 0:54guys at Swift. So that was in in stock but not
- 0:56online using smarter automation and this linked to really nice
- 1:01of that. We're getting deeper into
- 1:02product data itself and looking at how e-commerce teams works on
- 1:05the challenges they face with technology, the things that they
- 1:08want to do and sharing some insights from what we've seen
- 1:11work well and what we know from working with platforms and other
- 1:13E com teams and covering things like data trade and also looking
- 1:17at elements like site migrations.
- 1:19So let's take you on our ramble through product data
- 1:22optimization and shall we start at the let's start at the
- 1:25beginning. Oh, which is why is source
- 1:28product data often not fully fit for e-commerce?
- 1:32What gaps do you typically see retailers and e-commerce teams
- 1:36have and what impact does it have?
- 1:39Yeah. So I guess the biggest one that
- 1:43a lot of this started with and I'm sure you've worked with
- 1:46loads of businesses as well as a way where basically they just
- 1:53have a quite a high proportion of attributes or field.
- 1:57So just missing for like certain groups of products.
- 2:02And I think you see it less with some of the like newer brands
- 2:05where they've kind of started with a small catalogue, had
- 2:08grown out and they've kind of always had a certain set process
- 2:11in place that with a lot of the more traditional brands that
- 2:14kind of moved online and direct has historically been a smaller
- 2:17part of the business. Yeah, We've definitely had a
- 2:20number of clients that have been particularly poor and usually
- 2:24it's the, I guess the richer data.
- 2:27And we have one client that I loosely commented on here who
- 2:31are like a massive brand, great business, amazing product, but
- 2:35they're, they're a wholesale business.
- 2:37They're maybe 2% of the like global demand for the product
- 2:42and all of their resellers have like significantly better
- 2:46product data. And I flag it all the time like
- 2:48little things. So like compatibility of the
- 2:51product with other products you might store within the product,
- 2:55things like that. Like there's none of that on the
- 2:57direct side that all of the kind of resellers have, you know,
- 3:00seen enough rationale to, you know, actually invest in this
- 3:03stuff. And I personally think it has a
- 3:06big impact on their direct selling.
- 3:08And then we've got another client who's pretty similar or
- 3:13was pretty similar. It's a luxury watch brand and
- 3:16they probably had like 2530% like completeness within the
- 3:20product. And the impact on the customer
- 3:23experience, which we'll come on to was massive.
- 3:26And I think they're the perfect example of some of the stuff
- 3:28we're going to talk about, which is like wise.
- 3:32Yeah, I think it's really nice. And do you know what what I find
- 3:34really interesting is still there are enough e-commerce
- 3:37teams that don't have a standardized checklist of what
- 3:43the minimum eligibility criteria are for for products of it.
- 3:47It can't go live without ABCDEF and having that that gate, that
- 3:51stage gate check of homemade, we we fundamentally haven't got the
- 3:54attribute data right. We can't launch this there.
- 3:57There seems to be, and I know I get it, our trading pressures of
- 3:59get the products live, get products live.
- 4:01It needs strength in a team to push on other people business
- 4:05and say we only get it live when it's fit for trading and that
- 4:09takes some subtle one manipulation of personalities.
- 4:15One thing that's always surprised me is that we
- 4:17obviously talk a lot about like that like what's happening
- 4:20within the platform space and like where certain platforms are
- 4:23strong or like are investing etcetera.
- 4:26And I'm really surprised that because you look at like all the
- 4:30advancement around things like personalization, content
- 4:33management, like campaign management, all of that kind of
- 4:35stuff. And I'm really surprised that
- 4:38one of the bigger platforms has never created like
- 4:41differentiation around catalogue.
- 4:43So some of the like obvious functionality you might expect
- 4:46in a PIM like building that natively and then I think Dan
- 4:49was then maybe drive the market to start doing more of that.
- 4:52But yeah, I'm surprised that there hasn't been platforms that
- 4:56have like, you know, had some of those validation layers in
- 4:59place, like some of the processes around like reporting
- 5:02back on like completeness and all of that kind of stuff.
- 5:07So you've basically just pitched a new feature enholdsment to
- 5:11some of the legal platforms to challenge them to get this done
- 5:13right? What what, what?
- 5:16What should we tackle next then? Where we going next on this
- 5:18discussion? Yeah.
- 5:21So I think the one that we've got next I added in here as a
- 5:25bit of a random extra to your organized list of questions was
- 5:29like who should be responsible and how should a business
- 5:32approach it. So I talked about the scenario
- 5:35where you've got a business of like 30% completeness in their
- 5:38e-commerce catalogue, but who do you think should be responsible
- 5:41and how do you think that should?
- 5:44Be OK, so I'm going to, I'm going to do the usual consultant
- 5:47plaques and it depends because it depends on the size of the
- 5:50business and the size of the team.
- 5:51So I've got a client at the moment really small, they've
- 5:53literally got one person leading it and an admin person on the
- 5:57e-commerce side. That is it.
- 5:59Now I will split that out because I'm currently doing a
- 6:02process where we've got to sort out the product data and that
- 6:05process of getting products on to the right level complete.
- 6:07So #1, the person who has commercial responsibility for
- 6:11the channel needs to have ownership of the process for
- 6:14defining what the standard is. If you don't have a standard you
- 6:17can't complete new product data against it.
- 6:20So that that's you shouldn't give that to a genie person.
- 6:23Expect them to to have that knowledge and experience work
- 6:25with them to support them to understand what's required and
- 6:28why and show them what good looks like from other sites.
- 6:31But you as the as the commercial owner for that need to set that
- 6:34standard. So in this client they are
- 6:36affected the E com director role, then the the web
- 6:39assistant, web exec, whatever the job title might be then
- 6:42takes ownership for delivery against it when new products
- 6:45land. And they basically should have
- 6:48ownership of the relationship with the people who are
- 6:50providing the source data which sometimes is external suppliers,
- 6:54sometimes it's internal buyers. So it really varies from
- 6:58business to business to better on how how large and structured
- 7:01the teams are. But yeah I'd like to split it
- 7:03out between like own the process and the standards and only
- 7:06execution. That's mainly because at the
- 7:09moment I'm working with a lot more SMEs and not massive
- 7:11enterprise clients at the moment.
- 7:14And in terms of I know you said about do you bring in a PIM and
- 7:17I've this I find an interesting because often PIM is in to fix
- 7:21the wrong issue and you don't need the PIM.
- 7:23You just need to set up the product data and your structure
- 7:26and not spend on a license you don't need.
- 7:29I'm not denigrating PIMS because I I'm a big proponent of them.
- 7:32In the right use cases where you have like, you know, complex
- 7:36more complexity around pro channel, sorry, different
- 7:39channels, different product datas, different data feeds.
- 7:42When you've got translations, when you've got localisation,
- 7:46when you might have Excel third parties, you need to source and
- 7:49access product assets and you need that controlled
- 7:51instruction. So there are use places where
- 7:53PIM is useful, but half the time that's spacing mate, it's
- 7:55because there aren't enough quality controls.
- 7:58And so where's your head? What are you saying with your
- 8:00your client base? Yeah, I mean I completely agree
- 8:05with you just said one thing I guess we should add just for
- 8:08anyone that doesn't know what PIM is essentially is product
- 8:12Information Management system and it's like a master for
- 8:15product information. And as James said, probably I've
- 8:18seen maybe 90% of the people of our clients that use APM use it
- 8:24because of process limitations more so than you know needing to
- 8:28have the functionality within the PM or needed to like maybe
- 8:33past data around etcetera. I think going back to that watch
- 8:36example. So one thing and I think you're
- 8:39much better at you know driving change and process and being
- 8:44organized generally than I am. But I think with that watch
- 8:47example or just from like a practical standpoint.
- 8:50So it was quite interesting because it was a bit of a legacy
- 8:52issue because you've essentially got stores wholesale and then
- 8:57online and they're very, very separate and they've got very
- 9:02separate rooms. And I would say that direct was
- 9:06very heavily impacted by this lack of data and even things
- 9:11like the way they were structuring the navigation was
- 9:13like super, it was just wrong, like everything was like
- 9:17catering for existing customers and a lot of it was because of
- 9:20the lack of product data. So in the end what we did so
- 9:23took ages and they hired a head of E com which is the first
- 9:27senior hire in the E Com to that person advocated the need to do
- 9:31this and then they she basically got buy in from like the owners
- 9:37of the channels and then in the end exactly as you said.
- 9:40So we built out a sheet and we had a column per attribute.
- 9:45We started doing things like movement and you know strap
- 9:48material, all of that kind of stuff.
- 9:49And then basically they had to work with a couple of other
- 9:53people within the business sieve through loads of information,
- 9:56but it wasn't a small catalogue. And then basically once it was
- 9:59complete, the free senior stakeholders had to then sign it
- 10:03off. And then they reported into the
- 10:05E com platform and the ERP as well actually, which is quite a
- 10:10big change in how things are working.
- 10:13But basically once everyone had signed it off and it was
- 10:15complete and then the process with then in place afterwards
- 10:20ERP 1st to get the data in. But I would say that was a
- 10:24disaster. Like I think I've made that
- 10:26sound quite clean, but it was like major impact, took a long
- 10:29time. It's probably still not
- 10:30complete. But yeah, I think overall like
- 10:33usually there's a lot of variables that make this really
- 10:36complex and it's more like the echo law process or in different
- 10:40systems which are like, you know, it's not clear where
- 10:43certain things need to be entered, etcetera.
- 10:47Patchworks, the world's most modern integration platform.
- 10:49We can connect any key e-commerce system from
- 10:51storefronts to marketplaces and ER, PS to fulfilment solutions.
- 10:55This allows retailers and partners alike to simplify their
- 10:57tech stack integration and do away with point to point vendor
- 11:00specific apps. We can automate and streamline
- 11:02the flow of data across your entire e-commerce business.
- 11:05Find out more at We Are patchworks.com.
- 11:08Yeah, I think the political bits are really hard.
- 11:10One, because I've worked with businesses where you have
- 11:13friction between the whoever's sourcing the products, buying
- 11:17teams, merchandising, whatever the the the team structure is
- 11:20there and the people running the website.
- 11:24We've been trying to sell those products online and a a the
- 11:28friction being, you know, we want better product data.
- 11:31Well, no, that's not our job, but actually your job is to
- 11:34source the right products that sell and it's really hard
- 11:37sometimes to flip conversations to go well.
- 11:38Your products are not going to sell, you're going to have
- 11:41overstocks and you're going to have to discount heavily and hit
- 11:43your margins and not get as good trading figures.
- 11:46If you don't get this right because we know it, unless you
- 11:50have such a unique product and with such amazing demand where
- 11:54you can get away with with not good product data because people
- 11:57just want it which is rare if compared to market, then you
- 12:01fundamentally you've got to have it.
- 12:02Because otherwise people come onto a page you look at it and
- 12:04go, yeah that doesn't look the images are wrong or the product
- 12:07data doesn't really give you need or somebody who's got
- 12:10specific. We're getting to a bit more
- 12:12somebody who's got specific need especially around like a like
- 12:14dietary or lifestyle requirements.
- 12:17So let's look at the impact partial product.
- 12:19So let we know that you need good information because
- 12:22customers needed to make informed decisions.
- 12:24But let's look at the practical implications from trading in
- 12:27e-commerce store for what partial product data has.
- 12:30And the areas I normally focus people on when I'm stuck with
- 12:33clients trying to get into understand why their product
- 12:36data is not fit for purpose is number one is inaccurate search.
- 12:40So looking at our customers search data, what percentage of
- 12:43sessions is used in search? How does conversion rate compare
- 12:46for search to non search? What types of query are there?
- 12:49What are the matches, How many zero matches are there?
- 12:51How many people they have? I've been able to show them
- 12:55quite quickly some queries where they are not getting their full
- 12:58prop matches in there, they're missing products out of those
- 13:01matches or the relevancies of Rye and what the conversion loss
- 13:04conversion opportunity is. And this normally comes down to
- 13:07structure data like attributes. Then there's poor use of
- 13:10filters. You don't have the right
- 13:12attribute data structures in place, Whichever system you're
- 13:16using, whether you're using tags, attribute fields met,
- 13:19whatever it might be so that people can meaningfully refine
- 13:23filters on pages, and that filters are relevant to that
- 13:25specific product set and how customers want to search.
- 13:29Those are the two areas I typically focus on.
- 13:31Those are the ones I normally see the biggest issues where you
- 13:33might say if you go to thrill. I recommend anyone doing this is
- 13:37where you go through a search journey and you do the the
- 13:40comparative and browse journey. So I give an example to an
- 13:44online grocer where it was nut free chocolate was a popular
- 13:47search. Obviously allergy and food is
- 13:48huge. If you went through the online
- 13:51grocer into food cupboard into chalk and then filtered by like
- 13:55the nut free allergy, you had a totally different number of
- 13:57products and different products in those results for search and
- 14:01browse and some brands were not appearing in the browse journey
- 14:04versus the search. So clearly there is an issue
- 14:06with algorithms data driving that visibility.
- 14:10So that's where that's the what I know we see what what are you
- 14:13seeing around the the key impact partial product data has an E
- 14:16com teams. Yes, I think those two and
- 14:20probably data feeds to the free like highest impact, like
- 14:25consequence, but the big thing I'd say is probably more the
- 14:29opportunity as well. So a lot more platforms now are
- 14:34allowing for things like pool based category inclusion.
- 14:39A lot more people are using product data to support with
- 14:42automated visual merchandising maybe it's just the bottom of
- 14:47categories etcetera, maybe it's all or maybe it's just layers,
- 14:50but things like that. And then also even things like
- 14:56navigation like there's a lot of rules that can be put in place
- 15:00within ecom platform from third party and that can save huge
- 15:04amounts of time. And actually that product data,
- 15:07that luxury watch retail, the whole goal of it or the primary
- 15:12goals of it with the automate, like 50% of what people do it
- 15:16manually because you know they had free global websites and
- 15:20they were doing everything manually across three different
- 15:22stores. And I think it's probably like
- 15:25if you can get this in place suddenly you can start to look
- 15:28at some of these kind of automations that can like drive
- 15:31a better customer experience and reduce like manual overheads.
- 15:36But yeah, I think I think the impact's massive like and like
- 15:38you say there's variable as it depends on different industries
- 15:41at the silicon. And yeah, like I think if you're
- 15:44the worst I've ever seen, I don't think she'll mind me
- 15:47saying this because the company doesn't exist anymore.
- 15:49But so sham and my team used to work for BHS and BHS obviously
- 15:55went into administration and then came back.
- 15:57And when they came back they had to rebuild the catalogue and
- 16:00they were on an SAP platform. And essentially, one, it was
- 16:05really difficult for them to actually manage the process.
- 16:08And two, they had a really small team coming back into like an
- 16:12enterprise environment. But I remember when I was more
- 16:14involved with Claver trying to get their search working and
- 16:18their catalogue didn't have you know, gender like colour was
- 16:23missing in most cases like the most basic things.
- 16:25And I learned that certain people HS being really annoyed
- 16:29being like we're paying a fortune for Clavey.
- 16:31But it was literally impossible for Clavey to like unlike
- 16:36develop context. And I understand the product and
- 16:39so I think that's probably most dream, but I think it like goes
- 16:43up, doesn't it and influences so many different.
- 16:45Areas, Yeah, yeah. And interesting.
- 16:47You know, I I had a conversation relatively recently with a
- 16:51client who had been informed that that a couple of search
- 16:55tools were poor quality. Clay View was one, probably the
- 16:59other thing, I can't remember. The other one was mentioned
- 17:01because the search results were inaccurate.
- 17:03But the issue was not the search tool.
- 17:05It was a product catalogue and you know search providers.
- 17:08You know, picture search provider, Clay View, Algolia
- 17:10Search Green involved scores. It doesn't matter if you give
- 17:13them crap data. It doesn't matter how
- 17:15intelligent their algorithms are.
- 17:17You can't do anything because you're relying on on the active
- 17:20data. So it's critical when you're
- 17:22investing third parties and now the product data piece first.
- 17:26Complete. Do you know on that as well?
- 17:30Oh, sorry. Go on you go.
- 17:32Now you go. So one thing I was going to say
- 17:36on that as well is I think that so many people start to like
- 17:41berate their search tool without thinking about the potential
- 17:45impacts that data can have on it as well.
- 17:48Because long description. So a lot of people will have
- 17:52like equal weighting on things like title, long description,
- 17:56those two in particular. And then the long description,
- 17:58they'll have either really basic data or they'll have really
- 18:01random inclusions of like other terms and like stuff like that.
- 18:06I think if you start to face issues like that, you kind of
- 18:09have to work with the vendor because you can't necessarily
- 18:13expect the automation to start waiting things within your data,
- 18:19and that's just an interesting subway.
- 18:22Yeah, I'll give you an example of that.
- 18:24I pointed this out on the John Lewis site several years ago,
- 18:27and thankfully it's been fixed and I'm not going to be arrogant
- 18:29enough to think it's because of me pointing it out publicly.
- 18:33I'm sure that was just part of their ongoing optimization
- 18:36improvement because they've got a big smart team there.
- 18:39But they had, I was searching when when I when smartphones
- 18:44first went to dual camera mode, the front and back cameras, you
- 18:47saw a spike in search with your camera phone which is you know
- 18:51it's not a standardized, no iPhone or Samsung whatever.
- 18:54IB is going to have dual camera phone in the product title
- 18:57because it focuses on the brand, the model etcetera that's
- 19:00attribute data. But because the search was in
- 19:03deck was focused on title and description the the 1st 2
- 19:07results were dual cookers which which is bonkers.
- 19:11But yeah that the search engine couldn't detect that it was a
- 19:13phone, it was the primary driver and that dual camera was the
- 19:16attribute that was then next needed.
- 19:17So thankfully we're getting more intelligent and unsatisfaction.
- 19:21But if you don't structure your data correctly and you relied on
- 19:25your product title and you're right the product scripture is
- 19:28for me. I always say I I personally
- 19:32don't like indexing product descriptions unless there is a
- 19:35specific technical reason too because the product description
- 19:38is so technically specific to perhaps that it will add value
- 19:42to it. But I think one of the things we
- 19:47talked about earlier which is why do we get some of these
- 19:49issues and the politics and and the one thing that I have
- 19:51learned that can work or get them by in is is you model on
- 19:55some of the data, right? Especially with search journeys,
- 19:58they're easier to put numbers on of.
- 20:00How many search queries run in dead ends?
- 20:02What's our conversion rate? If we get better results, what
- 20:04could we potentially get as a as a conversion?
- 20:06Sales admit Find that taking that data back to the people
- 20:10involved in creating and providing source data, whilst
- 20:13not perfect, can open people's eyes and get more positive
- 20:16response and just say we need better data.
- 20:18But actually giving them numbers, if you give, if you
- 20:21especially get buying teams numbers of you could sell this
- 20:23many more units if you give us better data.
- 20:28Yeah, absolutely. And I think that long
- 20:29description points out, if you do get the product data right,
- 20:33you then don't need to index the long description.
- 20:36So yeah, I think that we could probably talk about that all
- 20:38day. So next question.
- 20:40And I think like we've talked about this briefly or not
- 20:43briefly, even like we talked about, it's on a few episodes.
- 20:45It's like when you have problems like this, like realistically
- 20:50the best time to fix it is probably going to be during like
- 20:53a migration or like a big change project.
- 20:56How do you think teams should approach this when they're very
- 21:00platforming or looking at a migration of some kind?
- 21:04So #1, it's determined by the platform because each platform
- 21:07has different data structures. I know fundamentally it's it's
- 21:11the same premise, right? We need product info, but it's
- 21:14going to be determined by what template you're using for your
- 21:17prep pages, what product data fields you've got and how you
- 21:21then use them to achieve what you want from a product data
- 21:25point of view. And I think there's an important
- 21:27consideration. So how much are you doing
- 21:29through like Pro attribute data? How much are you doing through
- 21:31tags that are visible or invisible to the customer?
- 21:36But the starting point to get there is the research bit.
- 21:40If you don't know what you need and what the problem is and what
- 21:43customers need, you can't possibly architect the site,
- 21:46right? So we've got to lead.
- 21:48I'm kind of answering the next question that we had to give you
- 21:51this answer which I know you're building because you've got
- 21:53loads of experience doing this and you've got I think you've
- 21:56got more experience doing this directly within the platform to
- 21:58help people the product data. But so I I remember working with
- 22:01Lumia Equestrian brand and one of the the precursors to any of
- 22:07this I did with them was Aux User Journey workshop on how do
- 22:11customers find products, what are their journeys, what are
- 22:14they looking for, what they searching for, what's in their
- 22:17mind, what questions they asking.
- 22:19Because the navigation wasn't. I knew the navigation wasn't
- 22:22wasn't aligned with what customers need, but I didn't
- 22:25understand the audience in depth enough to know exactly what what
- 22:29words, what language what were they looking for in the product.
- 22:32So we went through that process and we identified the fact that
- 22:34people shop not just by products or category, but people shop by
- 22:37event. So whether they are show jumpers
- 22:40or eventers exactly or they're just people gathered out on the
- 22:43weekend for a nice ride and then there are the people who are
- 22:47searching for gifts and that there are different different
- 22:49ways of finding and different needs in the product data.
- 22:52They've, you know people who are expert show jumpers want the
- 22:55product, the technical product information.
- 22:57They want to know that it's right for that also.
- 22:59Well that's right for them. And there are others who want to
- 23:03OK well call products for their granddaughter who know they know
- 23:07their granddaughter loves pink or blue or whatever the colour
- 23:10might be and they want to get therefore they're looking by
- 23:12colour. So it's totally different.
- 23:14So we we did that piece and that ideation LED into the next bit
- 23:18which was then the research around what people searching for
- 23:21externally like Google etcetera, what sort of keywords or
- 23:24questions and answers stuff then what are they searching for the
- 23:27site all of that intelligence which we then boiled down to
- 23:31right. OK, well where's the current
- 23:33date and what we missing what what actually what filters are
- 23:36we going to need to do this? What category navigation are we
- 23:39going to need? You know how do we pin the
- 23:42collections, what's the interdependencies, what's a
- 23:45fixed collection, what's dynamic And if it's going to be dynamic,
- 23:48what is the attributes and tags we need off the back of that to
- 23:50drive those dynamic categories and then map it.
- 23:54So that that's normally how I do it and that then if you do that
- 23:57piece right and structure your product data that and then
- 24:00inevitably leads on to much larger things like search
- 24:03accuracy. So that's that's more but how
- 24:06about you, how do you how do you tactics, I know you've got a
- 24:09pretty good methodology for this as well.
- 24:12I think it's pretty similar to what you just said, the same
- 24:15really. So I think we would usually like
- 24:18map out what you need the product data for.
- 24:22So a big part of that would be like yeah like you said category
- 24:24and collection inclusion filtering data feeds.
- 24:29The other one is things like internal exclusions and stuff.
- 24:33So if you want to exclude a product from search or you've
- 24:37got like a subscription product or a pre-order product or
- 24:39whatever it is like stuff like that.
- 24:41We'd build out all the potential like potential things you need
- 24:47the data for. And then we would start to look
- 24:51at like the optimal category structure or collection
- 24:54structure, which might not just be a short term thing, it might
- 24:57be like long term where they want to get to and the same for
- 25:00filtering. And then we would start to build
- 25:03all of the attributes. And then in an ideal world, so
- 25:07similar to what you just said, like I think you're best
- 25:09rebuilding A catalogue to fit in with the new structure.
- 25:13So we would then essentially get a spreadsheet and get a meeting
- 25:16in with as many people as we can that are relevant and start to
- 25:19kind of divide and conquer. And some, some data points might
- 25:24be just, you know, mapped straight over, but some might
- 25:29need to be created and that's where you might need to get like
- 25:31merchandise involved or whatever else.
- 25:34But yeah, that's my preference. Then import one by one.
- 25:38Yeah, I think that's a good process.
- 25:43Yeah, definitely. It's it's amazing actually when
- 25:46when you get sick people down and and stop because the problem
- 25:49with product data normally is that people are so busy
- 25:51day-to-day with we got to execute campaign we've got to
- 25:54get promotions set up we've got to add new products that you
- 25:57just carry on with your BAU process even if that process
- 26:00isn't necessarily like properly robust and fit for purpose.
- 26:03It's amazing if you can get people out of that day-to-day
- 26:05thinking and give them a mental break and then just challenge
- 26:08and support and and let them actually acknowledge that people
- 26:12have in a business about how you can make this better is really
- 26:15is is sharp little business having sometimes they eat hell
- 26:18and often time the the knowledge and abilities there.
- 26:21It's just people haven't had the time to think so.
- 26:24But I think my recommendation on this around the migration is
- 26:27you've got to ensure that in your discovery process that you
- 26:32have dedicated workshops around this and that the business takes
- 26:36time to prepare for that workshop with an agency or an A
- 26:39development partner to tackle some of these questions about
- 26:43what product data is needed and why and and what the current
- 26:46issues and and do some brainstorming to help make that
- 26:49session more productive. Then at the end of that session
- 26:52with the agency they will typically push back and go right
- 26:54we've got gaps on XY and Z. Now let's let's fix this as a
- 26:58piece of homework. So we've talked about, we mean
- 27:02we've talked about the tools really that teams can use to
- 27:05find product data gaps. I don't know if you want to call
- 27:09out any specific. I know there's there's loads of
- 27:11keyword research tools out there, pick your SEO tool of
- 27:14choice, pick Google external keyword, you've got various Q&A
- 27:18ones, there's loads of different reviews, engines of people to
- 27:23put out, customer insights etcetera.
- 27:25So I don't think, I don't is it is it worth pulling out any
- 27:28specific tools? I feel like it's the tools
- 27:30people use are going to be very, very dependent on on their
- 27:34business. Yeah, I don't really think there
- 27:37are any. I mean most of what we've done
- 27:39because I think the problem you've got with this is often in
- 27:42the average business probably doesn't know they have an issue.
- 27:46So we just recently did an audit for a computer company and there
- 27:49was a load of missing data, but we only found it by spending
- 27:53like loads of time or analyzing filtering and things and maybe
- 27:56like you know on some like or just products missing.
- 28:00So I think realistically you've just got to get the data into a
- 28:03spreadsheet and to start to like look to identify gap and it's
- 28:08like that's the kind of task that will just never come
- 28:10around. So that's why a lot of people
- 28:13share the PIM implementation or RE platform project.
- 28:18But yeah, it's the only one thing I'd say is like obviously
- 28:22AI is getting to a really good point to start to try and help
- 28:25with this. So Shopify built that into their
- 28:28core. So like populating additional
- 28:31black metal fields and thing and I think over time you'll be able
- 28:34to do a lot more, you know, via Google Sheets or Excel or
- 28:37whatever, and it'll be more kind of validation and less manual
- 28:42worth. Yeah, you drop the AI Klaxon.
- 28:45But I I agree, Jill, actually a tool that is really interesting
- 28:49for anyone listening, if you haven't used it for was
- 28:50Treejack. Because what you can do,
- 28:52especially around navigation and the field to try and look at
- 28:56navigation, whether you can get people quicker into product
- 29:00journeys, you can set up different navigation miles and
- 29:03you can invite people to come on the site and complete tasks.
- 29:06So it's not like you've developed your site.
- 29:08You do it in their tool and you can invite people in from your
- 29:11customer base. It's a complete task site.
- 29:13We want you to find X and it does time mapping of task
- 29:16completion, how many people get to complete it and how many P
- 29:19and how quickly. It's a really good way of
- 29:20testing you navigation menu structures that come out the
- 29:23back of like product data analysis pieces.
- 29:26I do love that one. Then it's like next we've got
- 29:29two last questions, how can we add richer data to product
- 29:32records? So I know you do a lot of work
- 29:36around custom meta fields and tags, so I'll leave that to you.
- 29:38I'm going to talk about the attribute data which is similar
- 29:40but you know custom meta fields and tags.
- 29:43People in Shopify world be used to attributes where EA people
- 29:47are used to work in and I also like custom fields in in the
- 29:50lights of like and a daily or big colours.
- 29:53But the most important thing for me around richer data is you've
- 29:56got to start with structured data, name value pairs.
- 29:59Like what is the attribute colour, size, dietary
- 30:03requirement for example and what are the name, value pairs that
- 30:06go with it. Because the most important thing
- 30:08with attribute data is standardization and
- 30:10normalization. You don't want for example open
- 30:14ones where anyone can put in a value without creating a a
- 30:18uploading new product because you get all sorts of divergent
- 30:21values so you get lowercase, uppercase.
- 30:23And yes, you can normalize through automation, say always
- 30:26capitalize the 1st letter. But you might have different
- 30:29spellings or or or different colour names and conventions
- 30:33that a system does not natively know or understand are the same
- 30:36thing. So you've got to be very, very
- 30:38clear on making sure there's a there's a consistent list to
- 30:42choose from and that really helps drive all consistent
- 30:45filtering. But also that you're using your
- 30:47customer feedback loops to identify where there might be
- 30:49gaps. So a good example I'll give you
- 30:51is I worked with a wine company where they there was a lot of
- 30:56demand for vegan wines. It's good for years back but
- 30:58there was no actually data in the product catalogue for vegan.
- 31:01So if somebody was searching for vegan wine, it would only find
- 31:04wines if the word vegan was used at any point in the
- 31:07descriptions, which it wasn't always.
- 31:09And simply adding vegan as an attribute instantly makes those
- 31:13products got those products straight into the top of
- 31:15searches and therefore selling more effectively.
- 31:18So actually data is always one of my starting points, but Paul
- 31:21do you want to talk through the the like custom meta fields and
- 31:24times and what valuable impact you see and how people do it?
- 31:29Yeah. So I think there's a lot to talk
- 31:31about here as much as before this episode we were talking
- 31:34about trying to make things a little bit less Shopify, the
- 31:37kids taking over the wild. But the so I think Shopify
- 31:41catalogue has changed loads. So you were just talking about
- 31:45free, free text fields and like the risks of that.
- 31:49And a lot of what I was referring to as well earlier, a
- 31:52lot of those issues were actually probably caused by tags
- 31:55because the way that it used to work is like an ad In a lot of
- 31:58cases someone would go into the admin and just, you know type
- 32:01words and that would drive inclusion and filtering,
- 32:05inclusion and pension and that would cause a lot of the issues.
- 32:07And I again like I kind of miss the old days of shop live
- 32:10because you know, we would come in and we'd build like a really
- 32:12nice little she and we would start to maybe pull certain
- 32:16things out of shop buy and you know build like standardized
- 32:19options and excel and stuff like that.
- 32:21But now they have improved a lot of this.
- 32:24So 2022 Shopify moved to like supporting Metafield properly
- 32:30and they became like a first party citizen.
- 32:32And then that developed a lot last year where they're kind of
- 32:36within the UI and everything. So a lot of stuff is now moving
- 32:40to meta fields and their performance benefits and
- 32:43everything else and they essentially work is like a
- 32:45standard attribute in another platform.
- 32:48And then last year they introduced meta objects, which
- 32:52is essentially allows for dynamic content and that could
- 32:56be like a cluster of meta fields that build a piece of content
- 32:59that you could use across different collections or PDPS
- 33:02etcetera. So that's a really nice addition
- 33:05that actually takes Shopify above a lot of platforms in some
- 33:08ways. But essentially with Shopify
- 33:12now, if you're starting from scratch, you probably wouldn't
- 33:15use too many tags like Shopify support meta fields filtering.
- 33:19Now, can't remember whether they do for smart collections or not,
- 33:24but all of the data feed providers, all of the third
- 33:26parties are indexing meta field. So yeah, they have farmer and
- 33:31the other thing is off the back of addition, save them moving
- 33:34towards essentially attribute sets where you'll be able to pre
- 33:37build sets of meta fields etcetera.
- 33:40And there's other improvements to product data around variants
- 33:44etcetera. So it's yeah that Shopify
- 33:46definitely getting a lot more in line with like a an agrarian or
- 33:50a sales force on that, yeah. Yeah.
- 33:52And that's so important because I speak to a lot of businesses
- 33:56and it's not, it's not system limitations obviously people
- 33:59often this people not understanding how they can do
- 34:01things in a more automated way where people are managing manual
- 34:04categories for basic things like you know new in or best seller
- 34:09whatever it might be and and and things where actually a lot of
- 34:12it could be or orchestrated through the system and you and
- 34:17some of it can be done through smarter tagging.
- 34:19So there's such an opportunity with a lot of businesses to just
- 34:22take away manual process that isn't needed and focus on more
- 34:25value add merchants as in. Yeah.
- 34:29And also a really nice case. Sorry, I've, I've spoken over
- 34:34you. I think there's a slight delay
- 34:35with Riverside actually, but I was just going to say it's also
- 34:39a really cool use case for Flow now that they can now you can do
- 34:42time based rules for things like new N and I've seen some brands
- 34:46do like new in this week, new in this month, etcetera used
- 34:51previously tags with roads for that.
- 34:57Yeah, Yeah, exactly. And also the ability to do it
- 35:00based on like dynamic around sale as well.
- 35:04So specific sale categories you could, you could have like you
- 35:08could define a a like hot deals. I'm working with a kind of
- 35:11listen about this premise of hot deals which is products of over
- 35:1550% discount. So that is a separate dynamic
- 35:18category to your main sale category and that is
- 35:20merchandising promoted differently.
- 35:24Let's finish up with the operational side because I know
- 35:27part of this conversation that we're having today came out of a
- 35:30discussion we were having with with with Luke, Luke Hodgson who
- 35:34who some of you listening will know well.
- 35:38And we've talked about our product piece and and tell like
- 35:41e-commerce teams need to take it more seriously.
- 35:44And often the operational side is where people don't think
- 35:46about where better product data or tagging can drive
- 35:50improvement. So let's go through that.
- 35:53A couple of examples I've seen and I guess this is dependent on
- 35:56what tools you're using, how strong order management
- 36:00capabilities are to automate this based on information that
- 36:03are all the file versus information that's missing.
- 36:06So good example is routing SKU specific carriers based on a
- 36:10particular tag in that SKU. Seeing this done where people
- 36:14have products that can't be shipped to certain destinations
- 36:16or they have to be shipped by specific Courier.
- 36:19Because a good example be two person delivery where only one
- 36:24Courier is able to do that and therefore the routing's got to
- 36:26happen based on the flag in the order for that item versus other
- 36:30items that can go out on standardized couriers.
- 36:34And even a coin of mine is now doing it for express shipping
- 36:38when when an express shipping flag is in it.
- 36:41I know this is separate the product days but related that
- 36:44that is a recognized as a specific tag by the WMS and it
- 36:48changes where those orders are pushed for pick and pack only
- 36:51view to different teams. So there's I think there's
- 36:54really, really small opportunities like order
- 36:56management. I don't know what what else
- 37:00you're thinking in terms of operational areas where better
- 37:02product data can help us. Yeah, I think, I don't think
- 37:07this is a something that a lot of our clients do or that is
- 37:12super simple with many third parties.
- 37:14But I think like our average client is trying to do a lot
- 37:17more around customer data at the moment and trying to pass more
- 37:21behavioural data into their CRM and you know try to move towards
- 37:24accounts and you know collect more first party data as part of
- 37:27that. And I think if you are tying
- 37:30like product and category affinity to a customer, you can
- 37:35then you could feasibly then use the attribute data to like build
- 37:40out better relationships with other products and things like
- 37:44that. But that being said, like I
- 37:46said, I don't have many people that sophisticated around that
- 37:48and I also think that a lot of that will be done by AI.
- 37:52Sorry to mention that again over time.
- 37:55But yeah, beyond that I mean I'm, I don't, I can't think too
- 37:58many examples really. I mean obviously you've got all
- 38:02the use cases if you get this right anytime and you don't have
- 38:06APM where you don't have that in your hours.
- 38:08Obviously the data could support other channels as well such as
- 38:11retail or catalogue or whatever that was.
- 38:13But yes, can't pick you too much beyond that.
- 38:19Yeah, it's I find it interesting that people that in e-commerce
- 38:22people don't sit down with the OPS people to talk about how
- 38:26this could actually benefit operational processes.
- 38:29And I I know it's not top priority for for most e-commerce
- 38:32teams when thinking about product catalogue optimization
- 38:35because they're thinking about front end trading, marketing
- 38:37channels etcetera. But if you can get operational
- 38:40efficiency as well by better Tate and and make the OPS people
- 38:44know about the data structuring and tagging that you are able to
- 38:48do to avoid is if you can, if you can append certain assets
- 38:54attributes to particular products that can be passed in
- 38:56order file to avoid somebody having to do any custom flows in
- 39:00back office systems where typically it's more complicated,
- 39:02that's a massive massive win. I know somebody did it around
- 39:05fragile goods because they they had a manual warehouse, it
- 39:09wasn't fully automated using like a modern WMS and they
- 39:13basically manually printed off pic sheets and then allocated.
- 39:17So they then had to go through each order to work out what had
- 39:19a fragile good because I had to go to a specific packing
- 39:21station. Well I actually the ability to
- 39:25it to have A tag flagged on the order so they lit through a
- 39:29sheet or they can print off only ones with that particular tag
- 39:33just made life so much easier. It was just quicker to go right
- 39:36here go those little orders over to the fragile station.
- 39:39So there's sometimes even little tweaks can can have a benefit to
- 39:42your operational staff. That feels like we've covered
- 39:48quite a lot. Is there anything obvious you
- 39:50think we're missing today? I don't think so.
- 39:55I think the only thing is we could probably do a follow up,
- 39:58maybe not quite full episodes but ordering customer as well
- 40:01like similar principles around like how you can kind of enrich
- 40:05data to maybe benefit other areas.
- 40:07But I think we covered loads of personally.
- 40:11Yeah, so we're going to stop rambling now.
- 40:13We hope you found that useful. Do do hit us up on LinkedIn and
- 40:17let us know what you know what. What are your tips and what have
- 40:20you found worked well for for improving product data?
- 40:24How are you using product data smartly to automate things
- 40:27either on the on the website or through back end OPS?
- 40:30We always like to hear really cool examples of people doing
- 40:33things smartly with data and do keep listening because next week
- 40:37we'll drop another episode. And if it was your first time
- 40:39listening, don't forget to give us that rating for you guys.
- 40:42Thanks very much everyone. For more information on this
- 40:52topic, head over to Replatform dot FM for our audio podcasts to
- 40:56discuss a project, or if you'd like to chat about any of the
- 40:58topics covered in this episode in more detail, please reach out
- 41:01to myself, James Gurd, or my Co host Paul Rogers via LinkedIn
- 41:04and Twitter. Thanks again for listening and
- 41:06keep your ears peeled for the next episode.