Transformation through Automation: Keith Mazanec, Brad's Deals

Varia: Hello. Welcome to People Changing Enterprises. This week, previous guest Keith Mazanec details how Brad's Deals reimagined their content operations using automation.
He tells the story of how content led the way to their digital transformation and how they automated key processes with Contentstack’s Automation Hub to scale, streamline and make life way easier for content teams. This was a keynote session recorded at our annual conference ContentCon. Enjoy.

Keith: I'm Keith Mazanec. I'm the director of Software Engineering at Brad's Deals. And this is our story [00:01:00] about how we transformed our content management experience with Contentstack. So to begin, let me first talk about what Brad's Deals is as a business.
We're an online shopping platform. We have a team of editors that go out into the market every day and find the best deals online. They're scouring different websites and newsletters and all sorts of different places for information to find the best pricing on products to share with our readers.
And we call ourselves a fully curated online shopping discovery platform because that's really what it is. It's all about the human touch and the editors that are actually finding great deals online. We're not trying to programmatically you know, using all this stuff to to fill in the hopper.
We just wanna find what's actually a great deal. And then we publish that content out to our millions of readers and email subscribers through all sorts of different channels, web, mobile native mobile email, etc. But again, every deal is vetted by hand. And because we're covering the consumer marketplace things are changing rapidly.
And so we're publishing hundreds of pieces of content every [00:02:00] day and need that to be smooth. The beginning of our story of our transformation at least begins during the pandemic as we saw during the pandemic that really scrambled the retail landscape, right?
There were runs on toilet paper and cleaning products, and it was really challenging to keep up with the, what was actually available. And consumers were coming to us, saying, where can I find these things? And then as people were, sitting at home the things they were buying were changing rapidly, right?
We saw the growth of home goods and furniture and those kinds of products really taking off. And us as a business, we spent a lot of 2020 and 2021 experimenting, trying new things. How can we offer content to people, that's relevant to what they're experiencing right now?
And every step of the way we kept running into the constraints of our legacy architecture. We're a 20-year-old business with a 10-year-old tech stack. And it was inflexible siloed and very difficult to make change. And uh, we really knew that something had to change.
And so for [00:03:00] us, that really meant beginning with the end in mind, where did we want to go? And I've heard a lot of other stories on the stage today about beginning with the end in mind, where do you want to be? What is the problem you're trying to solve for your market, for your business?
And for us it was the content model. That's really where things needed to start. And so we looked at how are we generating content, where is it coming from? It's really about our editorial teams and partners sourcing products and discounts and finding them out in the market, cataloging them into our system, and then sharing with our readers why this is a great deal and how you can save money, on, on X, Y, or Z, in your life.
And then the most important thing is then distributing that out to people through all of the different channels that we have. And so this content loop we identified was really the critical part of our business. And we needed to have a stack and a content model that could grow and evolve and change at the pace of the retail landscape, at the pace of marketing.[00:04:00]
And we started all of this months before we even started talking about the technology. And I think that's critical because the technology ultimately, is very powerful. But it's only as useful as what you do with it.
And so for us that's really where we began. We had an inflexible model. Change was very difficult. It involved a lot of developer work. We’re talking about a CMS migration here, but we actually have several CMSs, plural.
We had some homegrown CMSs in one's kind of programming language. We had others in a different language. We had some, traditional open-source projects and it was all siloed and fragmented. And because we're publishing about the retail landscape that means, we need to bring in the information about those services and products into our system.
And that meant a lot of rote manual data entry minimal reusability of content, no publishing workflow. It was time-consuming and painful for our teams. And so anytime we wanted to make a change we're essentially getting out the duct tape and gluing things [00:05:00] together in the most difficult ways.
So we set out then to make a change. And for us, that meant we needed one unified system for all of our content types. We did the work to define what that model needed to be, and we knew that it all needed to live within one system. Second, it was very important that our editorial teams and the content managers and the marketers were having a good experience using that content and creating it.
And then most importantly at the end, we needed to make sure we had something that was a stable foundation to build on top of for the future. The retail landscape is constantly changing and we need to be able to keep up with it. And for us that was really the sort of vision we set out with when we were evaluating what we needed to do.
And so then we started talking about the technology. And for us that meant, revisiting, do we need, are any of these CMSs that we already have viable? Can we, combine them together, rebuild our own and have a new solid foundation of our own? Or should we go out into the market [00:06:00] and find some solution that provides all of these things for us?
And ultimately our conclusion was we are not a CMS company, right? We are a publisher. We're helping people, helping consumers save money. And so that is the core of our DNA. We're not a tech company per se, right? We use technology to help tell people how they can save money and live a more beneficial life.
So that was our decision. We went ahead and joined up with Contentstack. We had a great experience onboarding and were able to produce our content model in their system. And that was all great, right? We were able to bring the content model into the system but we knew that wasn't the end.
That was just the beginning. And so next for us, we wanted to talk about how can we take this new model, this new system, and automate as many of the parts of it. As possible to reduce human error, to reduce the amount of rote work that people had to do and really to allow our editorial teams to share great deals with our readers faster.
And so I'm gonna talk now about a couple of the different things that we actually built using uh, the [00:07:00] Contentstack platform as our foundation. And so first again, when people are publishing onto our platform a lot of that really involves finding products for sale on a retailer's website.
And that involves a lot of manual entry of images, pricing information, shipping details generating tracking links and all those kinds of things. And we build a tool that essentially allowed our editorial teams to drop in a product URL from some, retailer they're working with.
And then have an automation system that goes and checks against a bunch of different data sources that we have access to, to find that information and bring it in automatically without that person having to do that manually. So our editorial teams can focus on what they're great at which is finding value for consumers.
And so that's been a great help. All right, number two. So when people are actually, bringing content into the system, publishing about products one of the important things that we need to have in there is the ability to categorize those things accurately, right?
Having a good categorization of all [00:08:00] your products is critical to doing things like recommendations segmentation and targeting, and it has to be accurate. And in our model, we have hundreds of different merchandise types as we call them, that certain content can fall in, and having our editorial teams have to select those manually was a, again a cause for error and yet another step in their process that we knew we could automate.
And so we took our corpus of data. We've been publishing for 20 years, and we're able to train a model. That could predict the merchandise type or category of the content we are generating and using Contentstack’s SDK, we were able to build that directly into the publishing interface with a custom component.
And it looks native. It doesn't look like something that, our team had to hack together. It's using their components and again, leveraging that platform effect to bring what's special about our business, to the platform. And so then, day in and day out, our editorial teams are spending their days in this tool.
That's where they, where they spend a lot of their time when they're not actually out in the market shopping and [00:09:00] finding things. And so we needed a central sort of starting point for them when they're in the tool. And so again, we were able to build a custom dashboard that shows them what they've been, what they've published what they're working on and, represent our own important information that we need to share with people. For example, we needed to have the images or the store name of where content's coming from. We want to know when that content was last updated, right? What did the price change, for example? And so we needed a home for that information to live within the CMS.
And again the beautiful thing about this is if you look at the actual UI here, it looks like native Contentstack functionality. And so that meant that when we were building this. Because we were building on top of the app SDK, we didn't need to reinvent the UI wheel, right? We had that toolkit to go to and pick from, and we were able to move very rapidly.
In developing all of these things that meant we were able to have a very tight feedback loop with our editorial teams and the product development folks that were working with them to get, feedback immediately and [00:10:00] have changes and new features delivered in days rather than in weeks or months.
Which in our previous tool, we would've just said, sorry, I can't do that. So that, that, that was amazing. So we have all these new, this new content model, and we have this new workflow and this new set of capabilities to automate many pieces of it. As we all know, these transformations are not something where you flip a switch one day and that's it, right?
It's just done. You need to have continuity and continue to run your business. And so for us, that meant and we needed a way to keep our content model in the new system, in sync with our old system while the teams worked on migrating over the app website, email publishing, and all the other core parts of our business.
And so again, using an automation, we were able to have that content automatically whenever it's man or updated in the new system. Push back into the old system obviously in a stripped-down form, right? One of the reasons we wanted to change our models because we wanted it to be richer and broader and larger.
[00:11:00] But being able to write back to our legacy system meant that we could transition incrementally and iteratively. And so for us that's been a huge source of, we wouldn't have been able to do this if we hadn't had a continuity between the old and the new. And lastly, I'm gonna talk about the future and going forward.
After we built all of these things and we have this continuity, the world doesn't stop changing while we're building, right? It doesn't, even if we spend a year or two years working on a project and working on a transformation migrating away from legacy technology the retail market is changing, right?
We see in, in, in the world today, inflation and other kinds of shortages and things like that, right? That's a whole different kind of, of dynamic, and we need to be able to move at the speed of retail. And so for us, a core component of the new system was the ability to change our content model over time, and we were able to do that with Contentstack using some of the automation capabilities there. And so now when we wanna make a [00:12:00] change to a content model, let's say to add a new field to handle pick up in-store, right? Which is a new thing in the pandemic, somewhat or you can buy online and have it brought to your car.
That wasn't something we'd be able to do in our old system. It would have to be a developer ticket, go in and make that change. And so now when a content editor or product manager has an idea for something they want to change, they can prototype that they go onto a separate branch within the content system and are able to modify that experiment, play around, try new ideas which is incredible, and then they can bring those back to our team and we can incorporate those into the core model through a controlled process of change.
And so we're able to use that model that the team prototyped, incorporate that into our own code base where we have it in inversion control and then use the SDK and the CLI, I believe to script all of that out so that change becomes predictable and change becomes something that's inversion control can be reviewed and reverted as needed.[00:13:00]
And so that was a, again, a huge unlock for us. And so now what's next, right? We've started this journey, but we're certainly not at the end, right? We're in the, still in the process of cutting over different pieces of our stack. And so what's next? For us, it's really about continuing to build on this foundation that we've laid and starting to build up from there.
So that means adding more data sources for automation new kinds of products and services that our editorial teams might want to publish about. It means bringing segmentation and targeting back into the tool itself so that when a marketing, a marketer wants to test different campaigns, they're not going into 17 different tools to do that.
We want to have that unified withn the content. Additionally, through some of the things that we've built our first experience actually with Contentstack, we built and launched within two or three months of signing on a new web stack. And that new sort of foundation we've continued to iterate on and ad capabilities too.
Such that now our marketing teams when they need to build a new page type [00:14:00] or they want to experiment with something, they have a whole toolbox, a whole set of Lego bricks going back to the beginning that they can then combine together in new ways and try out new things. And we don't have to then wait, weeks and weeks for the development cycle to make that happen.
And so it's really amazing to see what people can do when they have tools to do things. And then, lastly, we're looking at how we can continue to do things with automations and make even more parts of our life cycle event-driven. So one of the beautiful things about having your content all in one place is that you have an event stream of changes to that content over time.
And so we're very excited to start using that event stream for a whole bunch of different kinds of things, including personalization. As new content is published right in the in the consumer marketplace, things are not on sale for a month or even a week. Sometimes it's oftentimes, days or even hours in some cases.
And so it's very important for us to get that content in front of readers quickly and we can do that when we can [00:15:00] tap into the event life cycle in the system. Additionally, there's another big hairy Frankenstein legacy process, that's our email publishing and I'm very excited, to dig into that this year as well and bring even more automation to the fore. And remove the, I think two to three hours someone was saying on our team is how much someone spends every day putting together our daily deal newsletter that millions of folks receive. And so we want to reduce that down to near zero, right?
It shouldn't be something that requires a manual building every day. So yeah there's so much potential and we're super excited about all of that. And that's basically it. That is Brad's Deals. Thank you so much.