Expert Perspective on eCommerce Innovation: Kelly Goetsch

This month, YOTTAA had a chance to talk to Kelly Goetsch, Chief Product Officer at commercetools, 4x O’Reilly Author, and headless commerce pioneer. During this interview, Kelly expands on his professional experience, the innovation involved in today’s eCommerce technologies and MACH architectures, and advice to eCommerce leaders on what to do while moving through today’s uncharted territory.

Key Takeaways:

• Traffic, order volume, and order values are seeing substantial growth. This is the new “normal.” 
• eCommerce Innovation requires iteration. Going headless and adopting microservices allows for that iteration.
• Monolithic platforms cannot retrofit a microservices offering, and at the end of the day still have release and innovation limitations.
• During the current pandemic, physical retail is basically dead and brands have to focus on digital.
• Brands need a fast frontend, but also need a very fast backend. There is a very clear link between conversion rate and page speed.      
• People are notoriously difficult when it comes to changing their habits (like shopping online), but once they do, they change for good. COVID-19 has forced many to change their habits.
• To keep up with rising traffic, brands need an elastically scalable set of technologies.
• Brands should test everything to ensure performance is ready for the holiday.        

Pioneering with the Pros

Tell us about your professional background.

I started my career in retail with Footlocker as a summer job in college. My job was to convert Photoshop images into HTML wireframes — which just so happened to be the biggest ATG implementation at the time. ATG was the big commerce platform back in the 90s and 2000s, and it also happens that it was the last project for the most senior ATG architect at the time. For some reason, he really liked me and mentored me. I left that job after a year being a proficient developer with ATG. 

Then I moved to Chicago and finished up my undergraduate and graduate degrees while working for ATG. I really loved the commerce space. I was able to do so many fun projects and travel all over the world. My last three and a half years at ATG, I was responsible for all the Walmart implementations because Walmart, with the exception of Walmart.com, was entirely ATG based. 

Then my wife and I had a child — which was great — but it meant that I could not travel every week. I got a “real job” with Oracle, and ended up being a product manager in their new Cloud group. I worked on Java Cloud Service, Exalogic, and Java EE. In my last year there, I was responsible for microservices. By that point, I had written eCommerce in the Cloud for O’Reilly, so I knew I really wanted to get back into microservices and commerce — I love that space. I Googled “commerce cloud and microservices” and commercetools came up as eight of the top ten search results. And I thought “who the hell are these guys? They sound pretty interesting.” 

I went to LinkedIn, and it turned out that the head of the US group and I had 125 connections in common. I dropped him a note and said that I would love to catch up sometime. He said to talk to the CEO when he is in New York next. I did, and we had coffee, which turned into dinner, which turned into drinks and more drinks and I have been with the company now for four years. We have really been pioneering headless commerce, which is exciting.                       

What about eCommerce has kept you in the industry for so long?

We [commercetools] are building an entirely new approach to commerce. Those first generation platforms were built in the 90s — Websphere Commerce, ATG, Intershop — and obviously the world has changed a lot since then. Now, there are a lot more touchpoints, we are a mobile-first world, and we have the whole rise of Cloud and SaaS. 

Being able to get together with these first generation pioneers and doing it the right way the second time, but with all new technology, is very exciting. What is there not to enjoy about that? The Magic Quadrant is coming out and you will soon see where we placed. The market really recognizes what we’re doing now and it’s great to see that reflected in reports and elsewhere. 

eCommerce Innovation Requires Maturity

Here at YOTTAA, we advocate the importance of having a fast browser experience. Can you talk more about the importance of also having a fast eCommerce platform?

The total time that the shopper has to sit and wait is a combination of the frontend and backend performance. So brands need a fast frontend, which is what you (YOTTAA) accelerate, but they also need a very fast backend. There is a very clear link between conversion rate and page speed. I read a study and it was something like 5% conversion for every 100 milliseconds. Doubleclick by Google found that 53% of mobile sites visited were abandoned if the page took longer than 3 seconds to load. BBC found that they lost 10% of users for every additional second their site took to load. Pinterest increased search engine traffic sign outs by 15% when they reduced their perceived wait times by 40. So, it’s a well established fact out there that conversion rates, engagement, and all sorts of metrics are strongly linked to the speed of how long it takes those things to load.         

Do you know what’s impacting your site speed? Get a free Site Performance Snapshot Report to find out!

How can personalization improve the shopping experience?

Shoppers want a personalized experience so that they can find the right product. If you go back to the fundamentals, retail is about putting the right product in the right hands at the right time. Personalization makes that happen. So if brands are able to do one-to-one based personalization, or even segment-based personalization, they can give their customers the same feeling that would come from going into a store, where people know who they are, and where they have established a relationship with a sales person. Ultimately, it’s important to get that product mixture correct, and the way to do that is through personalization.       

How does a headless architecture enable brands to be more nimble and flexible?

Traditionally, commerce platforms had the head on them because at the time, there weren’t APIs, which made it very difficult to have a frontend decoupled from the backend. Also, the commerce platforms predated the content management systems (CMS). As a result, the commerce platforms, the pioneering ones, all had to build heads on their platform and include it as part of their platform — which in 95’ was pretty revolutionary. However, today we have a lot of new technologies out there. We have REST for example, we have GraphQL, we have really good standalone frontend technologies, we have progessive web apps (PWAs), we have all sorts of eCommerce innovations available. 

When you look at the monolithic eCommerce platforms that have their frontend tied to the backend, you have now tied the two together in a release cycle, which is devastating from a release cadence standpoint. The challenge is brands now have to release the two at the same time, which is difficult and slows down eCommerce innovation. If even a pixel is moved on the frontend, the entire backend now has to be retested, and vice versa. When a brand uses headless architecture, those two are decoupled, so the backend is working in parallel to the frontend. Now, brands are able to very quickly and easily innovate and two teams (or more) can move at different speeds. 

In short, monolithic platforms result in a much slower eCommerce innovation rate than using a modern headless architecture.                 

Do you think there is a way brands can architect microservices that won’t negatively impact their site speed?

They could use GraphQL, which is the topic of my last book. Facebook came out with it in 2015 because they were rewriting their native app. They basically had a webshell-like version of the Facebook app that can use many different clients, and brands can use to build native apps for. 

It can be difficult to call many different APIs to retrieve data to render a page or experience, and too many of these calls can also negatively impact page speed. GraphQL ended up building a query language for their APIs, so it sits on top of the APIs. It allows you to formulate a query from the frontend, almost like a SQL query, where you can say “for this product, or for this set of products, I want you to retrieve the name, description, price, inventory” — you can retrieve 500 different attributes if you would like. The GraphQL layer will call the respective APIs, get the response back, and pass to the client only the data that was actually requested. This results in making a single GraphQL query to your server instead of making so many different REST API calls. Now, all of the different API calls are happening within walls of the data center, which prevents over fetching, underfetching, and having to authenticate with the different APIs. It is one nice clean layer. 

Twitter has used this since 2016 and New York Times has used it. Shopify has actually announced that they are no longer API first, they are now GraphQL first which is cool. We are really proud of our support for GraphQL and it’s something most of our customers are doing now.         

What are some precautions brands should take when moving to a headless architecture?

I think the biggest precaution would be not having the maturity to deal with the technical complexity of using a headless architecture. There are some organizations that want a website in a box, which is fine. Think of Magento, Shopify, and Demandware. They are good platforms, but they are different from what commecetools offers. They are more for the brand that focuses on “I am a fashion retailer and IT/eCommerce tech management is not my core competency. Give me a website that allows me to sell shoes online.” I am simplifying dramatically, of course, but that is the value proposition.

Commercetools is on the other side of that spectrum. We are tools that IT and eCommerce teams can use. Think of us like lego blocks. It’s a different style of commerce, and to adopt our style with headless, brands do need some maturity. Brands’ tech teams need to be able to build a frontend on their own, must be able to call APIs, and call GraphQL if they want to have a faster/seamless approach. We are not a paint-by-numbers, everything-in-a-box platform. We are a platform for the more advanced, the more mature.      

How do microservices help brands create great experiences across all digital touchpoints, like websites, mobile apps, cars, wearables, etc.?

Microservices is a small granular piece of functionality that is built to manage one team, and it is exposed through an API. So microservices themselves don’t do much in terms of impacting the experience directly. Microservices allow lots of teams to innovate in parallel. You can now have 10, 20, or 50 different teams each iterating in parallel a new functionality which is exposed through a new API. These teams can be working on the website, mobile apps, cars, wearables and much more.

Being API based, being headless, you can incorporate that into any experience that you want. You can then build a progessive web app (PWA) or website — whatever you want to build. Ultimately, it’s not the microservices necessarily, it’s the APIs backed by the microservices that allow you to create that experience.   

Headless Enables Iteration and eCommerce Innovation

Tell us about commercetools and the ‘headless’ approach you took from the beginning.

Dirk Hoerig, CEO and Founder, founded commercetools back in 2005 as a hybris SI. We were actually the first hybris SI, and that was a successful business, but he saw the rise of cloud in 2008/2009, rise in mobile, social media, etc., and in 2011 he stopped the consulting side of the business, raised some money, and built the platform we have today from scratch. He went to market with the platform in 2013. He saw first-hand how difficult it was to get hybris to deliver a mobile first experience; it didn’t work. 

Dirk invented and was first to market with headless commerce in 2013, which really allows frontend teams and backend teams to all innovate in parallel. I am a firm believer that eCommerce innovation requires iteration. Going headless really allows for that iteration.            

What are some typical approaches and technologies brands take to build their frontend when using commercetools?  

It’s all over the place. We have a good chunk of our customers, probably 50%, that build a custom frontend from scratch. It’s React, it’s Angular, it’s pick whatever “hipster” framework is on Hacker News this week. But they will build something from scratch. 

Another chunk uses a Digital Experience Platform (DXP) of some sort. Bloomreach is popular. Adobe Experience Manager is still surprisingly popular, as is Acquia — there are a number of these DXP type platforms out there. We are also starting to see the rise of frontend-as-a-service platforms. So think of Frantastic, Viewstore Front, Mobify, and that kind of thing. We see those rising very very quickly.           

Why do you think it’s important for brands to use ‘native microservices’ eCommerce platforms vs monolithic platforms that are attempting to retrofit a microservices offering?

First, legacy platforms cannot retrofit their platform to support microservices. If you built it that way, it’s that way forever. It’s a problem with business model software, and we at commercetools will face the same issue 15 or 20 years from now as well. 

The challenge occurs as you become a successful company and you have 500, 1,000, 2,000 customers buy your product. They build their apps around your product and they expect your product to work in a very specific way. If you as a vendor go make substantial changes to that, it forces them to do reimplementation. If a customer is forced to do reimplementation, then they look around in the market and try to find someone who does what you are trying to make the product do, but better. 

In our case, the legacy vendors are trying to modernize with APIs, but then they look around and say “who invented API based headless commerce? Well commercetools does, so we should go take a look at them instead.” Someday 15 years from now, there will be some crazy new thing out there that we don’t do, that we can’t do, and we are going to go through the same thing. It’s the lifecycle of software. They may attempt to retrofit but they can’t.     

With native microservices, you get the very very unique ability to iterate in parallel and release in parallel. We at commercetools have seven vertical feature teams, each with a bunch of microservices underneath them. And any one of those seven teams can release independently to production. We have lots of releases to production everyday, which is great because our customers then get that functionality immediately; they do not have to apply anything. 

We are just updating the APIs overtime in a backwards compatible manner. The only way we can do that is because we have microservices in the backend, and that is why retailers and brands need to adopt some form, and I am not saying textbook microservices, of smaller pieces. eCommerce platforms can be a monolith, and they can have that monolith expose APIs, but in the end you still have a monolithic platform that struggles with the release and eCommerce innovation issues we have discussed. If you touch one part of it you don’t know how it is going to impact the other part of it.               

Habits, Once Changed, are Changed for Good

Due to COVID-19, have your customers experienced an increase in traffic and transactions on their eCommerce / digital sites?  Can you explain the types of growth they have seen?  Do you expect this growth to continue?

Yes, all of our customers have experienced massive growth. We have seen that first-hand with the amount of capacity we have had to provision in the backend, because we are multi-tenant SaaS. I don’t have exact numbers, but it has been substantial. 

We do most of the grocery store chains in Central Europe and the Nordics, and they have especially seen traffic skyrocket because it’s all digital. With the exception of Sweden, it’s all buy online, curbside pickup. We have seen that across the board for everyone. People are sitting and watching tons of tv these days, so our media customers are seeing a lot of growth. Even in the B2B space we see business activities still remain the same if not higher. We do Tottus in Chile, which is their Home Depot or Lowes equivalent, and lots of people are using this down time to do projects around their house. So, we see traffic across the board everywhere expanding and we have been able to handle it perfectly.     

There was a great study that came out by Goldman Sachs, and they basically said this is the new normal, and people are notoriously difficult to change their habits but once they do, they change for good. If I look at my own mother-in-law, she can barely use her phone to call, but does buy online curbside pickup from Walmart. She said to me two weeks ago “How did I ever live without this?” She figured it out, and now she will never change. She will die placing orders at Walmart and getting them curbside, and parking in the same spot. 

I think with apparel, especially luxury apparel, people want to go to those stores in person. I think a chunk of that will come back. A lot of people are trying to think about why they spent so much time in mass merchandised stores. It didn’t make sense, and it’s a lot cheaper and easier to just buy stuff online and have it shipped to you or have it picked up. I do think this is normal, and I do think it is a generational thing as well. 

If you look even just locally — I live in middle America and nobody under the age of 40 pays for gas at a gas station by walking into the store. But there is a good chunk of people in their 40s, 50s, that are walking into that store and paying with cash every single time. People pay for things with checks, like at grocery stores. You would not find somebody under the age of 40 paying with a check ever; it’s all debit or credit card. 

I think a lot of that is just older people (70s, 80s) that did not grow up with anything at all, just don’t get this stuff and they won’t touch it — and that’s where a lot of buying power is. Younger folks, people who grew up in this, people 50 and below, they get it and it makes a lot of sense. Certainly, many people have learned, but it’s hard to have people go back and learn those skills.             

What would be your advice to brands as they continue into what will be the most unpredictable holiday season ever?

It’s easy for me to say make big investments in technology because I am not the one who is going to get fired if it doesn’t work out. It’s August, and brands should not be adding anything significant right now. Realistically, it’s time to optimize what they have and get new eCommerce innovation projects cued up so that when holiday is over, they can start releasing a lot. My advice would not be to not make any big new changes and make sure you test everything to ensure performance is ready for the holiday.        

How important is it that brands maximize holiday revenue in 2020?

Of course, it’s important for everyone to maximize their holiday revenue. I would say since physical retail is basically dead right now or very impaired, you have to focus on digital. I mean, that has to carry your finances for you. I would say it is very important, but I can’t quantify it. I can quantify it by saying ‘if it’s not good, you are probably going to be fired.’   

What are the most important steps an eCommerce leader can take to prepare their site and maximize conversion when traffic and transactions increase, especially when the increase is unexpected?

To use an elastically scalable set of technologies. There is a big difference out there between hosted software, where a vendor is calling it “Cloud,” but it’s really just software that they manage on your behalf, and a truly elastically scalable Cloud solution — vendors like us at commercetools, and I am sure YOTTAA as well.

Like a lot of these newer vendors out there, we autoscale everything top to bottom. If Ronaldo has 260 million instagram followers and posts a link on instagram to a shoe that is one of our customers, we will survive that, because we autoscale within a couple of seconds the entire platform top to bottom. And if you are on one of those fake Cloud offerings, there is no way they could possibly scale the platform out that quickly. I would say the most important step an eCommerce leader can take is to use real Cloud technologies, and “real Cloud” implies elastically scalable multi-tenant SaaS — not fake Cloud (hosted), or as we did a marketing campaign awhile ago “Fake New.”