Omni-channel learnings from Sally Beauty – an IBM e-commerce fulfilment overhaul at agile speed
Omni-channel learnings from Sally Beauty – an IBM e-commerce fulfilment overhaul at agile speed
Thu, 01/28/2021 – 03:19
- In part two of this use case on Sally Beauty’s digital transformation program, the retailer’s CTO talks through the shift to agile deployment to address its e-commerce fulfilment needs.
The first part of this two part article on Sally Beauty Holdings omni-channel retail transformation – here – focused on the enormous uptick in e-commerce activity triggered by the COVID crisis and the consequent need to focus on the customer experience as salon doors temporarily closed and customers shifted to DIY beauty routines from their own homes.
The other challenge resulting from this e-commerce boom was a hugely practical one to overcome. Put simply, demand exceeded ability to fulfil orders. This caused the retailer to introduce new initiatives, such as curbside pick-up. Buy Online, Pick-up In Store (BOPIS) and ship-from-store, all at a considerable pace. As Joe Condomina, Chief Technology Officer at Sally Beauty Holdings, recalls, things “went like bonkers”:
We not only dipped our toe in the water, we jumped straight in with the most difficult implementation, not only ship-from-store, but same day delivery out of the store, out of the gate, and that was basically five months after we launched our first implementation on Sally.
Roll back to March 2020 and Condomina and his team are still working from the office and sitting down for an emergency brainstorm about what it now clearly an oncoming crisis. This resulted in the realisation that the firm needed to roll out a ship-from-store capability in a three week window:
If not, we were going to have our stores shut, we may likely run out of inventory, the distributors won’t be able to distribute merchandise to us to backfill the distribution center and we’re gonna have some serious problems.
Agility in action
With an end of March deadline locked in place to put a ship-from-store capability in place, work took place around the clock. It was, admits Condomina, something of a chaotic process, involving as it did different warehouses, different websites and different stores. The situation was further complicated by Sally not being well-versed in agile techniques across the enterprise, he adds:
We were being agile on the periphery, but we were not an agile shop. We quickly had to move to agile code development with our business. We had a lot of virtual sessions.
The underlying technology to enable ship-from-store was mainly structured around IBM’s Sterling Order Management System (OMS), which supported all the rules for order distribution and how they would get to the shipping nodes, explains Condomina:
But then we actually had to deliver it to the end point, which was in the store. We didn’t opt to choose to use the OMS storefront and then have it fulfilled there. We wanted to integrate it into our PoS [Point of Sale] to have a richer experience for our sales associates. So imagine a little bit of complexity there, along with the order brokering mechanism that Oracle Xstore [also in use] brings to the table. Then, underneath the covers, we also had to figure out a way on how to deliver messages on to the cloud side of IBM Sterling. We didn’t deliver that on prem, that was cloud-based, and our Oracle Xstore implementation is also in the cloud, a different cloud. So we had to figure out how to get all of that into the equation.
And as noted earlier, all of this had to be done at a very accelerated pace, something which did give Condomina some pause for thought:
I was a little bit intimidated because I’ve deployed [OMS] in the past and they’ve taken a lot of time. And everybody told me, ‘IBM Sterling OMS is no different, it’s going to be a tough implementation’. We’ve not had that experience. Has it been tough? Yes, it’s been tough, but it hasn’t been difficult. A lot of that is because of how we chose to deploy the technology, cloud-first, the plumbing that we already had in place to connect the clouds, and then the structure of our team and just a rock solid implementation team from [IBM partner] Proficient. We’ve been able to capitalize on those things and deploy really, really quick.
Ship-from-store was first up:
That was a very quick implementation and that’s where we really kind of carved our teeth and really sunk into the Agile delivery method. Each of the teams now functions in their own agile space, specifically on the e-commerce side and on the OMS side, and that’s that’s what really enabled us to go quick. We even brought our PoS team along into that journey from an agile delivery and as you know, PoS teams don’t like to deliver on a monthly basis – it’s quarters, it’s every other month, very calm structured deployments and that’s not what we’ve had with this.
Agile is now increasingly part of the corporate DNA, he says:
We’ve been really agile. We’ve been risk-takers, but we’ve been what I would say is calculated risk-takers. We’ve never done anything to risk the business. When you really start looking at your deployments, you can deploy these pieces of functionality [quickly]. You hear the term all the time, ‘Perfect is the enemy of good’ or ‘Fail fast, do small incremental releases’. We have been living that and we’ve been showing that you can be successful with that.
After ship-from-store launched at the end of March, the next step was to stabilize that environment to make sure that it could handle full capacity, that the firm didn’t have things just shipping here and there across the country, but that inventory was in the right location. That demanded up-to-date information at all times, explains Condomina:
With the velocity of orders being so high, a five minute delay in inventory [data] does not cut it; you have to go real-time and that was one of the things that we implemented in the summer. That was an incremental release. How long did it take? Probably about a month to implement. So, every single one of these things that we’ve done has been like two-to-three weeks, a month.
Once this foundation was in place, work on BOPIS capabilities began mid-September after an enforced delay:
The reason why we started so late and didn’t start that earlier, is because we had another program that was consuming a large part of the team and that was our private label credit card implementation. That took a large capacity. Imagine delivering and stabilizing ship-from-store, delivering a private label credit card program, and following that doing BOPIS!
But BOPIS was clearly going to be essential long-term as customer shopping practices altered, even once salon doors re-opened, he says:
Customers were already learning a learned behavior of, ‘Don’t walk into the store, socially distance’, and we knew that we needed to enable that. The only way that we were going to enable that with success was to deliver BOPIS and all the reporting around it, so that the business could know how good the program was doing and ensure that we’re meeting our SLA, [as well as] all the communication that you have to deliver to the customers via text via email, a follow up with a phone call.
Again, all of this was enabled in a very short space of time, with BOPIS piloted at 20 stores by the end of October. Fourteen days later, work kicked off on expanding the deployment to 3,000 nodes across the entire Sally enterprise, climaxing the week before Black Friday.
Most recently, Sally has completed its first implementation of real-time inventory across all of its distribution network, something that needs a bit more work, admits Condomina, as it’s “probably not best-in-class”. Other future plans are also taking shape, he adds:
There are some products from IBM that are coming out in the first part of this year, and also from Salesforce, that we’re looking to tie together to really get millisecond level updates from an inventory standpoint, from the store all the way to the consumer, to eliminate that friction point of when they click on it, they know that they’re going to get it. That’s probably going to be one of the big things underneath the covers that nobody’s going to see that we’re really working on to improve, because even a half second delay sometimes does mean disappointment for the customer, especially in some of our low inventory nodes. If you walk into a store, they’re small footprints. You can’t have thousands of products on the shelf just waiting for consumers to come in, so we have to balance some of that.
The other big objective for 2021 is to shift focus more firmly onto the professional beauty stylist audience, having spent 2020 catering to the uptick in demand for DIY amateur customers. The work done there will be transferable, says Condomina:
2021 is all about the Pro and bringing all of that functionality that we built for the retail customer into the Pro’s hand. It’s going to be a next level thing, because if you are a Pro and you shop with us, I will say that the experience is a little lacklustre online, because we’ve been really focused on stores and getting that store experience for those stylists to be very good. A lot of 2021 will also be on improvements in the underlying infrastructure. We still have some legacy integrations that are done on older technologies and we’re looking to move that into the 21st century, go more services-based, micro-services type of approaches.
He concludes with a simple maxim that will guide tech agenda priorities:
Those are the things that we’re really looking to improve, but all of this is with the intent of removing customer friction and thinking ‘customer first’. If it doesn’t impact the customer, we’re not going to do it, we’re not gonna spend time on it. There’s no sense in it.
Image credit – Sally Beauty Holdings
Disclosure – Condomino took part in the NRF Big Show -Chapter One. At time of writing, Oracle is a premier partner of diginomica.