Retail inventory is a game of balancing two competing interests:
- keeping inventory high enough, in enough places, to be able to guarantee good customer experiences
- keeping inventory low enough to minimize the associated risk and costs.
Getting this balance right has a direct impact on margins, and there has long been software dedicated the problem. But inefficiencies have endured due to the limitations of available solutions, and of brands’ legacy technical architectures.
Invent Analytics is intended to solve this problem. A composable inventory optimization tool, it’s designed to eliminate the guesswork and introduce greater profitability to brands’ demand chains.
Founder Gurhan Kok joined us on The Martalks Podcast to discuss how his technology works, and to describe his successes with likes of Camper, and Academy Sports + Outdoors.
Gurhan was also kind enough to share some insights from his commercial strategy, which are likely to be of interest to solutions vendors in analytics and beyond. In particular, we covered the importance of short-time to value as a competitive advantage, and how to achieve this despite the varying quality of clients’ data and technical architectures.
Listen to the full conversation here, or read below for a summary.
Counting the cost of poorly-optimized inventory
As Gurhan explains, “inventory is the blood of retail. You need to control the flow of it to make customers happy – and at the same time, do that in a cost-effective way.”
The formula for inventory optimization, therefore, is made up of product, location and time.
“The role of the supply chain is to bring those products to the marketplace fast, at low cost, and inventory is the outcome of that. In a retail store & distribution system network, product, location and time create what we call a chemical reaction that brings supply and demand together.
And when you bring the right supply together with the right demand at the right time, that chemical reaction is what generates the margin for the retailer.”
Failing to optimize this balance is a longstanding and costly problem for retailers. Gurhan referred to studies from many years ago which showed that, in cases where customers found the right product in a retail store, they were unable to the find the correct size on 30% of occasions.
For enterprise retailers, the cost of these poor customer experiences is multiplied, particularly when selling via multiple outlets in an omnichannel context.
Illustratively: if you have 100,000 products and 100 retail locations, you’re tasked with optimizing 10m combinations. This work falls in large part to the planning team’s intuition, supported by outdated technology.
Pre-year 2000, Gurhan was as PhD student in operations research at The Wharton School, studying big data, predictive analytics and retail optimization. There, he participated in projects with Best Buy and other enterprise retailers – and realized that the algorithms underpinning the technology in use at enterprise brands were more basic than those well-known in academia.
Beyond that, even retailers with large R&D departments were unable to improve on those algorithms and incorporate the improvements into their technical architecture.
That particular challenge is now gradually being mitigated by the rise of composable architecture – but only gradually. Clients new to Invent Analytics will often still operate legacy technical architectures, with various components hardwired to a legacy ERP, and varying standards of data management.
But with brands now increasingly selling in omnichannel contexts, there is now an urgent need for analytics solutions which can thrive regardless of technological constraints. The task for the vendor is to anticipate this complexity in their product design, and in their commercial strategy.
AI-powered inventory analytics
This complexity exists both within the business, and in the customer domain.
Within the business: a typical brand will have separate supply chain and planning departments. The different departments will have their own sets of KPIs – which may in some cases be conflicting – and data siloed in different systems.
In some cases, brands may have laid the groundwork for solving this problem by adopting a centralized cloud-based data platform. This does not, however, compensate for the limited ability of their legacy analytics software, to process large and diverse data sets in order to support favorable decision-making.
This consequence is that inventory planning, at many enterprises, is still based on incomplete data and human judgement. Gurhan describes the industry standard as ‘a series of post-mortems and guesswork’ – whereby businesses are making future decisions based on previous successes and failures.
This is as opposed to making decisions based on the most recently available data and a calculated probability of ROI.
In the customer domain: omnichannel retail has compounded these challenges. This is exemplified by Invent Analytics client Camper, a high-end European shoe brand, for which Invent Analytics carries out omnichannel forecasting. This entails forecasting…
“…the demand in-store and online, in each region, each metropolitan area, etc., and how they should position inventory in distribution centers and stores, given that a very strong percentage of their sales are online. And a strong percentage of online demand is fulfilled not only from DCs but from stores, and even sometimes from a store in another country, in the case of Europe.”
The consequence of this complexity is that Camper must keep more stock, in a greater number of locations, in order to ensure good customer experiences. Such a demand chain therefore comes with far higher direct cost, than the business models around which traditional inventory analytics evolved.
Invent Analytics’ answer to this what Gurhan describes as a ‘financial optimization system’, whereby inventory planning is guided by the probable ROI of individual decisions.
Gurhan compares it to…
“…playing blackjack at 1,000 tables at the same time, where if you count cards, you can reach 51% accuracy and win more hands than with the legacy system.”
Invent Analytics comprises an AI component which improves its ‘bets’ as it goes. By processing more data than would be possible through either manpower or with legacy technology, the system quickly produces a direct impact on the business bottom line.
Gurhan says that his solution is typically produces 1-1.5% improvement in business profits – and sometimes as a high as 5%. But even taking the lower 1% figure, this difference adds up to very significant savings.
“If you’re a $10bn retailer, every percentage point generates $100m incremental profit. A $100m incremental profit per year is $8-10m a month. So every month you’re delayed you’re losing $8m. We would rather start gaining that – earning that – and that money could fund many other internal projects: IT or otherwise.”
Such gains are all the more welcome in the present economic climate, where higher interest rates have increased the cost of money, and consequently, the cost of holding stock.
Needless to say, this makes short time to value a key factor for any solution vendor promising to solve these problems.
Resolving corporate inertia through the power of probability
Though larger retailers are ahead of the curve, even they would admit to “significant data accuracy problems” says Gurhan. This should make the promise of a unifying analytics solution an easier sell.
A widespread obstacle in the space, however, is the length and difficulty of implementation. Typically, it would take at least a year to secure resources, agree on requirements, carry out any necessary customization and train users client-side.
Part of Invent Analytics’ answer to this problem – as with many other modern SaaS vendors – is their internal architecture. As a cloud-based, API-first solution, the system connects easily to the brand’s existing technical architecture, while the work of maintaining and running the system sits vendor side.
But what struck me, as possibly Invent Analytics’ greatest competitive advantage, is how this technology enables a highly-effective commercial strategy.
With a new client, the focus is on showing quick ROI by optimizing only a small section of the client’s inventory, in order to win buy-in to then scale across the entire business.
Gurhan says,
“Perfect is the enemy of better, so we don’t wait for perfect. We wait for acceptable quality data: 70-80% in a month rather than 90-95% in a year.”
Invent Analytics compensate for these gaps in the data with its advanced algorithms and AI functionality – which drastically compresses implementation and training times.
Gurhan gives the example of Academy Sports + Outdoors. Despite the client having over a million SKUs, they were able to go live in only 3 months – compared to the 12-18 months implementation time that would be typical for the industry.
Being able to provide reliable insights in this way then frees up the planning team to carry out more effective work.
A traditional analytics vendor, explains Gurhan, would ask a planner what their service level should be for a given product and location. By contrast, Invent Analytics actually provides these answers to the planners – based on a degree of probability, but with greater accuracy than the planners themselves typically manage.
The same features allow the software to assist in ongoing discussion and experimentation in the client’s team. Where a planner has a hunch that they could improve optimization by tweaking inventory in certain areas, the software can simulate alternative scenarios – helping to control the associated risks of making any changes.
As a result, the brand gains far greater certainty in both short- and long-term forecasting, and in setting strategy the following year.
Composable architecture accelerating time-to-value for retailers
Benefits such as these are what one would always hope to gain from inventory analytics, and wider business intelligence tools.
The difference, today, is that composable architecture has enabled this software to fulfil its promise far more effectively. By integrating more easily, and by being able to process data more intelligently than with legacy tools, Invent Analytics has produced a solution that’s both highly productive and relatively easy to adopt.
Of course, only part of this can be attributed to the vendor’s internal architecture. Gurhan’s background in academia has brought with it the benefit of an outsider’s perspective, leading to a technically superior piece of software.
That uniqueness of vision, combined with a savvy commercial strategy, is the mark of many a pioneering software founder, and one that ought to inspire founders and innovators in any industry.
About Rosenstein Group
Supply chain & analytics are familiar territory for the Rosenstein Group, having spent over 20 years as the #1 executive search firm for martech scaleups.
In one case study, we helped an India-based retail analytics vendor break ground in North America by recruiting their first VP of US Sales. Within 2 quarters, the successful candidate had closed 2 record deals, and the company subsequently closed a $23m investment round.
Read the full case study here, or find out more about Rosenstein Group’s work in supply chain & demand chain here.
Apple Podcasts
Spotify
Google Podcasts
