A few years ago I worked for a customer that was under substantial financial pressure due to decreasing retail sales, and the incumbent CEO had initiated a programme of cost reduction. The part that I was involved in was an implementation of what is known as SAP Profitability Analysis (COPA).
SAP Profitability Analysis allows customers to evaluate the profitability of market segments and provides flexible analysis according to products, customers, orders, business areas etc. In the Retail Industry, it is particularly useful to flag an item as "like-for-like", which is when a product can be compared as like-for-like within a particular store. So we can do profitability analysis where the same items are not like-for-like (usually when a store has been refitted, to see if profit increases) or when they are like-for-like, which usually says more about store managers or economic conditions.
This is pretty powerful stuff and it allowed a cost-conscious CFO to come to a very interesting conclusion: they had always believed that their own designed and made products were the most profitable. But it turned out that that it was the items they outsourced the design and build for were by far more profitable. Empowered with this data, they moved to a programme of outsourcing more of their product collection.
In project terms, we had one major problem which was that getting good performance out of the system was possible for simple questions, but the business analysts wanted to ask complicated and ad-hoc questions. I have no doubt that if we did this project again today using an in-memory database like SAP HANA, we would have completed the implementation in 30% of the time (we spent months performance tuning), at half the cost, and it would have answered any ad-hoc question about Profitability Analysis. This brings me to my first point:
1) Invest in In-Memory technologies
Your project will happen faster (my calculation was 30% of the time for this example project), you will get your information faster and you will be able to ask any question, rather than being restricted. This means you can get on with running your business and making strategic decisions on the basis of facts - which is what most CEOs want to do in the first place.
Now unfortunately this story doesn't end so well for my customer. It turns out that whilst their house-designed and manufactured products were less profitable, they were responsible for the brand value. The value had already been eroded by the outsourced products already and moving to a mostly-outsourced model destroyed the brand value. Two CEOs and a sale to a restructuring company later, the company went into administration and was sold off into pieces. Which brings me to my second point:
2) In-Memory technologies do not a good business strategy make
I would argue it is possible that had my customer been able to implement SAP HANA at that time, they would have gone bust faster because they would have found the information they believed they needed to know faster.
On the other hand it is possible that they would have been able to ask much more sophisticated questions using business analysts - running statistical trend analysis or basket analysis on the core sales data, and figured out which products contributed to brand value and which products contributed to profit and therefore setting out the store in such a way that they sold customers a balance of the two. And this neatly brings me to my third and final point:
3) In-Memory technologies allow you to truly differentiate
against your competition
I think that history will bear out my last point here. You are either investing in in-memory or someone else in your industry will, because it is truly differentiating if you can figure the right strategy and business problem to solve in it. There is no point in my trying to template such a solution for you because then it wouldn't be differentiating, but consider the following scenarios.
The ability to recognise and react to a customer at point of sale. Bring in historic sales data, loyalty card transaction information and make decisions about offers to give to a customer during the sales process - increasing loyalty.
The ability to perform cluster and basket analysis on billions of line items of raw sales data without having to wait for weeks for the answer - just getting a response in a few seconds - allowing very rapid changes to store layout taking into account external factors like weather and news.
The ability to do clickstream analysis on all of your website data and historic sales in real-time and make offers to customers based on very complex rules - up-selling your customers and reducing the chance of returns.
These are three of the business problems that retailers I have spoken to are thinking about and I think any one of these could be differentiating to a retailer. However this is just the tip of the iceberg and the real differentiating scenarios are locked away in the minds of the CEO. In my opinion, those who have the courage to invest now will get a differentiating place in the market.