What happens when a startup company grows to the point where its existing systems for managing customer and order information can’t keep up?
That was the challenge facing one online retailer, which has grown rapidly over the past five years and seen its homegrown CRM system — as well as initial choice of ERP software — struggle to keep pace. Accordingly, the business turned to Cloud Sherpas for help with evaluating a move to new tools, including a cloud-based CRM application from salesforce.com.
Before selecting or implementing new CRM software, however, Cloud Sherpas is helping the retailer step back and ask some essential questions, including defining who counts as a customer, as well as detailing exactly how data will be shared between essential business systems.
Startup-Worthy MySQL Database Isn’t Scaling
According to Cloud Sherpas’ principal technical architect Grace Buchanan, the retailer’s legacy environment poses some interesting challenges. Notably, its “main brain” is a MySQL database. That was a good choice for a startup because it’s free and can be hosted locally. But now that the business has grown — and on its way to $100 million in annual revenues — the current approach isn’t keeping up.
Why is that? For starters, the CRM system and collection of ERP systems don’t talk directly, but rather draw what they need from the MySQL database. And, frankly speaking, as a result, the firm’s database schema is a jumble. Accordingly, before the company adds a new CRM system or ERP application — which will likely be SAP — it has to answer some essential questions.
Getting Existential: What’s A ‘Customer’?
First, the business is addressing this question: “What is a customer?” For example, when you add products to a shopping cart and check out, are you a customer? What about if you enter a credit card number, but it hasn’t yet been authorized (which can take up to two days)?
Accordingly, the company must decide whether to count someone as a customer before or after payment is received.
Similarly, the company must define what counts as an order. Is it a single shipment? Does the series of monthly shipments that are part of a service offered by the company count as one or multiple orders? These decisions will impact how the company configures and shares information between its CRM and ERP applications.
Debate: Where Should Customer/Order Information Live?
While Cloud Sherpas is helping the company select and implement a CRM application, we’ve also been helping it answer some of these more big-picture data questions. Based on the company’s current infrastructure, as well as long-term business goals, we’ve recommended that the company evaluate Informatica, to ensure that essential data can flow between the ERP and CRM systems, and in the future, the company’s business intelligence tool, amongst other systems.
For example, ideally the “golden copy” of all order information will live in — and be owned by — the ERP system, while customer information will reside in the CRM system. That’s the approach that Cloud Sherpas, together with the company’s lead technology executive, has been advocating. This approach will be backed by a tool such as Informatica PowerCenter or Informatica Cloud, which will enable the systems to communicate and share required information. (Other options that might also work include Cast Iron, which is owned by IBM, as well as Dell Boomi and Microsoft BizTalk Server.)
This will create a service layer — or enterprise service bus — that will let the company create a hub-and-spoke architecture, and not talk so much about where the data lives, but rather who owns it and needs to manipulate and access it, so it can be appropriately sent back and forth.
What Happens When HR Wants In?
Currently, however, there’s a debate inside the company, with some people advocating that the order and customer information remain in the MySQL database, with IT maintaining on-premise control. What’s wrong with that approach? Beyond overlooking the benefits of a cloud-based application — including ease of use and manageability — it would scale poorly.
By sticking with the MySQL database, every time the company wanted to tie in another application, such as an HR system, it would have to create connectors from the new system to the main brain, and then to the CRM and ERP systems. In fact, the HR department has already said it wants to pull data from the CRM system, for example to better calculate incentives for salespeople.
Required: Error Tracking, Complex Orchestrations
By creating a service layer, however, the company can add new systems to its heart’s desire, thanks to not having point-to-point integrations — and an aging database attempting to run the show — but instead having a distributed data storage model with a good deal of added intelligence.
Indeed, tools such as Informatica provide reliable delivery of data, meaning you have delivery queues, retries, errors and other essentials. You also gain real-time and batch processing and the ability to create complex mapping as well as complex orchestrations for long-running transactions.
In this case, the complex orchestrations involve data coming in through the website, then a customer account being created in the ERP system. This is followed by data going through the service layer to the CRM system, where a new prospect gets created and eventually becomes a customer. This process may take weeks, but a product such as Informatica can keep track of the entire process so that, a couple of weeks down the line, when a prospect does become a customer, the system can retrieve all information or preferences they’ve previously shared and help them pick right back up again.
Future: Nuke The Main Brain
Service layers also serve as an excellent foundation for future data sharing. Notably, going forward the company is planning to not just adopt this service layer approach, but also eliminate the main brain. But that’s a separate project, given a number of other legacy systems that tap into it. Even so, thanks to this being a retail environment, as the company’s lead technology executive says, the rate of decay of bad data is relatively quick, which means that as the company upgrades its CRM, ERP and other systems, the main brain will eventually just get left behind.
