The benefits of a cloud data warehouse platform
We are in the age of Big Data, Artificial Intelligence, and Machine Learning. There is a lot of discussion around how AI is going to transform life as we know it; a lot of it is hype, but there are many applications where Artificial Intelligence will have a significant impact. Despite all the hype around AI, one of the oldest technologies, Cloud data warehouse, still remains a vital piece of a company’s technology landscape. It continues to be the best way for organizations to deliver business intelligence that executives and managers alike can leverage to understand what happened in the business and make more informed decisions. It also continues to be an enabler for executives to assess overall business performance from a single dashboard that they can consume when they need to without waiting on someone.
Data warehousing has been a staple in large and mid-market companies for a few decades now to address questions regarding business performance and prediction of future performance based on historical data. However, despite its usefulness, its relevance in this day and age was questioned not too long ago. But, executives in the know, are aware of its importance and continue to invest in this crucial technology stack and use it to power decision making. This aspect is also punctuated by the fact that the most significant database company in the world, Oracle, released its own version of the cloud data warehouse, Autonomous Data Warehouse, a self-driving database that enables the adoption of data warehousing technology for companies large and small. Another startup, Snowflake, has raised over $300M for its data warehousing technology.
Let’s take a look at what a data warehouse platform is and why it has become even more important to have now than it was until a few years back.
There are two kinds of systems in a company. Transactional or OLTP systems and analytics systems. Most applications that are used on a day to day basis are transactional systems; order processing software, CRM, finance, etc. However, running queries on these systems, especially the ones that are historical or analytical in nature place undue burden on the production systems rendering them slow for users who use these systems for more important activities that are transactional in nature; ex. Placing an order in the OMS. It is never fun to keep a customer waiting on the phone because the system slowed down dealing with a report that was run.
However, reporting is also an important aspect that needs to be performed. To overcome the problem of historical reporting or analytical querying slowing down transactional systems, this data is extracted and loaded into a dedicated database that is designed to support reporting and analysis: a data warehouse. Additionally, since data warehousing by nature acts as a repository for all business data, relevant reporting and analysis can be produced that can give business leaders more actionable intelligence.
Until a few years back, there were still relatively few systems that supported day-to-day operations in comparison to what we have today. For instance, ERP software that comprised of inventory management, financials, supply chain, CRM, customer support, etc. have now decomposed into best of breed SaaS applications each with its own data store. These SaaS applications have gained tremendous adoption of the last decade and a typical organization now, regardless of the size, has a sizeable number of these applications in use to perform day-to-day business operations. These SaaS applications make feature-rich applications available to every company. The undesirable side effect is the creation of data silos. These data silos make data consolidation into a data warehouse even more critical now that it was in the past.
Let’s take a simple marketing use case at an e-commerce company, a marketer interested in finding out Return on Ad Spend (ROAS) and communicating it to their management.
ROAS = Revenue / Cost
ROAS = Revenue registered in ecommerce platform like BigCommerce /
Sum of all spend made on ads in ( FB + Google + Email + Instagram + Display Advertising)
Just to calculate one single metric, a marketer has to look at six different systems daily, if not multiple times a day.
A consolidated data warehouse set up to load Ad spend data from all these marketing sources and transaction data from BigCommerce, or another e-commerce platform can be achieved today in a matter of minutes. An analyst can now create an automated report that has the Ad-spend metric calculated automatically multiple times a day which a marketer or an executive can use numerous times a day without having to do any manual work.
Let’s take another example: an e-commerce brand is selling its wares on multiple channels including their online store, on Amazon marketplace and a couple of retailers like Nordstrom or Macy’s. The CEO wants to know their sales and profitability every day by 9 am. How would one achieve this?
Total Sales = Sales in E-Commerce store + Sales in Amazon + Sales in Nordstrom + Sales in Macy’s
Imagine how that formula will be calculated because we have sales data from multiple places.
You’d have guessed it right if you came up with this series of actions.
1) Log in to these e-commerce stores, Amazon, Nordstrom, and Macy’s sites.
2) Run reports on all four provider sites.
3) Download CSV or Excel versions of those reports to your laptop.
4) Consolidate these spreadsheets daily and email the management team.
5) Do this every day and pray that you are not making mistakes.
Cloud Data Warehouse – A compelling alternative!
1) Create a data warehouse with automated data pulls from these systems
2) Combine data from these sources in the data warehouse.
3) Create a report in one of the BI tools like Tableau, Looker or PowerBI
4) Set up the dashboard to be delivered to your management via email every day.
5) Did they ask you to send it more frequently? No problem, just change the frequency in step 4 instead of repeating steps 1-5.
If you are an executive at a company that doesn’t yet have a data warehouse, ask yourself these simple questions:
- What is the value of having easily accessible, automated, business performance data when you need it?
- What are you willing to pay for it?
- Will it be beneficial if the impact of customer service on your business is known to you immediately?
- Will it be beneficial if you catch poor marketing campaigns as they are happening?
- Will it be beneficial to know which products give you the best margins and how that changes over time?
- Will it be beneficial to optimize the inventory of your products based on historical performance rather than overstocking or understocking them?
- Will, my company, be better off if marketers focused on marketing, customer service focused on customer service, or will I be better off if these resources are spending 10%-20% of their time answering your questions?
All these and more can be answered quicker and cheaper than at any point in time in the past.