One of the challenges business leaders have using analytics is identifying the return on investment (ROI). While ROI seems like a relatively simple concept on the surface, the devil, as they say, is in the details. In this article, I’m writing about calculating ROI on analytics.
To keep this topic from spinning completely out of control, I’m dividing the article into a few specific sections:
- Basics of Calculating ROI
- Considerations for Calculating ROI
- An Alternative Calculation for ROA
- Other Analytic Performance Metrics
Let’s dive in…
Basics of Calculating ROI
At its core, return on investment is not a complex calculation. You need to know two things: the cost of your investment and the revenue from your investment. The ROI formula is:
\( \begin{equation} ROI = {\frac{\text{(Revenue from Investment – Cost of Investment)}}{\text{Cost of Investment}}}*100 \end{equation} \)Simply put, ROI is the net gain from an investment, expressed as a percentage of the cost of the investment.
If you want to apply ROI to analytics, a simple approach is to replace the word “investment” with “analytics.”
\( \begin{equation} \text{ROI on Analytics} = {\frac{\text{(Revenue from Analytics – Cost of Analytics)}}{\text{Cost of Analytics}}}*100 \end{equation} \)Your cost of analytics is relatively straightforward; it’s generally the cost of software, data storage, computing power, and analyst compensation.
Revenue, however, is more difficult for you to determine.
You must carefully consider which items count toward revenue from analytics. Additionally, you’ll consider how much of each revenue item should be attributed to analytics. Finally, you must also consider the time frame you are using to gauge revenues and costs.
Considerations for Calculating ROI
You could consider many elements when thinking about how to attribute revenue to analytics. I won’t claim to include them all here.
However, I include three important elements that you should consider every time you want to calculate ROI on analytics.
Which items count toward revenue?
When you think about the revenue or cost savings generated from analytics, which items will you include?
If your analysis leads to improvements in products or services, you will include those revenues.
If your analysis leads to reductions in staffing hours required, you will count the costs saved.
And if you are looking at improvements in process efficiency, translate those into cost savings.
To get more precise, you could dig into specific components of processes, products, or services to attribute to your analytics.
For example, your analysis might identify a marketing message that improves sales among one customer segment. In this case, you would attribute additional sales in that segment to your analysis.
The key is to critically assess the business aspects impacted by analytics when you identify revenue and costs.
How much revenue should be attributed to analytics?
Once you identify your revenue and cost items, you must consider how much revenue to attribute to analytics.
The key question is: how much revenue or cost savings can reasonably result from your analytic efforts?
Suppose your analysis identified a new market segment for your service. You might think you should attribute all of the revenue from that segment to your ROI.
But would that make sense?
Your marketing department still needed to create content and collateral materials. Your sales department still needed to identify prospects and close deals.
Sure, your analysis deserves some credit for identifying the new market segment, but not 100 percent of revenue.
Additionally, you may have business units working to make improvements independent of analytics. If this is the case, you may need to split the between analytics and other units.
For example, your analysis might indicate a growing preference among a key customer demographic from social media. At the same time, customer sales and support may receive feedback from clients identifying the same trend.
Analytics may ask marketing to target the customer demographic. At the same time, sales might revise their messaging to prospects in that demographic.
So, who gets the credit?
This is where attribution gets sticky and will require careful thought about your business processes, staff performance, and fairness and equity.
How long did it take to gain insights and generate revenue or savings?
A final consideration you must think about when calculating ROI on analytics is the time frame required.
Analytics take time to implement and execute. Your analysts’ time on a project should count toward analytic costs.
However, you must also consider how long it will take to see action and results from your analysis.
For example, if your analysis indicates making product revisions requires six months to execute, you’ll need to factor that in.
You’ll also need to consider the time it takes to achieve a return, either through increased revenue or cost savings.
Do you give yourself six months for a return? A year? More or less?
This will depend on the project and how quickly changes trickle through to generate returns.
Regardless of the analysis, carefully considering the time frames necessary to see a result is important for ROI calculations.
An Alternative Calculation for ROA
In the simple ROI calculation I presented above, time only plays an implicit role. We assume that revenues and costs are calculated over specific and equal time periods.
But they don’t have to be.
You could measure revenue over any length of time, regardless of the time frame for your costs.
Avinash Kaushik and Jesse Nichols[1] present a formula for ROA, which makes the time duration explicit.
\( \begin{equation} ROA = {\frac{(R_a – R_m)*(d)}{I_a}} \end{equation} \)where,
- Ra: Return using analytics
- Rm: Return without using analytics
- d: duration of time (e.g., # days, weeks, months, etc.)
- Ia: Costs to get the impact attributed to analytics
In the ROA calculation, Ra and Rm are captured under the same time duration (e.g., daily, weekly, monthly, etc.). The number of time units is captured in d. And the investment, Ia, is captured on a time horizon less than d.
As an example, you might have a company making $100,000 in revenue with a marketing budget of $25,000. That’s a 300% ROI, and worth every penny.
Then suppose you hire an analytics team for $10,000 a month on a three-month contract. That’s $30,000 in analytics, which then produces $150,000 in revenue each month for the next 12 months.
Your ROA would be:
\( \begin{equation} {\frac{($150,000 – $100,000)*12}{$30,000}} = {\text{2,000%}} \end{equation} \)So, a six-month investment to develop and deploy analytics boosted your revenue by 50 percent. Over the following year, with that performance, you achieved a 2,000 percent ROA!
You deserve a bonus…
Other Analytic Performance Metrics
In contrast to the revenue and cost framework to calculate ROI on analytics, you might also choose to track actual analytic performance. Stephen Tracy[2] addresses this with the following metrics:
- Time-to-Insight (T2I)
- Time-to-Action (T2A)
- Time-to-Deployment (T2D)
With each of these metrics, Tracy focuses on assessing the time investment of your analytics. However, additional business units may play a role in the time measured by these metrics.
Importantly, you should note that these metrics are relative to the complexity of the analyses, actions, and solutions you deploy. You will understand how long each phase should take by tracking these performance metrics across projects.
Time-to-Insight (T2I)
T2I assesses the time it takes from when your organization collects data to when your analysts produce actionable insights. In essence, T2I measures lead time for analytics to create value for your organization.
If your analysts take six months to produce insights from real-time sales data, you will constantly be behind the ball. Your goal, then, is to shorten the T2I metric for critical analyses.
Time-to-Action (T2A)
T2A assesses the lead time from data collection to taking specific actions toward deploying a solution. Because you can act on a solution without deploying it, T2A doesn’t require deployment.
Your analysts might produce insights quickly after data collection. However, if your internal processes add weeks or months to the time required to act, you have bottlenecks robbing you of analytic returns.
Time-to-Deployment (T2D)
T2D assesses the lead time from data collection to your organization rolling out solutions drawn from analytic insights. It captures the project’s complete life cycle for analyses leading to concrete organizational changes.
As with T2A, T2D will help you identify potential bottlenecks to realizing analytic returns. T2D also represents a metric that can benchmark your future efforts to improve efficiency.
Conclusions
Calculating ROI on analytics can be a challenging exercise. This is especially true when you attribute revenue and cost savings to analytics. You must consider the revenue and cost items to focus on, the portion attributable, and the time frames required.
Fortunately, you can approach the idea of measuring analytic performance in multiple ways. You could focus on traditional ROI-style calculations like Kaushik and Nichols’ ROA calculation. Alternatively, you could use a KPI-style performance metric like Tracy’s T2I, T2A, and T2D.
Regardless of your choice of metric, remember that there is no single “right” way to calculate ROI. Rather, there are better and worse metrics depending on why you are measuring return or performance.
Choosing the best metric will depend on you, your business, and your goals.
[1] Kaushik, Avinash, and Jesse Nichols. Excellent Analytics Tip #22: Calculate Return on Analytics Investment! February 25, 2013. Available at https://www.kaushik.net/avinash/calculate-return-on-analytics-investment/. Accessed on October 14, 2024.
[2] Tracy, Stephen. How to Measure the ROI of Data Analytics. June 22, 2015. Available at https://analythical.com/blog/measuring-analytics-impact. Accessed on October 14, 2024.
