How to Define KPIs for Your Strategic Digital Analytics Goals

by Apoorv Kashyap

Businesses cannot afford to pour money into digital analytics without a clear and achievable vision of the value return on investments. Often, this vision revolves around top-line and bottom-line revenue. A goal to achieve x% new customers within a fiscal year could be a good strategic business goal, or to achieve a reduction of y% of returned in-store orders, or even to achieve a z% increase in customer engagement (likes, comments, shares, etc.) on Twitter or any other social engagement channel.

Data science, machine learning, and artificial intelligence use historic and current data sets from a business’ back-end systems (e.g., transactional data), along with digital behavioural data (e.g., Adobe) and third-party data providers (e.g., credit scores) to generate customer insights. The insights produced from the above techniques could be directly actionable information, like a product recommendation for a given customer, based on their previously identified customer segment (or even directly based on that specific customer’s past behaviours). The insights produced could also be more strategic – for example, a merchandising analysis based on sales, inventory, procurement, and demographics data for the past 18 months.

In order to manage and succeed in your digital analytics program, it’s important to identify a handful of KPIs that reflect the true nature of activities most relevant to achieving the overall business objective (x% increase in customers, y% reduction in churn, etc.). Below are some examples of mid-level KPIs that are generally good indicators that your digital strategy is panning out as planned.

1. Increase in unique website visitors

Even though this is a very generic KPI, the devil is in the detail. A vision to increase unique website visitors will be heavily powered by marketing campaigns aiming to convert paper/non-digital customers into digital customers, as well as to new customers.

Target: Shoot for 10% increase and consider seasonality.

AI Modelling Strategy: Considerable customer analytics can be applied to find the potentially most likely customer segments to target with marketing campaigns.

Caution: In order to accurately identify “unique” customers and across all of your web traffic, you will need to create an ID graph that can identify a unique customer between the multiple visits they’ve made to your website, from various browsers and during multiple browsing sessions.

2. Increase in conversation rate

It’s quintessential to getting as many unique visitors on your website to be converted into paying customers.

Target: Achievable up to 20% with aggressive discounting and other personalisation strategies.

AI Modelling Strategy: This problem is well suited for dynamic price discounting.

Caution: The key in this scenario is to provide the right amount of discounting to convert a tentative buyer into a paid customer. The discounting model needs to perform to sharp ROC curve in order to make the most business sense. Therefore the discounting model needs to be monitored and reviewed much more frequently in order to gain the best value from it.

3. Increase in average order value

When you do have a customer who buys from you, it’s important to maximise the order value by offering the most appropriate product recommendations for the customer to choose from.

Target: 10% achievable with the right mix of approaches.

AI Modelling Strategy: Best suited for personalisation and product recommenders.

Caution: Needless to say, all the above KPIs are highly interdependent, so the customer segments that you identify to use in the first metric feeds into the rest of the approaches. Therefore it’s important to have clear AI pipelines that join the various applications together in a scalable flow which doesn’t break when there’s more traffic on your website.

4. Increase in up-selling

In our opinion, based on several decades of customer analytics, when the customer feels more comfortable with your brand and with your digital experience, they are likely to choose higher-end products when the option is available.

Target: Up to 50% up-sell opportunity achievable.

AI Modelling Strategy: Best suited for product recommender and next-best-action strategies.

Caution: Treading into softer side of human nature, product recommendation and next-best-action strategies can easily put off a customer if applied too aggressively and hence should be used with caution.

5. Reduced cost-per-click

As mentioned before, all of the AI advancements listed above have a cumulative effect on the customer’s digital experience and your spend on targeted digital advertising, thereby reducing the average cost per click for a given customer.

Target: 5% reduction year-on-year is achievable.

6. Number of FTEs refocused

This is a great KPI that has immediate impact on the bottom-line of a business. Freeing up a single business or IT resource equates to saving anywhere from $25,000 to $100,000 in direct costs. Again, this is much more of a by-product if the above activities are done correctly, as operational overheads will get reduced and business processes will streamline.

Target: If 10 FTEs can be refocused on more of revenue generating activities, that can be equivalent to putting up to $1M yearly investment back into business growth.

In summary, even though there are hundreds of detailed metrics that you can define to measure the success of your digital analytics program, the above 6 KPIs can give you a good starting point. Note: the above six KPIs were first published in an Adobe and Forrester report.

If any of this resonates with you, email me (Apoorv Kashyap) and share your thoughts or experience with KPIs and digital analytics. You can also request a briefing to learn about developing trends in marketing and analytics.

Share this:
Tags:   AnalyticsStrategy

Apoorv Kashyap

Head of Customer Value Advisory at Syntasa

Apoorv Kashyap heads the Advisory & Customer Success at Syntasa, with a goal to help early-mid maturity Digital businesses to increase Digital-channel sales, improve customer experience, and reduce operating costs by joining the Cloud technology, Data-led strategies, Mar-tech stack (re)architecture. Apoorv is a trusted advisor to the heretics of Digital, Data & eComm, and proudly carries two decades of battle scars in areas of Insights, Data pipelines & Machine Learning-led Digital Transformation programs. He is also a dad, a rock-climbing fanatic, and a motorcyclist.