Cohort analysis is a management tool to analyze time-dependent groupings of both customers and invoices. The more common of the two by far are customer cohorts, but invoice cohorts are also very interesting in the context of recurring revenue businesses. They allow for a waterfall display of the revenue recognition and deferred revenue remaining after grouping the invoices into cohorts according to their respective billing month, quarter or year. With this waterfall and drill-down to the underlying invoice detail and attributes in every cohort, it becomes easy to perform both a high level visual review and a granular audit of any cohort where the recognition waterfall appears to be out of line (e.g., due to early or late billing or unusually long or short subscription periods vs. the other cohorts).
Customer cohorts are groups of new customers for a month, quarter or year. Cohort analysis displays how the retention or attrition for each customer cohort trends over time. Retention or attrition can be displayed in either MRR or customer count terms. And totals and changes can be displayed in both revenue and percentage terms. Customer acquisition cost (“CAC”) is also often factored into the cohort analysis, as in one dashboard below, to highlight how the payback period for these up-front growth costs is changing over time by cohort.
Cohort analysis can be powerful tool to uncover both hidden successes and hidden failures at a relatively early stage, particularly after operational changes are made. For example, as new marketing campaigns are launched, or as new sales or support staff are hired, all such changes tend to have a significant impact on new customers. The impact also tends to be outsized on new customers, since a solid customer relationship has yet to be established.
By cross-referencing the timing of such changes to the affected customer cohorts, a cause-effect relationship can often be quickly established to explain significant fluctuations in the cohort analysis. For adverse fluctuations, this could then lead to immediate remedial action, such as retraining or replacement of underperforming employees. For positive fluctuations, it could obviously lead to reinforcing actions, such as continuing or ramping up of successful sales and marketing campaigns.
Cross-referencing operational variables and attributes to explain cohort analysis fluctuations can often be done “off the cuff” based on readily available knowledge. However, it is not always so easy or definitive in an enterprise-scale recurring revenue business. This is especially true when multiple BUs are involved who may share marketing campaigns and resources across multiple departments.
To address this problem, segmentation analysis adds a whole new layer to the cohort analysis by allowing a deeper dive into the potential root causes of positive or negative cohort fluctuations. By allowing both customers and products to be segmented by a virtually unlimited number of attributes – both standard and customized – one can change the analysis in one click to show how all the cohorts look for any particular customer or product segment. One no longer has to remember when every marketing campaign took place, when staff changes occurred, which sales or support staff were assigned to which customers, which customers got larger price increases this year, etc. To do a deeper dive, just select the campaign, assigned sales or support staff member, price increase category, etc. from the appropriate segmentation hierarchy to see if any of these can explain the aggregated cohort analysis fluctuations. Please see the “Segmentation Analysis” product page for more information on the segmentation functionality.
