By David Smith, ASA, CFA

Long gone are the days of old when a simple estimate of future attrition could be acceptably based on qualitative factors gleaned from an interview with company management. Attrition analysis – now a near absolute requirement when engaged in purchase price allocation projects that include the appraisal of existing customer relationships – has evolved. No more qualitatively-based estimates as to what the “turn-over” rate has been among a company’s “top customers”. Now it’s all about complex, attrition analysis models that can generate multiple indications of past attrition based on headcounts, revenue weighted headcounts, and revenue on both a static rate and variable rate basis. Some aspects of the analysis that you may need to consider...

No year-by-year customer revenue data

The lack of available year-by-year customer revenue data used to be accepted as justification for moving to a qualitative input based attrition estimate. But it just doesn’t cut it anymore. Those performing the AU336 review expect something more concrete and certainly more replicable. Ask for the person responsible for sales and you’re much more likely to find that year-by-year customer revenue data.

Headcount vs. Revenue Weighted Headcount

Sure, calculating attrition based on headcount is quick and easy. But if there is a reasonable expectation that attrition among the company’s “top customers” may differ materially from that among the “other customers”, a headcount based analysis may not be enough. Add the revenue weighted facet to the analysis and you’ll get an indication of attrition that is a better indicator of future revenue decline from existing customers.

Pure Revenue-Based Attrition

Sounds great in theory, but (i) the analysis is certainly more tricky than a revenue weighted headcount analysis and (ii) you can get lots of non-attrition factors (inflation and market/industry expansion/contraction to name a few) impacting the indicated attrition rates. This may work well in mature, stable markets/industries, but you can get some difficult to reconcile/apply indications of attrition if that is not the case.

Variable Rate Attrition

Great in theory, but it requires more years of customer revenue data to apply it. Forget the minimum two to three years that you might get by on for a static attrition rate analysis. Try at least five to six years of data if you hope to get meaningful results. I have not heard of this being pushed for in AU 336 reviews, and despite the fact that you can get a better indication of the “pattern” of future attrition, you may be adding a level of complication that isn’t required.

Measurement Periods

So what is the appropriate “measurement period”? Well, like so much in the business appraisal world, it depends. Using a one year measurement period works in many situations. But, if the relevant industry has unusually long sales or production cycles, you may have to adjust your analysis. Likewise, in other industries (telecommunication services comes to mind) the relevant measurement period may be less than a year. Just something to consider...and remember to adjust how you apply the indicated attrition rate if using something other than one year measurement periods for your analysis, but are applying it to a customer appraisal model based on one year periods.

Under-Analysis

Don’t just grab the indicated attrition rate output from your attrition analysis model and run with it. You really need to look at the individual indications of attrition rather than assuming the average or median “works”. A wide dispersion of indicated attrition rates may require some additional analysis and/or discussions with company management to better understand the “story behind the numbers” and may call for some adjustments.

Customer Differentiation

Sure it’s easiest to just toss all the customers into a single attrition analysis model, but that isn’t justified if there is a reasonable expectation that customers differ materially by profit margins, use of other assets in providing the product/service (differing capital charges), or yes, attrition rate. Consider the possibility that a single attrition model may not be appropriate. You may need to divide the customers into more homogeneous groups and run two (or more?) models to get an appropriately robust indication of attrition for each group.

 

This article originally featured in BV Success E-Letter 18-13: Attrition Analysis in Purchase Price Allocation
BV Success is a publication of the Business Valuation Committee of the American Society of Appraisers

 

J. David Smith, ASA, CFA
David Smith is a Principal with Hill Schwarz Spilker Keller LLC. He has over twenty years of financial consulting and financial services experience including fifteen years in the appraisal of businesses, business interests, intangible assets for corporate mergers and acquisitions, financial reporting, corporate tax, recapitalization, estate and gift tax, estate planning, employee stock ownership plan and litigation purposes.