Strategy Dynamics Breifing 17: The problems with correlation

A devastating implication arises from this interdependency between resources from the previous briefing. Here is a simple story…

A company’s profits come from its revenues minus its costs, and its revenues reflect the number of customers. Amongst its costs, marketing spend mostly acts to win customers, rather than increase their purchase rate

  • The company raises marketing spend in order to grow sales and profits
  • But first, its profits fall because of the higher marketing cost
  • However, the marketing spend does win a few more customers, but the small growth in sales is not enough to pay for the continuing higher marketing spend.
  • The following month, customer numbers and sales rise again, but still not enough to pay for the higher spend.

This could continue for many months, during the whole of which time the higher marketing spend seems to have caused lower profits.

  • The head of finance loses patience and pulls marketing spend back to a lower rate even than at the start.
  • The saving leads to an immediate jump in profits – to a higher rate than at first, due to the new customers who have been won in recent months.
  • With lower marketing spend customers are won slower than they are won, so sales fall.
  • Profits start to decline from month to month, but are still higher during the period of high marketing spend.

Lower marketing spend seems to have caused increased profits.

Diagram: Marketing spend inpact on profits

The data here is longitudinal, i.e. the same few observations repeated over many periods. But it could also be ‘cross-sectional’ – the same factors observed at the same time across many firms in the same market, each with its own history (on the lower chart). If you used normal regression methods to see if ‘marketing drives profits’, you could easily ‘prove’ a negative correlation. Positive correlation, or none, is also quite possible, but all possible findings are meaningless.

Problem is, the accumulating stock of customers destroys any direct link from marketing to sales and profits.

Today’s results reflect the whole history of marketing spend.
Simple rule: Don’t trust regression results when there may be accumulating stocks between the outcome you want to explain and the factor[s] you think are driving it.

… and here we only have one such item. Think what a mess you could get when several accumulating items are at work, for example when R&D spend grows new technology, which leads to more new products, which grows customer numbers…! No wonder we can’t work out what drives business performance by studying large samples of companies.

Until next time…


If you would like to receive the series from the beginning in your email inbox, please register on on our website and subscribe to Briefings in “My Account”

What ‘performance’ do we want?

We have largely taken it for granted up to now that what we want – at least in business cases – is growth in profits over time. Curiously, that’s not how academic research into business strategy generally views the question. Instead, this research tends to focus on levels of profitability – return on sales or invested capital, for example.

The reason for this is the micro-economic foundations of strategy research, where researchers want to know why some firms are more profitable than others. But investors value cash flows, specifically the present value of the cash flows they expect to receive in future periods. They don’t care, then, if firm A is less profitable than firm B, provided that A’s cash flows are growing faster. In extreme cases, investors accept very low, or even negative profitability for many years – still giving the firm’s shares a positive and growing value – because they expect to see rising cash flows in later years.

Strategic Management Dynamics book coverTo see how important this distinction is, consider who made losses right up to 2002, and continued to make very low returns on sales and invested capital for some years. On the first basis, researchers would be wanting to explain why Amazon had been so persistently unsuccessful, whilst investors were happy to enjoy its real-world success!

This briefing summarises discussion from chapter 3 of Strategic Management Dynamics,
pages 177-179

Read more about the book on our website

Leave a Reply

Your email address will not be published. Required fields are marked *