Banks in the United States share their customers with competitors in line with the size of those competitors. This law-like pattern, the ‘duplication of purchase law’, has been found in numerous other product categories, for both goods and services. The findings here inform us about how a brand will grow, and the extent to which targeting a particular segment will work or not.
Services marketing textbooks usually emphasise the importance of targeting, customised/differentiated services, and achieving high customer loyalty levels. Banks love the idea of cross-selling to their clients (see our report on Wells Fargo), ensuring a high level of ‘share of wallet’ and perhaps even sole loyalty: “our customers choose to bank only with us!”
But many bank customers aren’t solely loyal, they are dual-brand or multi-brand loyal. This raises questions for the marketing team: which other banks do our customers use? Does it tend to be another bank with a ‘brand position’ that is like ours? Or is it a competitor that offers something we don’t? Is there a key competitor that we need to focus on, to learn what it does to steal our business away? Should we survey our customers and ask them what particular things they were looking for that led them astray into the arms of a competitor?
These questions, it turns out, can be broadly addressed from an aggregate-level pattern regarding how brands share their customers with competitors – the Duplication of Purchase (DoP) law. The term ‘duplication’ is used to communicate that the same consumer can appear in the buyer base of several different brands (i.e., as if they are duplicated). The DoP law states that brands share their customer base with competitor brands in line with the size of those competitors (see Who do you Really Compete With? and Understanding How Brands Compete: A Guide to Duplication of Purchase Analysis). In other words, more of our customer base overlaps with the other big brands in the market and much less with the other small brands in the market. The DoP law is important because it tells us that competition between brands isn’t based on target segments (brand a appeals to that sort of customer, brand b appeals to a different sort of customer – see Brand User Profiles Seldom Change and Seldom Differ); rather, it is ‘wide open’ and direct. Evidence shows the DoP applies in a wide variety of markets: consumer goods (Scriven & Danenberg, 2010), cars (Colombo et al., 2000), quick service restaurants (Lynn, 2013), tourism destinations (Dawes et al., 2009) and – even banks in the United Kingdom (Dawes, 2014; Sharp et al., 2024).
Scientific laws are called laws if they hold over a wide variety of conditions. What about retail banking in the U.S.? At face value, there are some reasons to think the law might not hold here. One reason is that the U.S. is not necessarily one market – it is geographically very large, diverse, and not all banks have a strong presence in all parts of the country. For example, M&T Bank has 700 branches but they are mostly in the East, while Regions Bank is concentrated in the South. Also, many U.S. banks appear to pursue quite different marketing strategies.
Let’s look at some examples. Chase, the biggest – is putting a huge emphasis on branch proximity (Cocheo, 2022) – an aspect of physical availability. Perhaps its management has read How Brands Grow (Sharp, 2010)? Many others are pursuing variations of the brand purpose theme. Bank of America is heavily emphasising social purpose (TBH, 2024). Santander is promoting its leadership in global sustainability (Santander US, 2023). Truist seeks to carve out a differentiated position by trying to fulfil customers’ emotional needs, not just their functional needs (Fisher, 2022). Ally is a digital-only bank, spending heavily in women’s sports sponsorships and targeting female buyers (Hamill, 2023). Citizen’s Bank states it wants to stand out from the pack with a differentiated offering that targets young college graduates with a $50k+ income (Streeter, 2019).
So, many of these bank brands attempt to differentiate themselves and/or pursue specific target audiences. And some of them have a strong presence in different parts of the country. Given all that, could there be a coherent pattern in customer sharing? We address that question by examining how closely it conforms to the DoP law.
We used YouGov brand index data to determine if U.S. banks share their customers as the DoP law would predict. YouGov runs a large, population-weighted survey panel in various countries, including the US, where the panel includes over a million respondents. Respondents are asked which bank or banks they currently have a financial product with. From this data, we cross-tabulated the proportion of each bank’s customers that use each of the other top 22 banks that collectively account for 98% of the market.
We assembled data for the calendar year 2020. We assembled the brands in descending order of size (rows and columns – in line with good ‘data reduction’ principles) making the data pattern easy to see (see Making Data User-Friendly). The results are shown in Table 1.
We interpret the table in this way. The column labelled ‘Pen (% using) shows the percentage of people who use each bank brand. We see that 24% use Chase, 18% use Bank of America – down to 1% using Truist and BMO Harris. Reading along the rows we see the proportion of each bank’s buyers who use each other bank. So, looking at Bank of America, we see 35% of its buyers also bank with Chase, 25% bank with Capital One, down to 1% also banking with BMO Harris. The broad pattern is that a much larger proportion of any bank’s customers also use the other big banks (e.g., Chase, BoA, Capital One), and far fewer also use the other small banks. We calculate the average amount of duplication, or customer sharing, shown in the bottom row. This shows that on average, 34% of any bank’s buyers also bank with Chase, 24 with Bank of America, 27% with Capital One, 18% with Wells Fargo and so on, down to 3% on average for the small players Truist and BMO Harris. While the pattern isn’t perfect, we see the duplication of purchase law holds well in this market. The MAD, or ‘Mean Absolute Difference’ between actual duplication levels between these banks and our predicted level from the DoP law, is only 2.8 points, which is quite impressive. The correlation between actual and predicted duplication across all the observations is 0.89. It can’t get much better since correlations only go to 1.00.
But still, let’s now explore why a few brands share their customers more than what we expect, given the broad DoP pattern. For example, we see 52% of Citibank customers also bank with Chase, much more than the average figure for Chase of 34%. Another example is that Fifth-Third and Huntington share customers much more than expected, at about 10% in each case, compared to their average duplications of about 5%. Is there a simple explanation for these ‘market partitions’?
The Ehrenberg-Bass Institute has been doing research for decades on patterns in brand competition, including why some exceptions occur. We have found that it often boils down to a functional difference between the brands. An example in consumer goods is that diet soda brands share the ‘diet’ characteristic compared to normal soda drinks, so they customers to a greater extent than expected from DoP.
We first note that two banks specifically accommodate armed services personnel and their families (USAA and Navy Reserve Credit Union), and they certainly share customers between themselves to a greater extent than expected. That said, they do still also generally share their customers with competitors in line with the DoP law – more with the other big brands and less with the other smaller brands! In our following analysis, we code these two brands to control for the specialist brand effect.
We also consider simple geography. As mentioned, some of these brands are much more prominent in different parts of the country. For example, Chase has somewhat more customers in the Northeast and West than in the South or Midwest, while USAA has more customers in the South and West than in the other two regions. This means banks with a more similar geographic profile to another bank will share their customers with it more than expected. In contrast, banks with a different geographic profile to a particular competitor will share their customers less than expected. We mentioned two banks with considerable over-sharing earlier – Fifth-Third and Huntington – this could be because they both have many more customers in the Midwest than in other regions.
So, we calculated the geographic profile for each bank and then the profile difference between each pair of banks in the data to use in the subsequent analysis.
One other factor we considered is what is called the Natural Monopoly law. This law is that market-leading brands tend to ‘monopolise’ the lighter buyers in the category (Dawes, 2020). We see evidence for this in Table 1. In the rightmost column, we show the average amount of sharing each bank does with all the other banks. We see, for instance, that the average figure for the top two brands, Chase and Bank of America, are around 8% and 9%. When we reach the bottom of the column, the figures are 13% or 14% for the small banks Truist and BMO Harris. This pattern reflects Natural Monopoly in that the customers of the biggest brands tend to be lighter banking users, so they don’t need to use other banks as much. In contrast, customers of the smallest brands tend to be heavier category users, so they will be more likely to use multiple banks. Hence, their average levels of customer sharing are higher.
We then reran the DoP analysis, including these effects: specialist brands (armed services), geography and the natural monopoly law.
From this slightly more complex analysis, the MAD improved, but only slightly, from 2.8 to 2.7. The correlation between actual and predicted duplication increased slightly, from 0.89 to 0.91. So, allowing for geographic differences and the brands that accommodate armed services personnel improved our predictions, but only very slightly. Therefore, we find that a simple pattern, the DoP Law, explains almost all of the variation in how bank brands share their customers with competitors.
So, the overriding story here is that bank brands share customers in line with size.
But what about those that depart from the general pattern? The first answer is that we wouldn’t know they did without knowledge of the general DoP law: it gives us something to judge the sharing between particular brands. Second – even though some brands over-share with one or two other brands, such as Fifth-Third and Huntington – those brands still overwhelmingly share in line with size, as we can verify by reading across their respective rows. And the explanation for their over-sharing turns out to be reasonably simple, such as that they are concentrated in different parts of the country. In the case of USAA and Navy Reserve, these brands accommodate an identifiable segment: armed services personnel. Still, even here, these brands share their customers more with big ‘mainstream’ brands, Chase and Bank of America, than each other.
As we read earlier in the report, the marketing teams for many of these banks develop plans to present them in quite different ways to audiences or attempt to try and target or appeal to specific segments (females, young college graduates and so on). However, the simple DoP law shows that overall, competition between bank brands is very direct, very ‘head on’ and growth will therefore necessitate appealing broadly. That is – the size or market share of every bank in the market depends largely on how well it manages to attract some portion of every other bank’s customer base. Trying to cut oneself off from the majority of the market by pursuing a small segment would appear to be a recipe for being or becoming a small (or failing?) player.
The DoP pattern is also informative concerning loyalty. We know that banks extol the virtues of high loyalty: aiming for more ‘share of wallet’ or pushing to sell that one extra product per customer. And, of course, banks need to be alert to cross-selling opportunities. But the DoP pattern helps us think about how much we can grow by relying on loyalty improvements – we see that plainly, a very large proportion of any brand’s customer base uses at least one of the more than twenty other banks. This helps us to set realistic expectations for loyalty. And consider that Chase has become number one not because of its higher loyalty. We can see that in the data – most of its customers still deal with other banks! Chase is number one because it has attracted more customers – from every other bank. They still deal with those other banks, too. And therefore, for any of the other banks in the market to achieve significant growth they must emulate the same pattern of customer acquisition.
It appears that bank marketing plans are out-of-date, not in line with evidence-based marketing.
Can’t find what you are looking for? or have some feedback about the site? Contact Us
Now available as an eBook exclusively to Apple iBooks
The Ehrenberg-Bass Institute for Marketing Science is the world’s largest centre for research into marketing. Our team of market research experts can help you grow your brand and develop a culture of evidence-based marketing.
