One of the characteristics of core customers is that they resemble the typical customer imagined when the shop or service was first thought of. That is the way the service is from the product perspective is close to spot-on already for these customers. This fact also means any changes to site functionality risks a large volume of revenue but only a small number of customers. If, in an AB test, the unit of randomisation is a user of this detrimental effect can be lost because of the small numbers of core customers.
The key to growth beyond the current levels therefore is to find pockets of ‘near core’ customers with specific needs similar, but not exactly the same as, core customers and to try to cater for them specifically.
Who near-core customers are is context dependent, for instance they could be customers similar to the ones you have as core customers now except older, or younger, or with interests that are similar: a camping shop might find kayaking a good market to enter. What counts as a near core customer depends very much on who your core customers are.
Often sophisticated clustering analysis will help to identify good ‘near core’ customers because some of your customers are near-core already. They, like core customers, a homogenous group with very similar behaviour and although they form a small percentage of current users and can offer a good opportunity to grow the business.