This article originally appeared on Forbes.com.

Bank profitability has taken a significant hit since the global financial crisis, forcing many banks to focus on reducing costs. They have made less headway, though, on realizing the full potential of their relationships with corporate customers. Huge opportunities lie in a more rigorous approach to managing corporate client profitability, particularly if interest rates rise over the next few years.

Higher rates will likely increase the dispersion of return on risk-weighted assets, or RoRWA, among banks, with the laggards holding level at best and the leaders potentially doubling their rate of return. The catch is that banks must make money in the meantime. They have to figure out which clients are “profitable” today, which are not and how to serve the latter group in a different way—or move them off the books. Otherwise, banks will be limiting their upside on profitable revenue growth.

In a typical corporate portfolio of a large multinational bank, Bain & Company estimates that economic profit derives from roughly 20% of clients. Within that group, perhaps 1% account for most of the value. At the other end of the portfolio, some 30% of clients fall below the threshold for economic profit, with about 1% having the greatest negative effect for the bank as it currently serves them. The remaining 50% or so have moderate value for the bank or just break even.

The lowest profitability clients often account for a sizable share of a bank’s risk-weighted assets. Typically, the bank has made large loans at favorable rates and terms, without getting ancillary, higher-margin business—such as transaction banking, risk management or advisory services—in return. The level of profitability will also depend on factors that fluctuate over time, including the cost to serve, the dynamics and cycle of the client’s industry, and market interest rates. Many banks, though, lack a system that can track the ebb and flow of value for each client.

Banks that have been gaining control of their portfolio’s profitability generally have begun by categorizing clients into tiers based on economic profit and company size. Profitability models use some combination of data on revenue, RoRWA and return on equity. Once the bank has sorted its client base into the tiers, the discipline of profitability should pervade the life cycle of every corporate client.

Onboarding. When a bank is deciding whether to bring on a potential new client, calculating expected profitability should be a routine part of the decision. Onboarding has become more expensive because of compliance, regulatory and user-experience considerations, so it should be undertaken only when a bank can expect to have a sustainable relationship. The assessment will include indicators such as client size; typical needs for transactions, trading, bank guarantees or contingent loans; and the share of wallet that the bank can reasonably expect.

Deal pricing. When weighing a loan or other product offer, the bank should look beyond the marginal contribution for the specific transaction to whether the deal is accretive or dilutive to the relationship.

One global bank has integrated a pricing tool with its customer relationship management (CRM) system, to pull historical data about a client’s past deals and profitability. Certain elements, such as the bank’s desired profit level, are automatically programmed in, so that the tool produces a marginal contribution for the client as well as the expected future profitability. This bank also uses rules-based decision making, which limits exceptions on individual deals.

Account planning. Best-practice banks go beyond an annual account plan to collect information on historical client revenue and the profitability of the relationship, along with estimates of future potential. This profile feeds into a uniform planning process across the bank. The account plans inform many client decisions over the following year, such as the pricing of specific deals, coverage level and a broader discussion of other product offers. In the best case, the CRM system can automatically create a projected account plan based on the previous plan. That means staff don’t have to fill in every item from scratch and can update the plan at any point as circumstances change.

Client management. Not all clients will be profitable every year, given the nature of ancillary deals and the cyclicality of demand. So leading banks take a two- to four-year view.

An effective practice we have observed at another global bank is how it uses the CRM system as an intermediary between relationship managers and service functions such as the credit or internal capital teams. For example, if the profitability of a client drops below a certain level, the team responsible for overall profitability asks the relationship manager to devise an action plan housed in the CRM. When that manager logs in, an alert will prompt him to create the plan. The team can review it, share it with a senior committee if need be and respond through the CRM. Everyone shares the same information, sees what has been committed and knows what to track.

It’s never easy to nudge a client to profitability or to cast them loose if the bank cannot serve them profitably. But having solid data to present to the client, and an explicit plan to turn things around, injects rigor into a historically loose and overly personal process. Such rigor will allow banks to seize opportunities that will only grow larger as interest rates rise.

Jan-Alexander Huber, Iwona Steclik and Thomas Olsen are partners with Bain & Company’s Financial Services practice.