OnClarity Insights
How do we grow without cost to serve growing with us?
Asked by a technology leader at a fast-growing fintech.
6 min read
The short answer
Watch contacts per customer, not only cost per contact. Cheaper handling lowers the cost of each contact once and growth brings the volume back, while fixing the reasons customers contact you lowers contacts per customer, a saving that grows with every new customer. Put contacts per 1,000 active customers on the board pack every month.
Why does cost to serve keep growing with the business?
Most growing companies go after one number: what each contact costs. They move contacts to cheaper channels, add self-service and put an AI agent in front of the queue. Cost per contact falls, the business case closes, and a year later the support budget is climbing again.
That is not because the automation failed. It is because growth multiplies whatever ratio you started with. If each customer still needs to contact you as often as before, more customers means more contacts, just at a lower price each.
Technology alone does not take the cost out either. Gartner predicts that by 2028 over 50% of customer service organisations will double their technology spend without an equivalent reduction in talent (press release, 31 March 2026).
And the savings that do arrive often come in under plan. In Bain & Company’s 2026 survey of 951 companies, nearly 40% of those that measured AI cost savings landed below 10%, despite targeting 11% to 20%.
What is the equation behind cost to serve?
Cost to serve has three parts:
Cost to serve = active customers × contacts per customer × cost per contact
Growth raises the first part. That is the point of growing. Automation, self-service and cheaper channels lower the third part, usually once. Only the middle part asks whether the customer needed to contact you at all, and it is the one most service reports never show. The idea is not new: Bill Price and David Jaffe argued it in The Best Service Is No Service.
Here is what the equation does over three years, as an illustrative example with round numbers. Customers grow 30% a year. In the first path nothing else changes, so cost to serve grows with them. In the second, cost per contact falls by 20% once. In the third, cost per contact falls by 20% once and contacts per customer fall by 10% a year, because the causes behind them are being fixed.
Illustrative example, not customer data. Index: today = 100.
After three years, customers are at 220. Cheaper contacts alone leave cost to serve at 176. Cheaper contacts plus fewer contacts per customer leave it at 128. The second path saves once. The third keeps saving as you grow.
Which levers bend the ratio, and how long does each saving last?
Rank them by how long the saving lasts, not by how fast it shows up.
1. Remove the causes. The policy, product, billing or communication that makes customers contact you in the first place: a confusing first bill, a delivery update that never goes out, a setting nobody can find. Fix one and contacts per customer fall, for the customers you have and for every customer you have not met yet. The saving lasts as long as the fix holds.
Most of these causes sit outside customer service. In SQM Group’s 2024 benchmark of first call resolution, 49% of calls that were not resolved first time traced back to the organisation, not to the agent or the customer.
2. Make the first contact the last. Resolve fully the first time, tell customers before they ask (a known delay, an outage, a change to their plan) and stop the second and third contact about the same problem. This lowers contacts per customer too, but only while the process holds, so it needs watching.
3. Make handling cheaper. Self-service, AI agents and agent assist lower the cost of each contact that remains. It is worth doing, and worth doing fast. On its own, though, it is a one-time saving: the next wave of customers brings the volume back.
The order matters. Fix first, then automate what is left. Automating a contact that should not exist just makes it cheaper to keep paying for the same problem.
What should the board see every month?
One ratio: contacts per 1,000 active customers, or per 1,000 orders if you sell by transaction. Report it monthly with a 12-month trend, split by the top contact reasons, and put repeat cost removed next to it: the cost of repeat contacts before a fix minus the cost after it, counted only when the fix held.
Cost per contact on its own hides growth. A falling cost per contact next to a flat contacts-per-customer line means you are paying less each time for problems that keep coming back. A contacts-per-customer line that falls while customers grow is the sign that the business needs less support for every customer it adds.
Who owns contacts per customer?
The causes mostly sit in product, billing, logistics and policy, so the ratio belongs to the COO, or whoever runs operations across those teams. The CX leader measures it, brings the evidence (the conversations, the volume, the cost) and names an owner for each of the top causes. Customer service cannot fix a confusing bill on its own. It can make sure the person who can fix it sees what it costs every month.
The honest trade-off
Not every contact should go away. Sales conversations, retention calls and high-value questions are worth having, so track them separately or leave them out of the target. A new market, product or price change will lift the ratio for a while, so read the trend, not one month. Fixing causes needs other executives to act and is slower than buying automation; SQM’s own research notes that organisational causes take longer to fix than agent errors. And the ratio can fall for the wrong reason, because customers gave up, so watch complaints and cancellations next to it.
What to do this week
Split it by contact reason and rank the reasons by volume times cost.
Name an owner outside customer service for each of the top three causes.
Put the ratio on the next board pack, next to cost per contact.
Set a target to bend it, and review it monthly with the owners.
Questions people ask
What is a good contacts per customer ratio?
There is no universal benchmark. The ratio depends on your sector, how complex the product is and how you count contacts and customers. Compare against your own last 12 months instead, split by contact reason, and watch the direction as you grow. If customers grow and the ratio falls, each customer needs less support.
How do you reduce cost to serve in customer service?
Work three levers in order. Remove the causes that make customers contact you, make the first contact the last, then lower the cost of handling what remains with self-service and AI. Track contacts per customer next to cost per contact, so savings that fade as you grow stay visible.
How do you calculate cost to serve?
Take the fully loaded cost of customer service for a period (people, technology, outsourcing and overheads) and divide it by active customers. To see what drives it, split it into three parts: active customers, contacts per customer and cost per contact. Use finance’s figures so the number holds up in the boardroom.
Sources
Gartner, press release “Gartner Predicts Over 50% of Customer Service Organizations Will Double Their Technology Spend By 2028”, 31 March 2026.
Bain & Company, Your AI Budget Is Growing. Your Returns Aren’t. Here’s Why., 1 June 2026 (Automation and AI Pathfinder Survey 2026, 951 companies).
SQM Group, Call Center FCR Benchmark 2024 Results by Industry, 2024.
Bill Price and David Jaffe, The Best Service Is No Service, John Wiley & Sons (2011 edition).
The three-year example above is illustrative. It uses round example numbers, not OnClarity customer data.
