Support cost per conversation

Support cost per conversation is the unit cost of handling a single customer interaction. It takes the total cost of running the support function and divides it by the total number of conversations handled. If a support team costs £150,000 per month to operate and handles 10,000 conversations, the cost per conversation is £15.

The "total cost" should be fully loaded: agent salaries and benefits, team lead and manager compensation, software and tooling subscriptions, training and onboarding costs, office or remote work infrastructure, and any outsourcing fees. Including only direct labour costs understates the true cost and leads to poor decisions about automation and channel investment.

The metric varies significantly by channel. Phone support typically has the highest cost per conversation because it requires an agent's undivided attention for the entire duration. Live chat is less expensive because agents can handle multiple conversations simultaneously. Email and ticket support allows agents to work asynchronously and batch their effort. Self-service and chatbot interactions have the lowest marginal cost because they require no human agent time.

Support cost per conversation is not a metric to minimise at all costs. Reducing cost per conversation by cutting agent training, rushing interactions, or closing conversations prematurely will degrade service quality and increase downstream costs through repeat contacts, escalations, and customer churn. The goal is to reduce cost per conversation while maintaining or improving quality, typically through automation, process improvement, and channel optimisation.

A metric tree breaks cost per conversation into its cost components and volume drivers, revealing the levers available for optimisation.

The tree reveals that cost per conversation can be reduced by either decreasing costs or increasing the number of conversations handled for the same cost. Automation achieves both: bot and self-service conversations cost a fraction of agent-handled ones, and they increase total conversations handled without proportional cost increases.

The tree also shows the relative weight of each cost component. For most support teams, people costs represent 65% to 80% of total costs. This means that the most impactful cost levers are those that affect agent productivity: reducing average handle time, increasing concurrency, and deflecting simple queries to self-service.

The order-of-magnitude difference between phone and self-service explains why support organisations invest heavily in channel migration and automation. Moving 20% of phone conversations to self-service can reduce overall cost per conversation by 15% to 25% depending on the starting channel mix.

1. 1

   Invest in self-service and automation

   Every conversation resolved by a bot or knowledge base article costs a fraction of an agent-handled conversation. Analyse your most common conversation topics and build self-service solutions for those that are well-defined and repetitive. The ROI on self-service investment is typically the highest of any support initiative.

2. 2

   Reduce average handle time

   Shorter conversations mean more conversations per agent-hour. Invest in response templates, customer context panels, internal knowledge bases, and diagnostic tools that help agents resolve issues faster. Even small reductions in handle time compound across thousands of conversations.

3. 3

   Optimise channel mix

   Encourage customers to use lower-cost channels for simple queries. Make self-service prominent, offer chat before phone, and reserve phone support for complex or high-value interactions. Channel migration done well improves both cost efficiency and customer experience.

4. 4

   Reduce repeat contacts

   Every repeat contact doubles the cost of resolving that issue. Investing in first contact resolution and thorough initial responses reduces the total number of conversations needed to serve the same customer base. Track repeat contact rate alongside cost per conversation.

5. 5

   Address root causes to reduce conversation volume

   The cheapest conversation is the one that never happens. Analyse your most common issues and work with product and engineering to fix the root causes. A product bug that generates 200 support conversations per month costs far more in support labour than the engineering effort to fix it.

KPI Tree lets you model support cost per conversation alongside quality and efficiency metrics, ensuring that cost optimisation does not come at the expense of customer experience. The tree can break down cost by channel, team, issue type, and customer segment to reveal where costs are highest and where optimisation opportunities exist.

Linking cost per conversation to agent utilisation rate, average handle time, and conversation resolution rate creates a complete operational picture. When cost per conversation drops, you can verify that resolution rate and CSAT remain stable, confirming that the savings are genuine efficiency gains rather than quality cuts.

Finance and support leaders can use the tree to model the impact of proposed investments. Adding a chatbot, hiring additional agents, or expanding knowledge base coverage each affect different nodes in the tree. The model makes it possible to forecast the cost impact before committing resources.