How to Forecast a Customer Support Budget

Start from contact rate

Total volume is not forecastable because it moves with customer growth. Contacts per hundred customers is much more stable, so forecast that, multiply by projected customers, and you have a defensible volume number.

Apply cost per resolution

Once you have projected resolved issues, apply your current cost per resolution. This is why the metric matters: it converts a volume forecast directly into a cost forecast without a chain of assumptions.

Add the predictable spikes

Seasonality, product launches, billing runs, and marketing campaigns are all foreseeable. Build them in explicitly rather than treating each one as a surprise when it arrives.

Add an incident allowance

Incidents are unpredictable individually and reasonably predictable in aggregate. Look at the last two years, take an average, and include it as a line rather than absorbing it as a variance every quarter.

Bound the variable portion

If part of your cost is usage-based, a spend cap converts that portion into a known maximum. That is usually what a finance team needs to accept a usage model at all.

A worked forecast

• Current: 2,000 customers, 4.5 contacts per hundred customers per month = 90 contacts.
• Projected: 3,200 customers next year at the same contact rate = 144 contacts per month.
• At a cost per resolution of $3.86, base monthly cost = $556.
• Seasonal allowance: two peak months at 2.5x = add $1,668 across the year.
• Incident allowance based on the last two years = add $900 across the year.

Figures are illustrative and deliberately small for clarity. The method matters more than the numbers: forecast the rate, not the total, and build the spikes in explicitly.

Questions