Agency call tracking case study: a simulated example

An agency can measure clicks and forms and still not know what's happening on the phone. This example shows which questions appear when combining campaigns, call handling and sales. All data is fictional - it does not come from a CallsIQ customer or a provider test.

The business in the example provides home repairs. The cohort gathers one hundred initial calls from one hundred different contacts over a month. To simplify, there are no repeated calls. Closures are observed up to 45 days after the end of the period; The figures do not represent all the sales of the business.

Record before making decisions

The set assumes that the qualified opportunities and sales in the table come from answered calls. “Unallocated” does not mean zero cost: SEO work, call handling, branding and tools also cost money. The known advertising total is not the entire acquisition cost of the business.

First discovery: the cost per call hides differences

Google Ads and Meta Ads cost €15 per call: 600/40 and 150/10. This information could lead to treating them as equivalent channels. When looking at qualified opportunities, Google costs €30 per opportunity and Meta €50. The advertising cost per observed sale is €100 and €150, respectively.

Meta has only one sale in the example. A second sale would change its cost to €75. That is why an apparent difference does not justify cutting immediately: there is little evidence and part of the opportunities could remain open. The report must indicate the observation period.

Second finding: a call-handling problem needs investigation

80% of Google calls and 60% of Meta calls were answered. In total eighteen attempts were left unattended. This suggests reviewing schedules, diversion destination, mailbox, capacity and retries. It does not demonstrate that the eighteen calls were missed opportunities nor does it allow hypothetical billing to be assigned to them.

The agency could separate in-hours and out-of-hours care and note which calls are returned. To measure improvement you would need to compare consistent periods and consider campaign changes, seasonality and contact composition.

Third discovery: ten contacts have unknown origin

10% of calls have not been attributed in the example. The agency must investigate why: shared number, lack of link to the visit or measurement not available. You should not divide those ten contacts between campaigns to make the report complete.

I would also maintain the difference between observed source and commercial credit. If someone arrived through multiple channels, a rule of first origin is a choice of analysis, not a test of causality.

What income provides and what it does not explain

If each sale in the scenario has a value of €600 defined without VAT or returns, the attributed income would be €3,600 for Google and €600 for Meta. Your ROAS would be 6 and 4. The total of the twelve sales would be €7,200.

These ratios compare revenue and known advertising. They do not show profit: the cost of providing the service and other expenses are missing. A channel with more revenue may be less profitable if its jobs leave less margin.

The next agency experiment

The initial intervention would be to check for out-of-hours care and follow the same opportunity cohort. Before modifying large quotes, the agency would validate that each call is linked to the correct contact and that a sale is not counted twice.

To turn this exercise into a real case, reconciled exports, definitions, dates, costs and publication permission are required. Conclusions should include unfavorable and unknown results as well as any improvements. An attractive capture without that data does not demonstrate the effect of the tool.

The method is supported by the monitoring of commercial results of the CRM. Google considers the import of certain telephone conversions, with specific requirements. Importing call results .

Sources and limitations

Documentary review: 30 September 2026. Content type: Simulated educational case study.

Sources describe terms and capabilities stated by their owners. Proposed protocols and fictional examples do not establish product tests performed by CallsIQ.

• Importing call resultssupport.google.com

How to report a correction