Which Marketing Channel Actually Brings Customers?

A channel can create many views, messages or website visits while bringing few customers the business can serve. A referral may produce fewer enquiries but better matched work. The practical question is not which platform has the biggest number on its dashboard. It is where suitable customers come from, how much effort it takes to serve them and how uncertain the evidence is.

Choose a decision before collecting data

Perhaps you are deciding whether to renew a local directory listing, spend another month on a small ad or allocate staff time to a community event. Choose two or three relevant channels and a period long enough for customers to decide. Define a qualified enquiry: a distinct request that fits your offering, location and capacity. Decide what counts as a win: a confirmed booking or purchase, not a like or an unconfirmed appointment.

The SBA marketing and sales guide recommends tying marketing actions to a target market and goals. The Australian government's guidance on measuring digital performance explains how to choose measures that reflect the outcome you want. Both support starting with the decision and the customer rather than measuring everything an analytics tool can display.

Use one simple record

Channel	Hours	Spend	Qualified enquiries	Wins	Approximate booked value
Local directory	2	120	8	2	800
Customer referrals	3	0	5	3	1,050
Social posts	6	0	6	1	250

Illustrative assumptions: The figures describe one small repair business during a four-week period, in arbitrary currency units. “Hours” includes preparing and handling the activity, not delivery of paid work. Booked value is sales before service costs and is not profit. One unusual large job or a delayed decision could reverse the apparent ranking; do not treat these sample results as typical performance.

Collect the source without interrogating people

At first contact, ask “How did you hear about us?” once, as an optional question. Record the answer as given, including “friend and website,” “I do not remember” or “unknown.” Use a consistent dropdown only if it leaves room for free text. Someone may discover a shop through social media, see a review and later search for its name. Assigning all credit to the last click hides that path.

The Google Analytics explanation of user versus traffic acquisition distinguishes where a person was first acquired from the source of a later session. Even when you use web analytics correctly, it does not see every phone call, in-person conversation or offline referral. The simple customer question also depends on memory. Treat channel labels as approximate evidence, not a perfect assignment of credit.

Compare more than sales totals

In the example, referrals appear to bring three wins from five suitable enquiries, while the directory brings two from eight. But the referral activity used three hours and may depend on earlier service quality rather than a repeatable marketing action. The social posts generated one job and six hours of work. Before moving time or money, inspect the jobs: were they the sort you want more of, and did they cover direct costs and available capacity?

Record cash spend and staff time separately. An unpaid channel is not free if it absorbs the owner’s afternoon. A paid campaign may also generate future work whose result has not yet been observed. Compare a consistent window and make the decision that the data can actually support: perhaps pause one small directory renewal, improve the service description in the listing or test a more focused social post. Avoid claiming a precise return on investment from a handful of incomplete records.

Run a modest test

Suppose the repair business thinks its directory listing attracts requests outside its area. It changes only the service-area wording for the next month and records enquiries, fit and wins. It keeps spend steady. If the share of suitable requests improves without a sharp fall in total suitable enquiries, the change may be useful. If results are mixed, inspect individual messages and extend the test. A season, local event or competitor’s promotion can affect the comparison.

Agree on a cutoff for pending enquiries. The person who called yesterday has not had the same time to decide as someone who contacted you a month ago. Mark uncertain outcomes and revisit them rather than treating them as losses. Where the purchase cycle is long, use a longer window and fewer confident conclusions.

Common measurement traps

• Comparing raw clicks to orders without checking whether the visitors wanted the service.
• Counting the same person twice when they call and then email.
• Forcing one source when the customer names several influences.
• Using booked value as profit or ignoring staff effort.
• Changing the offer, channel and price together, then crediting one of them for the outcome.

If the website uses analytics or advertising tags, check the relevant privacy and consent requirements where the business and customers are located. A large ad commitment, complicated attribution setup or unclear supplier contract may warrant specialist review. A small shop can learn a great deal from a consistent enquiry log before buying sophisticated software.

If many qualified contacts do not buy, see Why Inquiries Do Not Become Customers. If quotes are going unanswered, see How to Follow Up on a Quote Without Being Pushy.

Distinguish the source from the decision

A source label tells you where a conversation may have started; it does not tell you which part of the offer convinced the customer. If several channels are involved, write that down. You can choose a consistent reporting rule for a simple comparison, such as “customer-stated first source,” but keep a second note about later influences. Never present the selected rule as a complete account of why the sale happened.

Some channels reach different customers or sell different kinds of work. Compare a local directory for urgent repairs with social content about scheduled maintenance only after you check that the jobs and decision windows are reasonably similar. If they are not, evaluate each against its own objective and available staff time. An expensive channel that fills an otherwise empty weekday may be useful even if its cost per enquiry is higher; another channel may create more leads than the team can deliver.

Before cancelling a channel, inspect whether the listing, message or service availability is the issue. A small, reversible wording test can provide a cleaner answer than ending the activity after one unusual week.

Next step: Record the source, fit, time, spend and outcome for your next ten distinct enquiries. Include “unknown,” then compare two channels using suitable enquiries and confirmed work.