Which of your traffic sources actually produces work

Channels are usually compared by how many visitors they send, which is the least informative comparison available. A small source sending people who buy is worth more than a large one sending people who read.

Talk to us about your site

Compare channels by enquiries and work won per visitor rather than by visitor count. Search, referrals, map listings, social and AI assistants behave very differently, and the largest is rarely the most valuable. Expect a large unattributed share and treat it as a gap rather than a result.

Volume is the wrong basis for comparison

A channel that sends a flood of visitors who never enquire is a cost, not an asset, however impressive the number looks on a chart. A channel sending a modest stream of people who become customers is the opposite.

So the first change is to rank channels by what they produce. That single reordering usually rearranges a marketing budget more sensibly than any strategy document.

How the main channels tend to behave

Patterns vary by business, but some tendencies are consistent enough to expect, and they explain a lot of confusing data.

  • Search for a specific service and place: fewer visitors, high intent, usually the best.
  • Map listings for local services: often produce calls rather than form enquiries, and are easy to undercount.
  • Referrals and recommendations: small numbers, high conversion, frequently checked against the website first.
  • Broad informational search: large numbers, low immediate intent, valuable as an introduction rather than a closer.
  • Social platforms: variable, and usually browsing rather than buying.
  • AI assistants: growing, hard to attribute, and often arriving already convinced.

The undercounting problem

Two channels are systematically undercounted in most reports. Map listings, because the resulting call never touches the website. And AI assistants, because they frequently pass along no source information at all.

Both then look weaker than they are, and businesses shift money away from them on the strength of a measurement artefact. Asking customers directly is the only reliable correction.

Judge introducers and closers differently

Some channels bring people who have never heard of you; others convert people who already had. Comparing them on the same basis undervalues the introducer every time.

If you record what customers say first brought them alongside what your analytics saw last, you can see both roles. A channel that appears often as a first contact deserves credit even when it rarely appears as a last one.

Deciding what to do with the answer

Where a channel produces good customers, the question is whether more of it is available. Some are capped: there are only so many searches for your service in your town, and reaching all of them is a ceiling rather than a growth plan.

Where a channel produces nothing after a fair trial, stop, and be honest about how long a fair trial is. Search visibility takes months; a social account takes longer; advertising can be judged faster.

  • Ask whether more of a good channel is actually available.
  • Give each channel a fair trial length before judging it.
  • Move money toward what produces work, not toward what produces visits.
  • Recheck quarterly rather than monthly.

How Licheo compares channels

We rank by enquiries and work won rather than by traffic, and we combine what the tools report with what customers say when asked directly.

Where a channel cannot be measured reliably, we say so rather than assigning its results to whichever source happens to look convenient.

Part of a larger guide

This page is one part of Revenue per visitor. The other parts:

Questions people ask

How do I know which traffic source is best?
Rank sources by enquiries and work won per visitor rather than by visits. The largest source is frequently not the most valuable, and the ranking usually changes once value is included.
Why do map listings look weak in my analytics?
Because the resulting call often never touches the website, so nothing is recorded. For local services this is one of the most undercounted channels, which is why asking callers how they found you matters.
How do I measure traffic from AI assistants?
Imperfectly. Many pass along little or no source information, so those visits land in the unattributed bucket. Treat a growing unattributed share as a signal and ask customers directly to fill the gap.
Should I stop investing in channels that do not convert?
After a fair trial, and after checking whether the channel is an introducer rather than a closer. Something that regularly appears as the first thing customers remember deserves credit even if it rarely appears last.
How often should I review channels?
Quarterly. Monthly reviews mostly capture noise and seasonality, and reacting to those produces a series of reversals rather than a direction.

Put the money where the customers are

Channels ranked by what they produce, with the gaps in measurement named rather than hidden.

Talk to us about your site Read the full guide