Which marketing numbers matter monthly, and which are vanity
Years ago I helped a business multiply its website traffic a hundredfold. The chart from that engagement is still the prettiest I have produced. Month after month, the line climbed.
The business changed far less than the line implied. Visitors arrived in crowds, read, and left. The revenue moved a little. The chart moved a lot. I learned the difference between the two the slow way, and it reshaped what I report.
Traffic is an ingredient. For years the industry has reported it as the meal.
Traffic measures less than it used to
Even as an ingredient, traffic now measures poorly, because the supply of clicks is shrinking underneath it.
SparkToro's 2026 analysis of US clickstream data found that 68% of Google searches end without a click to anywhere. Fewer than one search in three now sends a visit to any website. When an AI summary sits on the results page, Pew found clicks fall by nearly half, from 15% of visits to 8%.
Meanwhile the influence that matters most leaves no trace at all. A chatbot names you in an answer, the buyer reads it, and no click occurs anywhere. When that buyer shows up two weeks later, your analytics files the visit under "direct."
So a traffic report now works like a broken scale. It undercounts the influence you earned and overstates the progress you made. A number that behaves that way should not lead a monthly review.
The four numbers to watch monthly
The numbers that deserve the monthly look are the ones the business runs on. Four cover it for most expert practices.
- Consultation requests: how many arrived, and how many were a fit. Ten requests from the wrong buyers is a worse month than three from the right ones.
- Meetings: how many conversations were held, and how many proposals went out.
- Replies: responses to your emails, messages about a piece you published, a prospect quoting your article back to you. Replies are demand announcing itself early.
- What each new inquiry says brought them in.
That fourth number deserves its own habit. Put "what brought you to me?" on the intake form as an open text field, and ask it again on every first call. Skip the dropdown menu, because a menu only returns the channels you already believed in. Written answers surface the podcast mention, the forwarded newsletter, and the chatbot recommendation that tracking software cannot see.
Memories are imperfect, so read these answers as direction rather than precision. Over a year, the direction is worth more than any dashboard I have seen.
Every number on this list can be counted in an evening without software. Each one also forces a decision. Requests up while meetings stay flat points at the intake process. Replies up while requests stay flat means attention is arriving ahead of demand, and demand numbers move on a slower clock than presence numbers, in quarters rather than weeks.
The glance column
Traffic, rankings, impressions, followers, and open rates still earn a place, one level down. They are direction and diagnosis. When a primary number stalls, this column tells you where to look. When this column gets reported as the outcome, the report has failed.
Two newer lines belong in the glance column now.
The first is chatbot presence. Once a month, ask ChatGPT and Claude ten questions your buyer would ask, and note whether you appear among the sources. Twenty minutes, and you are watching the surface where a growing share of evaluation happens.
The second is AI referrals. Google Analytics added an AI channel in 2026, though much of this traffic still lands in "direct," so treat the count as a floor. The column stays small and matters anyway. Ahrefs ran the numbers on its own site: AI assistants sent 0.5% of visitors and produced 12.1% of signups. By the time a chatbot's reader clicks, the comparison shopping is already done. A small column of visitors like that is one more reason a raw traffic total misleads.
Why the dashboard celebrates the wrong column
The standard reporting follows incentives rather than malice.
Platforms report what they can count, and clicks are countable while conversations are not. Providers report what their work visibly moves, and a campaign moves impressions long before it moves requests. Both incentives point at big numbers, so big numbers fill the reports, and owners learn to nod at charts that predict little.
The buyer-side test is simple to apply. A monthly report should end in decisions: what gets more effort next month, what gets less, and why. If a number cannot change what you do next month, it does not belong in the report.
The short arithmetic of content ROI
The industry publishes ROI figures for content, and most trace back to people selling content. Skip them. For a high-ticket practice, the arithmetic fits on an index card.
Take what one client is worth to you across an engagement. Take what the marketing engine costs you for a year. Divide. Suppose a client is worth $30,000 and the engine costs $24,000 a year: one client clears it, and the second is margin. For most expert practices the whole return question comes down to one or two clients a year. That is why the measurement problem is smaller than the software vendors need it to be.
You do not need attribution modeling to notice two clients. You need to ask each one what brought them in, and write the answer down. The arithmetic assumes those quarters are spent on work with a compounding record, and it explains why a practice can win while its traffic chart stays modest.
The evening a month
The whole review fits in an evening. Four numbers counted: requests and their fit, meetings, replies, and what new inquiries said brought them in. Ten buyer questions run through the chatbots. A glance at the direction column. Then the only part that matters, the decisions.
The prettiest chart I ever produced taught me the order of operations. Count what the business runs on first. Glance at the rest.
