Most agencies report traffic. We report customers.

The SEO KPIs that matter are new customers won from search, and what they are worth. Everything else, including traffic, rankings and impressions, is a leading indicator at best. Traffic used to be a reasonable stand-in for customers because more visits usually meant more enquiries. With AI answers now satisfying many searches on the results page, that link has weakened, and an agency that still leads with traffic is reporting the wrong thing.

That is our stance, and the line we run Pieo on: most agencies report traffic, we report customers. Here is why, and what we report instead.

Why is traffic the wrong SEO KPI?

Three reasons, one old and two new.

Traffic never paid the bills

A thousand visits from students researching a topic are worth less than ten visits from a finance director with a problem. Traffic counts both the same. Every experienced marketer knows this, and yet traffic charts still lead most SEO reports, because they usually go up and they are easy to produce.

AI answers mean fewer clicks for the same visibility

Pew Research analysed 68,879 Google searches made by 900 US adults in March 2025. When an AI summary appeared, users clicked a traditional search result in 8% of visits. Without one, it was 15%. Links inside the summary itself were clicked in 1% of visits. Pew also found that around one in five searches in its sample produced an AI summary.

So you can be more visible, cited in the summary and named in the answer, while getting less traffic. If your agency reports traffic, that looks like failure.

Google puts AI traffic in the same bucket

Google's documentation on AI features says clicks from AI Overviews and AI Mode are counted within the "Web" search type in Search Console. You cannot separate them. That makes "organic clicks" an even blunter measure than it was.

I wrote about this problem in a post on the declining value of a click. The short version: when clicks stop tracking demand, you need to measure demand directly.

What SEO KPIs should you track instead?

We split them into three tiers. The first tier is what we lead every report with.

Traffic is still there, in tier three. It is useful for diagnosing problems. It is not the result.

How do you count customers from SEO?

This is simpler than it sounds, and it is where most reporting gives up.

1. Ask every enquiry how they found you. A required field on every form, with options including "Google search" and "ChatGPT or another AI assistant", plus free text. Ask again on the first call.
2. Tag what you can. OpenAI's publisher FAQ says ChatGPT adds utm_source=chatgpt.com to referral links. Use call tracking on key pages. Keep UTM tags on Google Business Profile links.
3. Record it in the CRM, or a spreadsheet if that is what you have.
4. Let finance count the result. Once a quarter, match won customers back to their source. Use the invoice value, not an estimate.

Step four matters most. If the agency counts its own results, you are marking its homework with its own pen.

What does this look like in practice?

On one B2B services engagement, we generated enquiries that turned into 18 new accounts over eight months, from August 2025 to March 2026. The client's own finance team counted and invoiced them, at about £40,000 (AED 194,852). That is the number we reported. Not sessions, not rankings.

On a separate site rebuild, Google clicks rose 66% and average position improved from 25 to 11, compared with the same months a year earlier. Those are good visibility numbers, and we were pleased with them. But they are tier three. They only matter if they turn into enquiries and customers.

And, for honesty: Pieo's own site gets very little search traffic today. By our own measure, what counts is whether our clients win customers.

What if my agency only reports traffic?

Ask them three questions:

• How many qualified enquiries did SEO produce last quarter, and how do you know?
• How many became customers, and who counted them?
• If traffic fell but customers rose, how would your report show it?

If the answers are vague, the reporting was designed to make the work look good, not to show whether it paid.

From our work

Here is why I distrust traffic as a headline number. A US accounting firm we looked at in September 2026 had grown its estimated search visits from 699 to 1,758 this year, which would look great in a monthly report. But most of the growth came from one article about a tax threshold, read by people checking a rule, not by businesses looking for an accountant. Meanwhile its outsourced accounting page, the service it sells, sat on page two for the searches that bring clients.

What we'd do in your first 30 days

• Week 1: agree the definition of a qualified enquiry and a won customer with your sales and finance teams. Write it down.
• Week 2: add a "how did you hear about us?" field to every enquiry route, including phone intake. Set up AI referral grouping in Google Analytics and call tracking on key service pages.
• Week 3: pull 12 months of history. Match past customers to sources where you can, even roughly, to get a baseline.
• Week 4: build the three-tier report and agree the monthly rhythm. From then on, every meeting starts with tier one.

Then the SEO and AI search work itself begins, aimed at the services and customers that the numbers say are worth winning. For law firms that means signed matters; for accountants, engagements and their annual fees.

Checklist

• Written definitions of "qualified enquiry" and "won customer"
• Source question on every enquiry route
• AI referrals grouped in analytics
• Quarterly revenue match done by finance, not the agency
• Reports lead with customers, then pipeline, then visibility
• Traffic used to diagnose, not to judge

FAQ

What are the most important SEO KPIs?

New customers from search and their value, followed by qualified enquiries. Rankings, impressions and traffic are useful leading indicators, but they are not the outcome.

Is organic traffic still worth tracking?

Yes, as a diagnostic. A sudden drop can signal a technical problem. It just should not be the headline, especially now that AI answers reduce clicks.

How do you attribute a customer to SEO when they also came through a referral?

Honestly, often you cannot fully separate them. Record every source the customer mentions. A referral who checked you in Google or ChatGPT before calling still depended on what they found.

How often should SEO KPIs be reviewed?

Visibility monthly, enquiries monthly, customers and revenue quarterly, because sales cycles in professional services rarely fit inside a month.

See what search could be worth to you

Want more revenue from Google and ChatGPT? Enter your website at our free Search Check. Tell us what one new client is worth and our team will review your site. You hear back within one working day.