Reading a Representative Scorecard

In this lesson

The final module pulls everything together in one document: the free scorecard. Everyone works with the same representative scorecard, so we all learn to read the same numbers the same way.

A reminder from Module 2: the free scorecard asks ChatGPT only, with 5 automatically generated questions that don't include the brand name. It can run once every 30 days for the same business. Its headline number is Discovery Visibility.

What we know about Sample Coffee Co.

Representative example, not a real measurement. Sample Coffee Co. is a fictional brand. The questions, answers, names, domains and numbers were written for this lesson.

• An independent coffee shop in Clifton, Bristol. Website: samplecoffee.example.
• Open every day, 08:00–20:00.
• Serves breakfast, but the breakfast menu on the website is only a photo, not text.
• It sells beans only in the shop; the website doesn't mention them. It has no store on the marketplace.
• No structured data (JSON-LD) on the website.

Representative scorecard: Sample Coffee Co.

Header

• Brand: Sample Coffee Co. · samplecoffee.example
• Measured: 1 October 2026
• Engine: ChatGPT
• Measured scope: Coffee shop · Bristol · United Kingdom · English · in-store · 5 questions

Headline

• Discovery Visibility: 50% · non-branded
• 2/4 answered questions found the brand
• Answer coverage 80% · 4/5

Question × AI evidence table

Simplified: the real scorecard also has an Engine column. The last column shows a short quote from the answer; the free scorecard has no source list, because the ChatGPT measurement returns no citation evidence.

Other names in AI answers · not the official SOV

Platforms and chains named in answers:

• marketplace.example · marketplace · 1/4 answers
• A national coffee chain · chain · 2/4 answers

Source and measurement coverage

Sentiment and prominence

• Sentiment: positive 1 · neutral 1 · negative 0 (across the 2 answers that mention the brand)
• Prominence: Mid-answer (average of the 2 answers that mention the brand)

Site Readiness · separate diagnostic · 3 passed · 1 issue · 0 not measured

Top 3 prioritized actions

1. Move brand evidence earlier. Why: on average the brand appears mid-answer.
2. Check your listings on these channels (marketplace.example). Why: AI answers route buyers to a marketplace. It's not a competitor; it's a place buyers can find you.
3. Fix technical entity signals (JSON-LD). This adds no points to Discovery Visibility, but it helps AI systems identify the brand correctly.

Reading it section by section

Header. Start with what was measured: one engine (ChatGPT), 5 questions, one date, one scope. Every number on the scorecard holds only within that frame.

Headline: Discovery Visibility, not SOV. 50% means Sample Coffee Co. came up on its own in 2 of the 4 answered questions. It isn't an SOV Score. SOV needs a confirmed competitor set and the weekly five-engine measurement, and this scorecard has neither.

x/y and answer coverage. "2/4 answered questions found the brand" is visibility. "Answer coverage 80% · 4/5" shows how many of the 5 questions got a usable answer. That second figure isn't visibility. Question 5 wasn't measured and stays out of the denominator, so the right calculation is 2 ÷ 4 = 50%, not 2 ÷ 5 = 40%.

Evidence table. Each row is one observation. "Absent" is a measured result: an answer came back and Sample Coffee Co. wasn't in it. "—" means there was no answer to check. Prominence only exists where the brand was found.

Other names: not the official SOV. Café A appearing in 3 of 4 answers is an interesting signal, but none of these names has been confirmed as a competitor. Use Module 6's four boxes. The marketplace is a channel. The national chain is a major player, though it could be a direct competitor if it has a branch in Bristol.

Sources: not measured on the free scorecard. The "Answers with citations" and "Owned-site citations" rows show no number, just "Not measured". Here's why: the free scorecard measures ChatGPT only, and that measurement returns no structured citation evidence. We can't reliably tell which sites were cited, so we don't calculate these rates. That's also why the last column of the evidence table shows a short quote from the answer, not a source list. Even when a site name appears in a quote (like marketplace.example in answer 4), it isn't citation evidence; it's a name in the answer text.

"Not measured" isn't 0%. It's the same rule as in Module 2: an unmeasured value isn't counted as zero. "Owned-site citations: Not measured" doesn't mean your site was never cited; there's simply no evidence to measure. This is the third case from Module 3, Lesson 2: with no citation evidence, the answer stays out of the denominator and isn't counted as zero. Sources and owned-site citations are measured in paid multi-engine tracking, on the engines that provide citation evidence.

Site Readiness: a separate diagnostic. "3 passed · 1 issue" is a technical checklist. It adds no points to Discovery Visibility or SOV. A missing llms.txt isn't counted as an issue.

Common misreadings

• "Sample Coffee Co.'s SOV is 50%." Wrong. 50% is Discovery Visibility.
• "Visibility is 40%, because we appeared in 2 of 5 questions." Wrong. The unmeasured question 5 stays out of the denominator.
• "Answer coverage is 80%, so our visibility is 80%." Wrong. Coverage shows whether answers came back.
• "We were absent from question 5." Wrong. Question 5 wasn't measured, so it's unknown.
• "Owned-site citations are not measured, so our site was never cited." Wrong. Not measured isn't 0%; no citation evidence is captured in the ChatGPT measurement.
• "Site Readiness looks good, so we're doing well in AI." Wrong. They're separate measurements.
• "Café A is our strongest competitor." Not yet. It's an observation; you decide with the four boxes whether it's a competitor.

More: free scan rules, Discovery Visibility definition and Owned-Site Citation Coverage.

Summary

• The free scorecard is one engine (ChatGPT), 5 non-branded questions, one date and one scope.
• The headline number is Discovery Visibility, not SOV. In the representative example: 2 ÷ 4 = 50%.
• "x/y answered questions" is visibility; answer coverage (4/5) shows whether answers came back. An unmeasured question stays out of the denominator.
• Other names in the answers aren't an official SOV; sort them with the four boxes.
• On the free scorecard, "Answers with citations" and "Owned-site citations" aren't measured, because the ChatGPT measurement returns no citation evidence. Not measured isn't 0%; sources and owned-site citations are measured in paid multi-engine tracking, on the engines that provide citation evidence (Module 3, Lesson 2).
• Site Readiness is a separate diagnostic and adds nothing to visibility.