Ten to twenty minutes and up to 250 words each, on the fictional Kalvenor sample (Sample data, fictional company) and on category level Absence Index data. Pick one, write your answer.
Exercises
PracticeFictional company15 minutesUp to 250 words
Fictional Kalvenor: choose three checks
Using the fictional Kalvenor sample, choose three checks you would complete before recommending a change. For each, name the visible evidence you would inspect and one result that would change your next step. Separate an observation from a possible explanation.
PracticeFictional company15 minutesUp to 200 words
Fictional Kalvenor: investigate one omission
Open the fictional Kalvenor sample and choose one buyer question where Kalvenor Systems is not named. State the question and the engine view you inspected. Give two plausible explanations, one piece of evidence that would distinguish them, and one conclusion the sample cannot support. Do not infer cause from an omission alone.
The construction project management release checks 22 vendors. Its highest observed naming count is 109 of 150 valid answers (72.7%, Wilson 95% interval 65.0% to 79.2%). Observed answers on stated dates, not a ranking. This is not a judgment of product quality. Name two checks needed before interpreting an observed absence, and write one sentence that a buyer could safely repeat. Do not identify absent vendors or judge products.
In a hypothetical exercise, a company is named in 12 of 150 valid answers and then 19 of 150 a week later. Give two checks needed before comparing the runs. Explain why overlapping intervals alone cannot settle the question, propose a suitable comparison if the same questions were repeated, and write a cautious update for a CMO.
Choose one public, dated claim about AI search visibility and link the original source. Identify the observation, the inference and any missing evidence. Propose one fair test that could weaken or support the claim. Critique the claim, not the author, and avoid confidential information.
In a hypothetical case, a software company is named by ChatGPT but not by Claude, Perplexity, Google AI Overviews or Google AI Mode. Before proposing a change, identify one missing fact. Then propose a small test, a comparison, a fixed observation window and a result that would make you abandon your explanation.
PracticeFictional company15 minutesUp to 220 words
Reconstruct one receipt
Choose one answer in the fictional Kalvenor sample. Write a receipt another reader could use to locate exactly the same view: question, engine, date, answer excerpt and cited source. Identify one field that the view does not reveal. Do not fill that gap with a guess.
PracticeFictional company10 minutesUp to 180 words
Citation is not endorsement
Find one cited source in the fictional Kalvenor sample. Write three separate statements: what the answer names, which source it cites, and what you cannot infer about why that source was used. If the view cannot support one statement, say so.
This category's highest observed naming count is 120 of 149 valid answers (80.5%, Wilson 95% interval 73.4% to 86.1%). Observed answers on stated dates, not a ranking. This is not a judgment of product quality. Explain why replacing the valid base with the number of planned answers would change the reported result. State what you would inspect before calling an excluded answer a technical failure or an omission.
The highest observed naming count in this category is 64 of 150 valid answers (42.7%, Wilson 95% interval 35.0% to 50.7%). Observed answers on stated dates, not a ranking. This is not a judgment of product quality. A colleague wants to compare it with another category to declare one market easier to influence. Explain two reasons this conclusion does not follow and propose a narrower, testable question.
Choose one public claim that a content change improved AI search visibility. Link its original dated source. State what changed, what was measured and which alternative explanation is strongest. Describe the smallest extra evidence that would improve your confidence. If the source makes no causal claim, explain that instead.
The highest observed naming count is 113 of 150 valid answers (75.3%, Wilson 95% interval 67.9% to 81.5%). Observed answers on stated dates, not a ranking. This is not a judgment of product quality. Write a short limits paragraph for a reader considering a purchase. Keep the figure intact if you use it. Explain what this observation can and cannot tell them.
In a hypothetical answer, an engine says, "Some teams consider Example A, but it does not fit the requested use case." Example A is fictional. One reviewer records a mention; another records a recommendation. Propose separate fields and an adjudication note that preserve what was actually said. Explain how two reviewers would use your rule consistently.
Read the published buyer questions. Pick one buyer need the set covers and one plausible need it might not cover. Propose one additional question and explain how you would report the expanded set without presenting it as directly comparable with the original release. Do not identify an absent vendor.
PracticeFictional company20 minutesUp to 250 words
Design a repeatability check
Using one question from the fictional Kalvenor sample, write a plan for collecting comparable answers next week. State what remains fixed, what is recorded when it cannot remain fixed, and how failed captures are handled. Name one result that would make the comparison inconclusive.
The highest observed naming count is 126 of 150 valid answers (84.0%, Wilson 95% interval 77.3% to 89.0%). Observed answers on stated dates, not a ranking. This is not a judgment of product quality. A hypothetical headline claims the category's answers prove which product buyers should choose. Replace it with an accurate headline and a two-sentence explanation. You may omit the number instead of abbreviating away its context.
I'd first establish what the software does, who buys it and what problem sends that buyer looking for alternatives. That tells me which question to test and which companies belong...
Check 1: Margin Trend Verification. Observation: Inspecting the trailing 12-month gross margin breakdown in the category audit report, I see a 4% drop in product-line...
Question: “How can a service business quote and invoice from the field on the same day?” Engine view inspected: ChatGPT view for this question. Kalvenor Systems is marked as not...