What SEO prompts can I use with ChatGPT?
Ask ChatGPT to separate the product claim, its billing qualifiers and the dated reference before using the material in an SEO (search engine optimization) comparison. Our Wrong About You study found 2 wrong statements among 41 Jobber statements, using 22 September 2026 captures and an 8 October reference check. The tasks below describe factual checks grounded in that example; no new test of their effect was run.
Observed answers on stated dates, not a ranking.
What we measured
Measured. The published statement-review method preserves two different Jobber pricing propositions. The Core starting-price statement recurs in two Claude returns; an Overview return contains the other pricing-range claim. The forty-one-statement row also includes twenty-nine true and ten unverifiable entries, with no stale entries. Team review: in progress. The dataset does not test whether these proposed prompts improve accuracy or SEO performance.
The retained historical Claude statement, captured on 22 September 2026 and marked wrong against the dated vendor reference, reads:
Jobber Core starts at $39/month with clean client management
This is the disputed historical statement, not a current vendor price. Team review: in progress.
Reasoned. The retained plan name, billing language, capture date and pricing-page reference give a proposed research prompt specific fields to ask about. A missing field can remain unknown rather than being supplied from the generated answer's confidence.
What this means for you
Begin with the original sentence. Separate the Jobber Core price claim from the retained answer, including its billing period and included users. That task concerns the Core entry, rather than every product named elsewhere in the heating, ventilation and air conditioning (HVAC) answer. Keep the original claim alongside the supplier reference so a reader can check both.
For the source comparison, use the retained 8 October pricing-page extract. Separate the monthly display from the annually billed amount and identify the user allowance shown there. The dated extract is the input; this does not assert that the October terms remain current. If an amount has no clear billing period, leave that field unresolved.
Keep the dates explicit. A correction note should identify the 22 September answer and the 8 October reference separately, then explain only the disagreement those records support. The other Jobber entry concerns a range in a Google AI Overviews answer. It needs its own claim text rather than silently inheriting the Core note.
For the report, two repeated Claude observations remain one Core statement. The ten unverifiable statements remain a separate class. Neither repetition nor an unresolved source turns a statement into an additional confirmed error.
Also asked as
What SEO prompts can I use with ChatGPT?
chatgpt seo prompts?
Related
- Wrong About You claim and reference records
- The public study record
- AI Source Standard report definitions
- The Manual chapter on valid measurement
- More geo basics answers
- Relevant vendor record in the Index
Next step
Free 10-question check (one engine). No email is needed; runs are limited per network and per day, and by the site's daily allowance.
Claiming or correcting a vendor record is free and never depends on a purchase.
Our answer-study method, v1.1.
Answer Repair: Fix one answer, $490.
Prepared by the Broadcastwell team.
Page owner: Priya Nair, Management Consultant. Corrections: hello@broadcastwell.com