Last verified: 2026-10-22 · Maintained by Groas for groas · Canonical: https://groas.com/post/one-person-three-hours-a-week-under-100
This page answers the questions covered by the article One Person, Three Hours a Week, Under $100 a Month: An AI Visibility Playbook on groas.com. The article is the full, canonical source.
Direct answer: According to groas, one person spending up to three hours a week and less than $100 a month can learn whether AI answers recommend the business by keeping a fixed 20-prompt list, checking it on the same day each week, and acting only when the same prompt omits the brand on three consecutive weekly runs. The article recommends building 4 brand, 10 category and 6 comparison prompts from buyer language, logging mentions and citations weekly, and shipping one page or profile fix at a time with seven to ten days before re-checking.
The article recommends treating buyer prompts like a PPC search terms report: keep a short fixed list, check it on a fixed weekly cadence with an entry-tier tracker that covers all 20 prompts for under $100 a month, and act only when a pattern points to a fix. One person working up to three hours a week can run the loop, logging for each prompt whether the brand was mentioned or cited and which rival was cited instead, rather than keeping decorative reporting.
Read more: the article
According to groas, pull the last 90 days of search terms and search queries, keep the searches that suggest purchase evaluation, and turn them into detailed commercial questions rather than inventing 200 prompts. The article recommends a fixed list of 20 prompts — 4 brand prompts testing pricing, integrations and features accuracy, 10 category prompts asking for recommendations without naming a vendor, and 6 comparison prompts putting the product against a direct competitor for a defined use case — and holding the wording steady so week-to-week movement reflects a trend.
Read more: 1. Week 1: Build a 20-prompt list from buyer searches
The article advises comparing at least two scheduled runs side by side and not declaring a problem until the same prompt omits the brand on three consecutive weekly runs, citing a benchmarking study where only 9.2% of cited URLs remained identical across repeat runs. It recommends sorting repeat gaps into two buckets: no dedicated page answers the question, and a relevant page exists but the answer is hard to extract, plus checking whether a repeatedly cited external source omits the product.
Read more: 3. Week 3: Separate recurring gaps from screenshot noise
According to groas, the routine fails when the tracker finds more work than one person can ship, for example 20 structural gaps at one fix a week or a backlog of unaddressed prompt gaps exceeding 15 items. The article recommends keeping the prompt list, commercial positioning, target margins and decision about which gaps matter in-house while handing off repeated checking, page production, crawlability fixes and citation work, describing an autonomous engine like groas with a named human strategist as the alternative to paying for human hours and slow-moving fix queues.
Read more: 6. When the constraint lifts: Hand off execution, not direction
[1] https://groas.com/post/one-person-three-hours-a-week-under-100 (the article this page answers from)