# Case Study: A 90-Day Plan for a SaaS Startup — AI Visibility and LLM Brand Discovery

Source: https://www.geekswithgeeks.com/en/ai-visibility/wrap-90-day-case

> Plan the first 90 days for Acme Tasks, a small SaaS product that is rarely mentioned by assistants.

## Plan

**Days 1-15 (baseline)**: build a 60-prompt set in English and Hindi; run it 3 times on 3 assistants with search on and off; record mention rate with ranges, citations, accuracy and sentiment; check robots.txt, server logs and rendering with `curl`. **Days 16-45 (fix)**: unblock search crawlers deliberately, make pricing and docs server-rendered, add Organization/Product/FAQ JSON-LD, create a brand facts page, correct two stale directory listings and one outdated blog post. **Days 46-75 (create)**: publish pricing, three honest comparison pages, five how-to guides, an FAQ and a Hindi pricing and FAQ page; release an original survey. **Days 76-90 (earn and measure)**: ask happy customers for reviews, contribute to two communities with disclosure, pitch the survey to trade sites, re-run the prompt set, report with ranges and decide next quarter's focus.

## A complete 90-day plan

Fix the basics, publish what assistants need, earn credible mentions and measure honestly.

![Four phases: baseline, fix, create, earn.](assets/figures/ai-visibility/section-8-map.svg) — Figure 8.1 — Baseline, fix, create and earn.

## The plan on one page

Each line maps to a section of this course.

```text
Baseline    60 prompts (en+hi), repeated runs, ranges, logs, curl   (Sec 4, 3)
Technical   crawler rules, server-rendered pages, JSON-LD, sitemap       (Sec 3)
Facts       brand facts page, fix stale sources, correction log         (Sec 2, 6)
Content     pricing, comparisons, how-tos, FAQ, Hindi pages, survey     (Sec 2, 5)
Authority   honest reviews, disclosed community help, trade-press pitch (Sec 5, 6)
Operate     monthly cycle, change log, honest scorecard                 (Sec 7)
```

## Compare like with like

Re-run exactly the same prompts, assistants and settings at day 90. Otherwise a change in the questions, not in your visibility, might explain the difference.

**Quiz:** What should happen at day 90 so results are comparable with the baseline?

- [ ] Skip measurement
- [ ] Use completely new prompts
- [x] Re-run the same prompts, assistants and settings
- [ ] Change the assistant each time

*Answer:* Re-run the same prompts, assistants and settings. A like-for-like comparison isolates the effect of your changes.