How to Track Your AI Search Rankings by Hand (Free Method)

To track your AI search rankings for free, keep a fixed list of buyer prompts, ask each one in the main AI assistants on a set schedule, and record in a spreadsheet whether you appear, in what position, and who else is named. The key is consistency: same prompts, same method, same day each month. That turns random answers into a usable trend.

What “ranking” means in AI answers

There is no page one in an AI answer. Instead, you are either mentioned or not, and when you are, you may be the main suggestion or one name in a list. So a practical “rank” has three parts: present or absent, position in the answer, and whether a link to your site is shown.

Step 1: Build your prompt list

Write 10 to 20 prompts. Keep them natural and close to how buyers talk. A good mix for a yoga studio in Montreal might be:

• “best yoga studio for beginners in Montreal”
• “hot yoga classes in Plateau Mont-Royal”
• “is [studio name] good for prenatal yoga”
• “yoga studio with English and French classes in Montreal”

Save the exact wording. Changing a single word can change the answer.

Step 2: Set up the spreadsheet

Use one row per prompt per assistant per date. Columns:

1. Date
2. Assistant (ChatGPT, Gemini, Perplexity, Copilot, Claude)
3. Prompt
4. Mentioned? (yes/no)
5. Position (1 for first, 2 for second, and so on)
6. Link to your site shown? (yes/no)
7. Competitors named
8. Notes, such as wrong facts

Step 3: Use a clean method each time

• Start a new chat for each prompt so earlier messages do not influence the answer.
• Use a logged-out or fresh session where possible, as personal history can shape results.
• Note your location if the question is local, because answers can depend on it.
• Run each important prompt twice. If the answers differ, record both. Variation is normal.

Step 4: Summarise monthly

Add a small summary tab with three numbers per assistant: the share of prompts where you appear, your average position when mentioned, and how many times your site was linked. Add a line chart over the months. Three months of data is usually enough to see a real direction.

Step 5: Connect it to your analytics

In your web analytics, look at referral visits from AI domains such as chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com. Also watch searches for your brand name. These show whether visibility is turning into visits. Our page on getting traffic from Perplexity explains why some tools send more clicks than others.

Reading the answers, not just ticking boxes

Once a month, read five full answers slowly. Look at how the assistant describes you and your rivals. Words like “affordable”, “beginner-friendly” or “premium” tell you how you are positioned, and whether that matches the image you want.

Limits of manual tracking

• It takes time. Twenty prompts across five assistants, run twice, is 200 answers to read.
• It is easy to drift: different wording, skipped months, different people checking.
• It shows what happened, not why. You still need to study the cited sources and competitor pages.

For many small businesses, manual tracking is a good start. When it becomes a chore, a tool can take over the repetitive part. See Keboost vs DIY for an honest comparison.

How Keboost helps

Keboost automates this process across ChatGPT, Gemini, Claude, Perplexity, Copilot and others, compares you with competitors, and creates focused content for the prompts where you are missing. Run a free AI visibility check to compare its results with your spreadsheet.