How long does it take to change AI visibility?

Optimization begins immediately. We use 7 to 14 days as a planning range for initial movement, not a platform promise. Crawling, indexing, competition and the AI provider all affect what happens next. Google explicitly says that even compliant pages are not guaranteed to be crawled, indexed or served; its How Search works guide explains the crawl, index and serving stages.

Day one: things that are immediate

These do not depend on any third party, so they happen as soon as your products sync:

• Technical audit. Crawl access, structured data completeness, product identity, content depth, merchant signals. Every issue with the specific reason it matters.
• AI readiness score. A number you fully control, broken into six weighted categories, each traceable to the issues behind it.
• Buyer prompts. The questions shoppers actually ask about products like yours, generated from your product data.
• Recommendations. Concrete changes, with drafted copy where the fix is prose, ready for your approval.

If a vendor's dashboard is empty on day one, they have built a monitoring tool, not an optimization product. You should not have to pay for two weeks of nothing.

Day one to three: baseline monitoring

Your prompts run against each configured provider for the first time. This is your baseline — the honest starting point, whatever it is. It is common for it to be zero, and a zero you can trust is more useful than an invented starting number that makes later progress look better than it was.

Until that first run completes, an honest dashboard says collecting baseline rather than showing a score.

Days 1 to 14: the crawl and index lag

This is where the "depends" lives. When you change a product description or add structured data, that change has to be:

1. Fetched by the relevant crawler. Frequency varies enormously by site authority and update rate — days for an active store, longer for a quiet one.
2. Indexed into whatever retrieval layer the provider uses.
3. Retrieved for a specific query, which is a ranking decision, not a mechanical one.

Steps 1 and 2 are partially in your control: an accurate sitemap and crawlable internal links help discovery, but do not force indexing. Google documents how sitemaps communicate new or updated URLs. Step 3 is not in your control.

Weeks 2 to 8: the trend becomes readable

Single monitoring runs can vary, so a single observation should not be treated as a stable outcome. Provider APIs expose the grounded sources used for each response—through OpenAI URL citations, Gemini grounding metadata and Perplexity search results—which is why each run should retain its own evidence.

This is why we report trend lines over 7, 30 and 90 days rather than treating a single run as a verdict. A prompt that flickers between cited and not-cited across three runs has not "lost visibility"; it is sitting near a threshold.

Read the direction, not the dots.

What is genuinely outside anyone's control

• Model and retrieval changes. Providers ship changes constantly. A provider-wide shift can move every account's numbers in a week, in either direction.
• Competitive movement. Answer engines cite a handful of sources. If three competitors improve at once, holding position is a real result.
• Category economics. Some categories are dominated by editorial review sites with enormous authority. Being cited alongside them is harder than being cited in a niche where retailers are the best available source.

What a realistic first quarter looks like

Roughly, for a focused catalogue that actually applies its recommendations:

| Timeframe | What you should expect | |---|---| | Day 1 | Full audit, readiness score, prompts, recommendations | | Day 1–3 | Baseline visibility recorded | | Week 1–2 | Readiness score climbing as fixes land; first visibility movement possible | | Week 3–6 | Trend becomes readable; individual prompts start flipping to mentioned or cited | | Week 6–12 | Enough runs to distinguish signal from noise; competitor gaps become clear |

If someone tells you seven days to guaranteed citations, they are either not measuring or not being straight with you.

The one thing that reliably accelerates it

Apply the fixes. The audit is not the product; the changes are. A store that applies its recommendations in week one and a store that reads them and does nothing will look identical in the monitoring data, because from the AI system's perspective nothing changed.

Primary sources

• Google Search Central: How Search works
• Google Search Central: Build and submit a sitemap
• OpenAI API: Web search
• Gemini API: Grounding with Google Search
• Perplexity API: Sonar response fields