An owner's case study, not a customer result
goldprice.dev is our own site. We chose the work, built the pages and read the dashboards ourselves. Nothing here comes from a SearchSeal customer, and we are not presenting it as one. We publish it as our clearest example of the method in use, backed by real numbers we can stand behind.
Finding the searches worth winning
The gold price API market has well known names already. Buyers search for them directly: goldprice.org, GoldAPI, Metals.dev and the gold endpoints of Twelve Data. A smaller site will not outrank those brands on their own names by accident, but it can earn a place on page one beside them when someone is comparing options.
We began with the category’s top sites and the queries people use to find them. For each query we asked three questions. Does the search mean the person is choosing a provider? Is page one made of pages we could match or beat on facts, such as free tiers, limits and coverage? Do we already get some impressions for it, which shows the search engine sees us as relevant? Passing all three made a query winnable. Queries that only wanted the rival's own login page were left alone.
That pointed at two groups: rival-name searches, such as a competitor's brand with "api" or "free tier" added, and plain API searches, such as gold price API in another language. The same pattern is what SearchSeal looks for in AI answers: the questions where a rival is named and you are not.
What we built and how it was indexed
We wrote comparison pages, alternatives pages and roundups for those queries. Each one answers the question the searcher actually has, for example what a rival's free tier includes and how goldprice.dev differs, with the facts in plain text and tables that both search engines and AI tools can read.
We rolled out indexing in paced batches from 12 Oct 2026, instead of all at once. Each batch went through Google Search Console, the Bing sitemap and IndexNow submissions.
The observed numbers
We checked all figures below on 18 Oct 2026 in read-only views. Each row names its source and window, because the windows are not all the same.
| Metric | Observed value | Source | Window |
|---|---|---|---|
| Google clicks | 444 | Google Search Console, web search | 18 Sep to 15 Oct 2026 |
| Google impressions | 14.3K | Google Search Console, web search | 18 Sep to 15 Oct 2026 |
| Google average position | 7.7 | Google Search Console, web search | 18 Sep to 15 Oct 2026 |
| Queries with 10+ impressions averaging position 10 or better | 41 of 58 | Google Search Console, web search | 18 Sep to 15 Oct 2026 |
| Bing clicks | 1,810 | Bing Webmaster Tools | 18 Sep to 15 Oct 2026 |
| Bing impressions | 101,219 | Bing Webmaster Tools | 18 Sep to 15 Oct 2026 |
| Referral visits from ChatGPT | 128 | Site AI-traffic log | 20 Sep to 18 Oct 2026 (WIB) |
| OpenAI search-bot requests | 4,103 | Site AI-traffic log | 20 Sep to 18 Oct 2026 (WIB) |
| ChatGPT user-triggered page fetches | 3,661 | Site AI-traffic log | 20 Sep to 18 Oct 2026 (WIB) |
| Perplexity search-bot requests | 2,339 | Site AI-traffic log | 20 Sep to 18 Oct 2026 (WIB) |
| External signups attributed to AI assistants | 64 of 460 (13.9%) | Signup UTM and referrer | 20 Sep to 18 Oct 2026 (WIB) |
In Google, 41 of the 58 queries with 10 or more impressions averaged position 10 or better. The example queries below are among them. Positions are averaged over 28 days; a query at 6.9 appeared both higher and lower. The only query where goldprice.dev averaged a top three position was its own brand name, goldprice.dev, at 1.1.
| Query | Impressions | Average position |
|---|---|---|
| goldprice.org | 573 | 6.9 |
| goldprice.org api | 332 | 6.6 |
| goldapi | 176 | 8.5 |
| goldapi.io free tier | 102 | 6.1 |
| twelvedata gold price api | 57 | 7.4 |
| metals dev | 40 | 5.9 |
| api giá vàng (Vietnamese for gold price API) | 37 | 5.2 |
Source for the query table: Google Search Console, web search, 18 Sep to 15 Oct 2026.
Bing was the larger channel in the same window, with 1,810 clicks from 101,219 impressions.
AI assistants and signups
The site logs its own AI traffic. Over 28 days to 18 Oct 2026, it recorded 128 referral visits from ChatGPT. It also recorded 4,103 requests from OpenAI's search bot, 3,661 page fetches triggered by ChatGPT users and 2,339 requests from Perplexity's search bot. Crawls and fetches are retrieval: the assistant reading a page. They are not citations and they are not visits from people.
In the same window, 460 external accounts signed up. We identified 64 of them, or 13.9%, as coming from AI assistants: 62 from ChatGPT and 2 from Google AI. Only UTM tags and referrers identify those signups. Other people may also have found the site through an AI tool without leaving that trail.
What this does not show
- It does not prove the pages caused the results. These are observed numbers and before and after movement does not prove causation. Seasonality, rival changes and search engine updates also matter.
- Most of the Google and Bing window, 18 Sep to 15 Oct 2026, came before the paced indexing rollout began on 12 Oct 2026. The table covers that period for the site, rather than the rollout’s effect alone.
- Average position is an average across many searches. It is not a fixed rank, and it differs by country, device and person.
- Bot requests and page fetches show that AI tools read the site. They do not show how often the site was cited in an answer.
- We do not publish paid conversion numbers here. Signups are accounts, not customers.
- One site in one niche is not a forecast for yours.
Who ran this work
We did this on a site we run, using the approach we build into SearchSeal. SearchSeal and goldprice.dev are owned and operated by Nusantara Ventures, LLC. Read it as an owner's account. The Narwhal case study covers a second site run the same way.