How Taman Safari Indonesia Became the Most-Cited Wildlife Brand in AI Search.
When Search Agency first measured how AI assistants answer questions about Indonesian wildlife parks, Taman Safari Indonesia was already ahead on the metric most brands chase. Assistants named it more often than any competitor, which made it the category leader for share of voice. The catch sat in where those answers came from. When ChatGPT or Gemini described a Taman Safari park, the underlying facts came off review sites, travel forums and booking aggregators instead of tamansafari.com, so the brand was in almost every answer while other people supplied the source behind it.
Between late June and early September 2026, that changed. Taman Safari's English AI citation rate more than tripled, from 12% to 38%, while the brand held its number one share of voice, and in Indonesian answers it moved from the third or fourth name mentioned to almost always the first.
Who Taman Safari Indonesia is
Taman Safari Indonesia is the country's largest wildlife and conservation group. It spans six locations, eight parks and four resorts, from the flagship safari park in Bogor to Bali Safari and Marine Park, Taman Safari Prigen in East Java and Solo Safari in Central Java. Millions of families pass through the gates every year, and most of them start planning on tamansafari.com, where the group sells tickets, lists show schedules and lays out what each park offers.
Search Agency has run the group's SEO and generative search program since 2025. The brief covers the whole family of parks and both languages its audience searches in, English and Indonesian, which is why the AI-answer problem was measured on both fronts from the start.
The attribution gap
More travellers now open an AI assistant instead of a search box when they plan a trip. They ask which Indonesian wildlife park is best for a family with young kids, or how Taman Safari compares to Singapore Zoo, and they treat the answer as a recommendation rather than a list of links to sift through. That shift is where Taman Safari carried a hidden weakness.
The brand led the category for share of voice, so it turned up in plenty of answers. Its citation rate told a different story. In English, assistants cited Taman Safari's own pages in only 12% of answers about the parks. In Indonesian the figure was 3%. Most of the time an assistant would describe a park, quote a price or rank it against a rival using a third-party page, a review aggregator or a forum thread, never the source that actually runs the park. Being mentioned kept Taman Safari in the running, but it did nothing to put the brand's own facts, prices and schedules into the answer.
Total Graph Authority
Total Graph Authority is the framework Search Agency uses to turn a brand into the source an AI model reaches for when it builds an answer. It rests on a simple observation about how assistants behave. They do not cite a name, they cite evidence, and evidence has to line up across the whole web before a model treats it as safe to quote.
The framework works on three layers at once. The first is the brand's own site, where Taman Safari needed structured, citable pages for the things people ask about, tickets, show schedules, experiences, individual animals and its conservation work, each written as a clear answer with the specific facts an assistant can lift and attribute. The second is everything the wider web says about the parks, so encyclopedic and knowledge-graph records, booking platforms and review sites all had to carry the same park names, locations and facts, letting a model that checks one source against another find agreement instead of contradiction. The third is the useful content that sits around a decision, price clarity, park comparisons and animal-welfare information, published where travellers and assistants both go looking for it.
None of the three works alone. A polished set of pages that nothing else on the web backs up still reads as a single unverified claim, and a strong third-party footprint counts for little if the brand's own site cannot answer the question in a form worth quoting. When all three line up, a model stops hedging and starts pointing at the brand's own pages.
How the work was organised
The framework tells you what to build. The order of the work came from how people actually shop for a day out. Search Agency mapped the high-intent questions a traveller might put to an assistant and grouped them into five buyer-intent hubs, tickets and schedules, experiences, animals and conservation, planning a visit, and choosing which park to go to. Each hub became a cluster of pages aimed at the real questions inside it, so the content Taman Safari published matched the phrasing people use with AI rather than the phrasing a marketing team might reach for.
Inside each hub the three layers ran in parallel. On the site, Search Agency wrote answer-shaped pages that put the quotable fact first, the ticket price, the opening time, the animals a child will see, the distance from the city, then the supporting detail underneath. Off the site, the team worked through the entity records and third-party listings that describe each park, correcting names, locations and facts so a model cross-checking sources found one consistent story. The social and comparison content filled the gaps that reviews and forums used to own, the honest price breakdown and the straight comparison against a rival park.
The aim was narrow and hard to fake. Every hub was built so that when an assistant answered a question in that space, the page it pulled from and attributed was Taman Safari's own.
What changed in the answers
Search Agency measured the program the same way before and after the work, running a fixed set of buyer-intent prompts through the same five AI platforms, ChatGPT, Gemini, Perplexity, Google AI Overviews and Google's AI Mode. The English read covered 50 prompts and 250 results, the Indonesian read 18 prompts and 90 results. The baseline was taken in late June 2026 and the latest read in early September 2026, all of it tracked through StoryMint's AI Brand Performance tool.
The headline is the English citation rate. It moved from 12% to 38%, a 3.2x gain, while Taman Safari held on to the number one share of voice it started with. The brand went from being described through other people's pages in most answers to being the cited source in well over a third of them, without giving up any of the ground it already led on.
| Metric | Baseline | Latest | Change | What it means |
|---|---|---|---|---|
| English AI citation rate | 12% | 38% | 3.2x | AI now cites Taman Safari's own pages |
| English brand mention rate | 30% | 40% | +10 pts | Named in more answers |
| English share of voice | 55% | 54% | Held #1 | Still the category leader |
| English avg answer position | #3.8 | #3.6 | Improved | Appears higher in answers |
| English ChatGPT mention rate | 15% | 48% | +33 pts | Weakest platform became a strong one |
| Indonesian share of voice | 44% | 48% | +4 pts | Extends the lead at home |
| Indonesian AI citation rate | 3% | 5% | +2 pts | Early, still building |
| Indonesian avg answer position | #3.1 | #1.9 | Improved | Often the first brand named |
Where the biggest gains showed up
The platform-level numbers show where the shift was sharpest. In English, ChatGPT went from Taman Safari's weakest platform to one of its strongest, with the brand's mention rate climbing from 15% to 48%. Google AI Overviews reached 42% and AI Mode 50%, which matters because those two surfaces sit directly under Google's search results, in front of the exact audience already looking for a day out.
The competitive picture moved too. In English answers, Taman Safari now leads ahead of Bali Zoo, Singapore Zoo and Australia Zoo. In Indonesian, it leads ahead of Ancol, and when it appears it is usually the first brand named, at an average rank of #1.9 against #3.1 before the work. Its Indonesian share of voice climbed from 44% to 48%, extending a lead it already held at home even though the citation rate there, now 5% against a 3% baseline, is still early.
Why this is a durable advantage
AI assistants are turning into the first place a travel decision gets shaped, and they lean on brands they can cite with confidence. Once Taman Safari became a citable source, its own prices, park details and conservation story started flowing into answers directly, instead of reaching travellers secondhand through a review site or a forum, so the family reading that answer sees the park's real information framed the way the park would frame it. That kind of position holds, because a citation compounds as the pages and entity records behind it keep earning trust, and it is far harder for a rival to knock out than a paid placement that vanishes the moment the budget stops. Taman Safari did not buy its position in these answers. It earned a place in the evidence assistants trust, and that place holds until someone builds something more credible.
What comes next
The English result is the proof of concept, and it sets the template for Indonesian. Share of voice and answer rank in Indonesian already lead the category, which leaves citation rate as the next lever, the same move that worked in English. That means pointing structured, quotable pages at the Indonesian questions people ask about tickets, park comparisons and animal welfare, and correcting the entity records that feed them.
That work is underway, with a close eye on the platforms where Taman Safari still has the most room to climb. The method is proven in one language, and the second is a matter of running the same play again.
About Search Agency
Search Agency is an SEO and GEO/AEO agency that helps brands win visibility across Google and AI assistants. The Total Graph Authority framework behind the Taman Safari program is how we turn a brand from a name assistants mention into a source they cite.
If you want to see where AI assistants get their facts about your brand today, and where those facts should be coming from instead, start with a free AI Visibility Audit at search.agency/ai-search. It is the same first read Search Agency ran on Taman Safari before any of this work began.
See where your brand stands in AI answers today, benchmarked against your competitors, no pitch required.
How Prenagen Came to Own AI Search in Indonesia
Generative engines now answer many of the high-intent questions that brands once captured through traditional rankings. When an AI Overview resolves a query inside the results page, the underlying website often loses the click, even when it ranks well. Across the Indonesian maternal-health category, keyword visibility contracted by roughly 9% as this repricing took hold.
read_post →
When Ranking #1 Stops Being Enough in the Age of AI Search
Prenagen was pulling close to 48 million search impressions a month and still watching its traffic slip. Here is what its data taught us about doing SEO in the age of AI Overviews, and what it means for your strategy in 2026.
read_post →
How We Grew DiscoverASR's AI Search Audience by 450% in a Single Quarter
In one quarter, DiscoverASR, a global serviced residence and extended-stay hospitality brand, went from being a quiet challenger inside ChatGPT, Gemini, Google AI Mode, and Perplexity to a brand that AI engines actively recommend.
read_post →