BCA and Bank Mandiri show why AI visibility needs context
BCA and Bank Mandiri lead different AI banking questions in Indonesia, showing why visibility audits need customer context and verifiable evidence.
Bank Central Asia and Bank Mandiri do not occupy one fixed position inside AI-generated answers. New research on Indonesia’s banking sector found that BCA appeared more prominently in questions about personal banking and trust, while Bank Mandiri led more often when prompts concerned business banking and economic contribution.
That split turns a familiar marketing question into a more demanding one. The issue is not simply whether a brand appears in an AI answer, but which customer need brings it forward, which system is answering, and what public evidence supports the result.
The study, produced by PRecious Communications, should be read as a diagnostic snapshot rather than a league table. It measured answer presence and first mention across tested AI routes. It did not measure bank quality, customer preference, security, or the likelihood of a sale.
Table of contents
Jump to each section:
- What the Indonesia results actually show
- Why one visibility score hides the market
- Evidence becomes part of brand positioning
- What marketers should know about AI answer visibility
What the Indonesia results actually show
The Indonesia findings separate reach from prominence. BCA appeared widely across the tested answers and led in personal banking and trust-related questions. Bank Mandiri was more likely to be named first when the question shifted toward business services or the bank’s role in the economy.
86% of the tested Indonesia answers included BCA, while the bank’s first-mention strength varied by customer need.
This matters because an AI answer is not a neutral shelf where every brand has one stable position. The wording and intent of a question change the comparison set. A consumer asking which bank is trusted for everyday needs creates a different competitive frame from a business owner asking about trade finance or corporate services.
Brand visibility is becoming situational before it becomes measurable.
The research also found substantial variation across the wider regional sample. Identical customer questions did not reliably produce the same first-mentioned bank when the tested AI route changed.
43% of eligible cross-service comparisons named a different bank first when the AI route changed, even though the customer question remained the same.
The result does not prove that one AI service is more accurate than another. It shows that a single audit can reveal only one slice of a brand’s machine-mediated visibility. The study itself notes that the findings can change as models, search features, sources, and public information evolve.
Lena Marlowe
Why one visibility score hides the market
Marketers are used to compressing performance into rankings, share of voice, or a dashboard score. That habit is understandable because a single number creates an easy benchmark. In AI discovery, however, compression can remove the context that gives the result meaning.
The common assumption is that a brand with the highest overall visibility owns the strongest position. The contrasting reality is that BCA and Bank Mandiri led different kinds of questions. The strategic implication is that communications teams need to define the decisions they want to influence before choosing the prompts and metrics used to evaluate visibility.
A bank can be present in many answers without leading the questions most relevant to its growth priorities. Another can appear less often overall but hold a stronger position in a valuable service category. Those are different communications problems and should not produce the same response.
Presence is not positioning, and first mention is not preference.
This distinction also limits how the findings should be interpreted. The study used a set of tested AI routes with different search and source conditions, rather than reproducing every consumer’s experience in a live application. Answer order is therefore evidence about the recorded responses, not proof of customer belief or future market share.
For marketers, that limitation is useful rather than disqualifying. It clarifies what an AI visibility audit can do well: identify inconsistencies, reveal which needs a brand is associated with, and surface where the public evidence may be thin or fragmented.
Evidence becomes part of brand positioning
PRecious Communications recommends that brands connect their desired position to clear facts, customer evidence, and verifiable information. In the banking example, that means a claim about trust, business support, or service capability needs enough current public material for an AI system and a skeptical customer to inspect.
This changes the role of communications assets. Product pages, executive commentary, media coverage, help content, and market-specific explanations are no longer only destinations for human readers. Together, they form part of the evidence environment from which AI systems may construct an answer.
The strongest message is the one the evidence can keep intact.
There is a temptation to treat this as a volume problem and publish more material around every possible query. The study points toward a narrower discipline. Brands first need to identify the customer questions that matter, then examine whether the available evidence answers those questions clearly, accurately, and in the language used by the market.
Indonesia makes that requirement especially visible. A broad national reputation can coexist with very different positions across personal banking, business services, trust, and public contribution. Generic brand copy may establish awareness, but it may not give an AI system enough specificity to connect the institution with a particular need.
The deeper shift is that positioning now has a machine-readable layer. A brand still needs a distinctive promise, but it also needs public proof that can survive retrieval, synthesis, and comparison.
What marketers should know about AI answer visibility
The banking study offers a practical model for auditing AI visibility without mistaking a snapshot for a verdict.
Start with the customer question. Define the need, decision, and service category before testing visibility. A general brand prompt can hide the contexts where the competitive field changes.
Separate reach from prominence. Track whether a brand appears and whether it is named first, but do not treat either measure as proof of preference or performance.
Audit across meaningful contexts. Compare relevant AI routes, languages, and markets using a repeatable set of questions. Record the testing conditions so changes can be interpreted responsibly.
Inspect the evidence behind the answer. Check whether product facts, third-party coverage, customer proof, and owned content are current, consistent, and specific enough to support the position the brand wants to hold.
AI visibility will be attractive to marketers precisely because it can be scored. The risk is allowing the score to become more authoritative than the customer context it compresses.
For communications teams, the more durable objective is not to win every answer. It is to make the brand’s relevant strengths legible when a customer asks a question the organization is equipped to answer.
That widens the remit of search and reputation work. Brand strategy, public relations, content operations, product marketing, and regional teams all contribute to the evidence an AI system may encounter. No single dashboard can substitute for that coordination.
The emerging competitive advantage is not universal first place. It is contextual credibility: being visible for the right need, supported by evidence that remains clear when the interface changes.
