Your Bank Cannot Buy Into a PH AI Overview

On 19 December 2025, Google quietly expanded ads inside AI Overviews to eleven countries beyond the United States. The Philippines was on the list, alongside Australia, Canada, India, Indonesia, Kenya, Malaysia, New Zealand, Nigeria, Pakistan and Singapore. No announcement. Advertisers found out because a help page changed.

Buried in that same documentation sits a sentence that matters far more to some businesses than the rollout itself. Google states it currently does not show ads in AI Overviews for sensitive verticals including adult content, alcohol, gambling, finance, healthcare and politics.

Read that list again with a Philippine business in mind. Banks. Insurers. Lending companies. Investment platforms. Hospitals. Clinics. Pharmaceutical brands. Health insurers. For every one of them, the answer box at the top of the page is not something a budget can reach. If a Filipino asks an AI-assisted question in those categories, the only way to appear inside the response is to be a source the system chose on merit.

What actually shipped, and what did not

Two separate things get confused here, so worth separating them properly.

AI Overviews themselves, meaning the organic generated summaries, have operated in the Philippines since Google's expansion to over a hundred countries in October 2024, later extended to more than two hundred countries and territories in May 2025. Google's own support documentation lists English, Filipino and Cebuano among supported languages for this market.

Ads inside those overviews are a separate product with a separate rollout. Google previewed the APAC expansion around July 2025 at Google Marketing Live in Singapore, naming Singapore, Malaysia, the Philippines, Indonesia, India, Pakistan and Australia. The actual switch flipped on 19 December 2025. Eligible formats are Text and Shopping ads drawn from existing Search, Shopping and Performance Max campaigns, in English, on both mobile and desktop.

So the correct statement for a Philippine business: organic AI answers have been running here for well over a year, and paid placement inside them arrived in December 2025 with categories carved out.

Why the exclusion list is not an oversight

It would be easy to read the carve-out as a temporary gap that will close. That reading underestimates why it exists.

Finance, healthcare and the rest of that list are the categories where a wrong or misleading generated answer causes concrete harm to a person. They are also the categories under the heaviest regulatory scrutiny in most markets, the Philippines included. Placing a paid message inside a synthesised answer in those areas creates a liability surface that a search platform has good reason to approach slowly.

Whether the exclusion eventually narrows is unknowable and not worth planning around. What is plannable is the present state. In these categories, an appearance inside a Philippine AI answer is earned or it does not happen.

How much is actually at stake

The obvious next question. If paid placement is off the table, how much traffic is sitting behind that answer box in the first place?

Two numbers help, and both need their caveats attached.

On prevalence, there is one Philippine estimate. Ahrefs analysed 108 million AI Overview queries in its Brand Radar database and put the Philippines at 29.1 percent of keywords, level with Mexico, behind Indonesia at 37.2 percent, ahead of India at 26.8 percent and Nigeria at 26.4 percent. The United States sat thirteenth at 20.5 percent. Single source. Method-dependent. And other research reviewed for this cycle stated plainly that no Philippine prevalence figure should be presented as Philippine-specific. Both of those positions are on the record and we are not going to resolve them by choosing the more useful one.

Global comparators show how much the corpus drives the answer. Semrush found approximately 15.69 percent across over 10 million keywords in November 2025. Conductor found 25.11 percent across 21.9 million queries in the first quarter of 2026. BrightEdge found 48 percent, but across nine commercial industries rather than a broad corpus. Xponent21 found 60.32 percent in April 2026. Spread of nearly forty-five points, all defensible for what each measured, none interchangeable with the others.

On the cost side there is genuinely a Philippine study, and it tends to get overlooked. LeapOut Digital published "Ranked But Bypassed" in June 2026, tracking GA4 and Search Console for 11 brands, 7 of them Philippine, from January 2023 to June 2026. Across roughly 59 million search impressions over twelve months, fewer than 1.3 million visits resulted. Philippine brands at average positions 8.6 and 9.1 recorded click-through of 1.9 percent and 2.3 percent, against a pre-AI benchmark of roughly 6 to 9 percent for those positions.

The authors state that the decline cannot be attributed to AI alone, and they rate their own claim that Philippine brands trail Western markets by 18 to 24 months as a hypothesis rather than a finding. Both caveats deserve repeating rather than stripping.

Put the pieces together and the shape is clear enough for planning. A substantial and possibly growing share of Philippine result pages carry a generated answer. Click-through on Philippine properties in the positions most businesses occupy has compressed well below historical norms. And in finance, healthcare and four other categories, the paid shortcut does not exist. What remains is being cited.

What that means for the categories affected

Category	Paid route into an AI Overview	What this changes about the plan
Banking, lending, insurance, investment	None at present	Organic citation becomes the only entry. Entity clarity and third-party corroboration move from nice-to-have to primary channel
Hospitals, clinics, health insurers, pharma	None at present	Same, with an added premium on named authorship, credentials and cited primary sources, because health answers face a higher trust bar
Alcohol and gambling	None at present	Organic only, and worth checking Philippine advertising rules for the category before building anything at all
Retail, e-commerce, consumer goods	Available	Shopping and Text ads from existing campaigns are eligible. Paid and organic citation compete for the same real estate, so track them separately
B2B software and professional services	Available	Available, but frequently not where the buyer is. Vendor-shortlisting questions run through assistants carrying no ad inventory at all
Every category, on ChatGPT and Claude	None	These surfaces carry no equivalent ad product. Appearing there is earned regardless of sector

That last row generalises the point beyond the excluded categories. ChatGPT is used monthly by 42.4 percent of Philippine internet users aged 16 and over, sixth highest of any market surveyed by GWI in the second quarter of 2025 and roughly 1.6 times the global average of 26.5 percent. Inside surveyed Philippine organisations, reported ChatGPT use runs at 83 percent. None of that traffic can be bought.

What earning it actually involves

The categories locked out of paid placement share a useful property. They are also the categories where authority signals are most legible and most checkable. Works in favour of anyone willing to build them properly.

Name the regulator, and name it correctly. A Philippine financial page referencing the Bangko Sentral ng Pilipinas, the Insurance Commission or the Securities and Exchange Commission where each is genuinely relevant reads as Philippine to a retrieval system in a way that a generic finance page does not. Same holds for health content and the Department of Health or the Food and Drug Administration.

Put a real person on the page. Named author, stated credentials, a reviewer where the subject warrants one. In regulated categories this is not a formality. It is the difference between content a system can attribute and content it treats as anonymous.

Answer the question in the first sentence. A direct, self-contained answer at the top of a section, then the evidence and the exceptions. Retrieval systems extract from structure, and content that buries the answer three paragraphs into a narrative is harder to lift cleanly.

Get corroborated somewhere you do not control. A claim appearing only on your own site is a claim with one source. The same claim reflected in Philippine business or trade publications, in a directory that matters, or in a discussion where someone else raises it, is a claim with a network behind it.

Keep the entity consistent everywhere. One legal name, one address format, one phone number, one description of what the business does, across the site and every third-party profile. Inconsistency gives a system a reason to hedge, and hedging is how you end up described vaguely instead of named.

None of this is novel. What changes in a locked-out category is that it stops being best practice and becomes the channel.

The measurement that should follow

One warning about how this gets reported. Because paid placement is unavailable in these categories, there is a temptation to treat any appearance of the brand in an AI answer as a win and screenshot it. That is not measurement, and it will not survive a procurement conversation.

Freeze a panel of the questions your buyers actually ask. Run them repeatedly, from Philippine sessions, across the engines that matter in this market. Record the full answer, the cited URLs and their positions, the engine and the model date. Keep brand mention in a different column from source citation, because in regulated categories being named without being cited is a common and meaningfully different outcome.

Then compare against the baseline you took before starting. Without that baseline a favourable result is unfalsifiable, and so is an unfavourable one.

Worth knowing that measurement capacity in this market is genuinely thin, which is part of why so little of this gets done properly. Swarm's Philippine AI Report 2025, surveying 175 organisations, found 92 percent had used AI in some capacity but 65 percent remained at proof-of-concept, with talent scarcity named as a barrier by 57 percent and security or privacy concerns by 40 percent. Broad adoption, shallow depth, and not many people available to build the infrastructure.

Our GEO service for the Philippines builds that baseline first, and pairs with our SEO service for the Philippines, because engines still retrieve from crawlable, indexed content.

---

Frequently Asked Questions

When did ads inside AI Overviews go live in the Philippines?

19 December 2025. Google expanded ads in AI Overviews to eleven countries beyond the United States through a documentation update rather than a formal announcement, covering Australia, Canada, India, Indonesia, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore. Placements are available in English on both mobile and desktop.

Which categories cannot buy ads inside AI Overviews?

Google states it currently does not show ads in AI Overviews for sensitive verticals including adult content, alcohol, gambling, finance, healthcare and politics. For Philippine banks, lenders, insurers, investment platforms, hospitals, clinics and health insurers, there is no paid route into that answer at present.

Do we need a new campaign type to appear there?

No. Eligible placements draw on Text and Shopping ads from existing Search, Shopping and Performance Max campaigns. Ads may appear within the AI Overview or above and below it depending on keyword match type and relevance.

Are AI Overviews themselves available in Filipino?

Yes. Google lists English, Filipino and Cebuano among supported languages for AI Overviews in the Philippine market. Note the asymmetry: the organic surface supports three languages while paid placement is English only, so the earned route reaches a wider audience than the paid one even in categories where paid is available.

How often do AI Overviews appear on Philippine searches?

There is one Philippine estimate and it is disputed. Ahrefs analysed 108 million AI Overview queries in its Brand Radar database and placed the Philippines at 29.1 percent of keywords, level with Mexico, behind Indonesia at 37.2 percent and ahead of India at 26.8 percent, with the United States thirteenth at 20.5 percent. Single dataset and method-dependent, and other research reviewed for this article holds that no Philippine prevalence figure should be presented as Philippine-specific. Global comparators run from approximately 15.69 percent across over 10 million keywords in November 2025 to 60.32 percent in April 2026, which shows how much the query corpus drives the result.

What does an AI Overview actually cost us in clicks?

One Philippine study addresses this. LeapOut Digital's "Ranked But Bypassed", published June 2026, tracked GA4 and Search Console for 11 brands, 7 of them Philippine, from January 2023 to June 2026. Roughly 59 million impressions produced fewer than 1.3 million visits over twelve months. Philippine brands at average positions 8.6 and 9.1 recorded click-through of 1.9 percent and 2.3 percent against a 6 to 9 percent pre-AI benchmark. The authors state the decline cannot be attributed to AI alone, and they rate their own claim that Philippine brands trail Western markets by 18 to 24 months as a hypothesis rather than a finding.

Can we buy placement inside ChatGPT or Claude answers?

No. Those surfaces carry no equivalent advertising product, so appearing in them is earned regardless of sector. That matters commercially here: 42.4 percent of Philippine internet users aged 16 and over used ChatGPT in the previous month according to GWI's Q2 2025 survey, sixth highest of any market surveyed, and reported use inside surveyed Philippine organisations runs at 83 percent.

Will the sensitive-vertical exclusion be lifted?

Unknowable, and not worth planning around. The exclusion covers categories where a wrong generated answer causes concrete harm and where regulatory scrutiny is heaviest, which are reasons for a platform to move slowly rather than a temporary technical gap. Plan for the present state, which is that appearance in these categories is earned or it does not happen.

---