How did a restaurant in Michigan end up on a Makati list?

The Google Places text search that built the list set regionCode to PH, which biases results toward a region and formats addresses for it but does not filter by location. The request also restricted results to the restaurant type, so the type was constrained and the geography was not. When the search ran short of strong local matches it returned a venue in Flint, Michigan, and nothing in the pipeline rejected it.

How many venues were affected across the network?

Eleven venues across three of the eight publications were filed under a section they do not belong to. One was in the United States. Four were in the wrong city inside Metro Manila. Three were on mainland Bohol rather than the island of Panglao. Three more were moved between sections on the same site after the overlap rule was added.

Did the monthly verification not catch this?

No, and it reported zero problems on the day the error was found. The verification asks whether each published venue is still trading. The Michigan branch is open and well reviewed, so it passed. A check for closures cannot detect a venue that is open in the wrong country. The geographic test now runs on every verification pass rather than only when a section is first built.

Why does the BGC list only have four venues now?

Four of the six entries were not in BGC and were removed. A search inside the 2km boundary found only four qualifying venues, meaning poke or salad places carrying at least 25 Google ratings. The page says four rather than padding the list back to six with venues from neighbouring areas.

How do I report something else that looks wrong?

Email hello@localenhance.com. Every venue on this site carries the Google rating, the review count and the date it was checked, so a claim here can be checked against Google directly. If something is wrong it gets corrected on the page rather than quietly left.

Makati

A restaurant in Michigan was ranked first for Makati

Q: Why use a radius instead of checking the city name?

Because Google reports many genuine BGC addresses with the locality Makati City, including Bonifacio High Street and 28th Street. A test comparing city names cannot separate BGC from Poblacion, which is a different area 2.8km away and also reads as Makati. The coordinates Google returns for a venue are exact even when its address formatting is not, so a circle can distinguish them.

A poke bar at 1214 University Ave in Flint, Michigan held first place on this site's Makati ranking. The Google Places search that built the list filtered by business type but not by location, so it accepted a venue 13,000km away. Every section now has a geographic boundary and eleven misfiled venues were removed.

By Tony Hilden 28 July 2026 4 min read

Contents

What actually broke
It was not one venue
What changed
The check that was not enough
What this list is now

For a few weeks, the best place to eat a poke bowl in Makati was, according to this site, a restaurant at 1214 University Ave Ste 1 in Flint, Michigan.

It sat at position one. It had a 4.9 Google rating from 163 reviews, which is a genuinely good score, and an editorial score of 92.2 that was calculated correctly from real data. Everything about the entry was accurate except the part that mattered, which is that it is 13,136km from Makati and has never been in the Philippines.

It is worth being specific about how bad that was, because the error was not confined to the page. The structured data this site publishes for search engines carried the Michigan street address with addressCountry set to PH. The FAQ block answered the question “What are the best poke and salad bowls in Makati?” by naming it. Anything reading this site programmatically, which includes Google, was being told a restaurant in Michigan was the best bowl place in Makati.

What actually broke

Every ranking here is built from the Google Places API. The query for the Makati section asked for poke and salad places, restricted results to the restaurant type, and set regionCode to PH.

That last part is the problem. regionCode tells Google which region to bias results toward and how to format addresses. It is not a filter. Nothing in the request said the venue had to be inside Makati, or inside the Philippines at all, so when the text search ran out of strong local matches it reached further, and one of the things it reached was Flint.

The type filter was strict. The geographic filter did not exist.

It was not one venue

Once there was a reason to check, the same audit ran across all eight publications, comparing every venue’s section against the address Google holds for it. Eleven were in the wrong place.

Five of the six venues in this site’s BGC section were not in BGC. One was in Kapitolyo, Pasig. One was in Mandaluyong, at Shangri-La Plaza. One was in Poblacion, which is Makati but is not BGC and is 2.8km away from it.

On the sister site for things to do, three of the six entries under Panglao were not on Panglao. Bilar Man-Made Forest, Camugao Falls and Pangas Falls are all on mainland Bohol, between 28km and 39km from the island.

None of these were close calls dressed up as errors. They were a filter that did not exist.

What changed

Every section now carries a centre point and a radius, and a venue outside that circle is rejected. BGC gets 2km, which is about right for a 240-hectare district. Panglao gets 13km, enough for the whole island. El Nido gets 30km because its attractions are islands offshore.

The reason it is a circle and not a comparison of city names is worth explaining, because the obvious fix does not work. Google reports a large number of genuine BGC addresses with the locality “Makati City”. Bonifacio High Street comes back as Makati. So does 28th Street. A test that compares the city name against the section cannot separate BGC from Poblacion, which is a different place 2.8km away. The coordinates Google returns for a venue, though, are exact. A circle can tell those two apart. A string cannot.

Sections on one publication also compete now. Makati’s circle contains BGC, because the two are under 3km apart, so a venue that falls inside both goes to whichever centre is nearer. That is how a branch of Raging Bull Burgers at Shangri-La The Fort stopped being one of the best burgers in Makati. It is in BGC.

The check that was not enough

The monthly verification job re-checks every venue against Google and removes anything reported permanently closed. It ran the day this was found and reported zero problems.

It was right, and it was useless. The Michigan branch is open. It has good reviews. Asking only “is this place still trading” will never catch a venue that is trading perfectly well in the wrong hemisphere. Worse, the verification had recently been rewritten to stop re-rolling the lists every month, which means it now trusts what was already published. It was actively protecting the error.

The geographic test now runs on every verification pass, not just when a section is first built. That is the change that matters. A filter that only applies to new entries leaves every existing mistake in place forever.

What this list is now

The Makati ranking on this site is six venues, all of them in Makati. The BGC ranking is four, because after removing the four that were not in BGC, only four qualifying venues were found inside the boundary.

Four is a worse-looking list than six. It is also true, and the page says four rather than pretending otherwise. A ranked list is only worth reading if everything on it is where it claims to be.