Why a recent month in Thailand made me question some of the Maps offered user experience
OK, so this will be a long story… hope you got the time to spare.
I hadn’t been to Thailand in 26 years.
Last time, my kids were young, and while we’ve travelled extensively since, somehow we never made it back. Life, work, other destinations—you know how it goes.
This January, that changed. We’d skipped travelling over Christmas, and originally thought about Florida for January. But then we discovered that various family members were already planning Thailand trips, and one thing led to another. My wife Atara had recently retired, I negotiated some remote working with the office, and what started as “a quick hop to Thailand” turned into a full month: Bangkok, Ao Nang and Koh Lanta.
Family scattered across different countries, all converging on the same beaches at roughly the same time.
It wasn’t an exploration trip. We weren’t trying to see as much of Thailand as humanly possible. The plan—if you can call it a plan—was to find few places where family could gather, do relaxed activities together, and I’d keep half an eye on work while everyone else lounged by the pool. It worked surprisingly well.
I probably should tell about the Thai experience but that would take an even more time. What I thought was telling about my Maps experience.
Here’s the thing about being a Local Guide: even when you’re decidedly not local, you somehow become the designated navigator.
Phone in hand, running slightly ahead of the group, everyone following. “Sam, where are we eating tonight?” “Sam, where’s breakfast?” “Sam, find us somewhere interesting.”
So I opened Google Maps.
A lot.
My experience left much to be desired and I’ll explain the typical problem.
Search for resturants around our hotel within walking distance:
What I got (again and again) was a sea of red pins, each proudly displaying ratings between 4.5 and 4.9. Occasionally, a rebellious 3.9 would pop up like a warning beacon. Everything else? Excellent. Amazing. Outstanding.
Pick one—they’re apparently all interchangeable.
Here’s the thing: they’re not. Not even close.
The Variance Problem: When a Shack and a Five-Star Kitchen Score the Same
Here’s something you need to understand about eating out in places like Koh Lanta, (and it’s very different from dining in London or New York).
In Europe or the US, there’s a baseline. Health inspections, building codes, safety standards. Every restaurant has proper flooring, normalised furniture, walls that keep weather and wildlife out. You might not love the food, but you’re unlikely to question whether the kitchen has running water.
Thailand - as far as I can tell - operates on an entirely different spectrum.
On one end, you have restaurants that could sit comfortably in any modern city. Proper kitchens, trained staff, air conditioning, the works.
On the other end, you have what I can only describe as a shack: no flooring, plastic chairs, open to the wind (and the occasional wandering dog), hygiene standards that would give a European health inspector palpitations.
Now don’t get me wrong, sometimes the shack serves the most phenomenal food you’ve ever tasted. Fresh, vibrant, cooked by someone who’s been making the same dishes for thirty years. Sometimes the fancy place serves generic tourist fare that tastes of nothing in particular.
But they both score 4.7.
This is where the rating system fails spectacularly. The variance in Thailand—in infrastructure, facilities, ambience, safety, and yes, food quality—is enormous.
You’d expect the scoring to reflect that. You’d expect to see genuine range: this place is basic but brilliant, this place is polished but disappointing, this place gave three people food poisoning, this place is spicier than you can handle.
Instead, you get a narrow band of near-identical excellence.
As if the plastic-chair shack and the marble-floored restaurant represent roughly equivalent experiences.
The Mobile Experience Is Where This Falls Apart
I assume many of us has similar Maps experience: You’re standing on an unfamiliar street, hungry,(mildly jet-lagged), squinting at your phone and you want a quick answer: where should I eat?
What Maps gives you is a wall of near-identical ratings.
Danny’s Restaurant: 4.9. Fill-Feel Restaurant: 4.8. Uncle O Restaurant: 4.7. Sam-Pe-Nong Kitchen: 4.6. Scroll down. More 4.6s. More 4.7s. Occasionally a bold 4.5 to break up the monotony.
The actual differences between these places? Staggering. We ate at restaurants that were genuinely wonderful—fresh ingredients, careful cooking, the kind of meal you remember.
We also ate at places that were… fine. Acceptable. The definition of “OK.”
You’d never know which was which from the ratings. When 95% of restaurants score between 4.5 and 4.9, you’ve created a system that communicates nothing useful.
I’m sure this had been discussed 100s of times but… nothing changed.
The Case of the Suspiciously Popular New Restaurant
In Ao Nang, we visited an Indian restaurant. Lovely spot.
The staff mentioned, casually, that they’d been open for a few weeks. Being a Local Guide myself, I naturally had a look at their Maps listing later.
Hundreds and Hundreds of reviews.
Now, I’m not saying every single one was fake. But I’ve been contributing to this platform long enough to recognise the difference between organic growth and… let’s call it “accelerated enthusiasm.” The language patterns, the timing, the suspiciously comprehensive praise for a restaurant that had barely finished unpacking its cutlery.
Reviews, it seems, can still be bought in bulk.
And Maps still hasn’t quite figured out how to stop it.
Gemini’s Diplomatic Summaries Aren’t Helping
Google’s Gemini AI-generated summaries are meant to help you cut through the noise. In practice, they read like a particularly non-committal reference letter.
Every restaurant “offers a welcoming atmosphere.” Every hotel “provides comfortable accommodation.” Any reservations—and there are always reservations buried in the actual reviews—get relegated to a brief aside at the end, phrased so diplomatically you might miss it entirely.
“Guests praise the friendly service, generous portions, and central location. Some visitors noted occasional wait times during peak hours.”
Translation: the food’s probably fine and you might have to queue.
Thanks, that narrows it down to roughly every restaurant on the strip.
What I Actually Want From Maps
I’m a heavy Maps user. I’m also a Local Guide. I contribute because I believe the platform works better when real people share real experiences. But here’s what frustrates me as a consumer of that information:
Why can’t I filter by review count? A restaurant with 2,000 reviews and a 4.5 rating is fundamentally different from a restaurant with 3 reviews and a perfect 5.0.
One has survived thousands of assessments. The other has been likely reviewed only by the owner’s cousins.
Why isn’t the rating displayed with context? Show me “4.5 (2,847 reviews)” versus “5.0 (3 reviews)” at a glance. Don’t make me tap through to discover that the “top-rated” option is statistically meaningless.
Why no weighting for verified Local Guides? Someone who’s contributed thousands of thoughtful reviews probably has more credible opinions than an account that’s reviewed exactly one place: this one, today, with perfect marks.
Do we check review velocity? A place that’s suddenly accumulated 200 reviews in two weeks after opening might warrant a raised eyebrow.
The Rating Scale Is Broken Because We Broke It
Part of this is cultural. We’ve collectively agreed that anything below 4 stars is basically an insult. Businesses genuinely get upset when they’re rated “3 stars”—which, if you actually read the scale, means “OK.” Average. Neither good nor bad.
But “average” somehow has become synonymous with “avoid at all costs.”
According to Google the average map score is: 4.11 (!)
So everyone gives 5 stars for anything remotely acceptable, and reserves 1 star for genuine disasters. The entire middle of the scale—the part designed to help people distinguish between “decent” and “exceptional”—has collapsed into irrelevance.
I’ll admit, I’m part of this problem.
I’ve given 5 stars to places that were merely good because I didn’t want to hurt a small business. But when I’m using the platform rather than contributing to it, I desperately want those distinctions.
The Expectations Paradox: Why the Friendly Bungalow Outscores the Professional Hotel
Here’s something I noticed with accommodation that made me properly question the entire system.
We stayed at beach bungalows—charming places, run by lovely people who clearly worked hard to make guests happy. The facilities were “charming” but basic. Pavements uneven. Showers questionably installed. Nothing that would pass muster against any Western motel standard.
But the owner was wonderful. Friendly, accommodating, making genuine effort on a personal level. So naturally, people rate it 5 stars.
Meanwhile, proper hotels—the ones with front desks and professional staff and actual safety standards—get reviewed more harshly.
Towels weren’t replaced three times daily - 2 stars.
The pool bar closed at 5pm - 1 star.
The staff were courteous but not personal - 1 Star.
So the improvised bungalow with the cheerful operator outscores the professional establishment, because we’ve decided that warmth trumps real infrastructure.
Don’t get me wrong—personal service has enormous value. But does a place deserve a perfect score purely because the owner is friendly? Does a professional hotel deserve a lower score because our expectations were set higher?
We’re no longer comparing like with like. We’re comparing how a place made us feel against our expectations rather than what the place actually offers.
The Price Complaint: Your Fault, Not the Restaurant’s
While I’m airing grievances, a quick one I noticed regarding price.
I’ve lost count of how many reviews I’ve read where someone demolishes a restaurant because “it was expensive compared to other ‘out there’”.
They sat down, ordered, ate, enjoyed the food… and then rated it poorly because the bill was as high as priced on a menu.
You knew the prices going in. The menu was there. If it’s too expensive for you, step out—nothing wrong with that (We try checking pricing in advance to avoid time waste but still it happens that we will walk out if prices are not what we assumed).
But you chose to stay. You chose to eat. And now you’re punishing the restaurant for charging what they clearly stated they would charge?
If something was hidden—the drink prices weren’t listed, there was an undisclosed service charge—fair enough. That’s worth mentioning.
But “I thought the meal was expensive” when you had every opportunity to check before ordering? That’s not a review of the restaurant. That’s a confession of your own poor planning.
Do the Photos Help? Well, Not Really.
You might think: okay, the ratings are useless, but surely the photos tell the story?
In theory, yes. In practice, you have to dig for it.
Open any restaurant listing and you’ll see the same pattern. The gallery leads with a video now—Google’s been pushing video content heavily—usually someone zooming in on a beautifully plated dish. Then a few more food photos: curries glistening, noodles steaming, everything looking exactly as appetising as everything else. Maybe a shot of the menu. Perhaps a distant photo of the entrance.
That’s it. That’s what floats to the top.
And here’s the thing: all food looks fine in close-up. A plastic-chair shack can serve a photogenic pad thai just as easily as a five-star kitchen. The food photos tell you almost nothing about the actual experience of eating there.
What about the vibe? The seating? The hygiene? The view from your table versus the view in the carefully framed promotional shot? That information exists—buried somewhere in the gallery, uploaded by actual visitors rather than the business. But you have to scroll past dozens of food close-ups to find the one photo that shows the questionable wiring near the kitchen, or the plastic chairs, or the stunning sunset view that actually makes the place worth visiting.
Hotels are even worse. The business uploads their best angles: the pool looking pristine, the room at its most spacious, the entrance freshly cleaned. The reality—the dodgy shower electrics, the uneven pavements, the noise from the bar next door—you’ll only find that if you dig through to the guest photos. And who has time to do that for every option?
The photo system, like the rating system, has been optimised to make everything look good. The differentiating information exists, but it’s hidden. You have to work to find it. And when you’re standing on a street corner, hungry and jet-lagged, you don’t have time to archaeological dig through photo galleries for every restaurant in the area.
The surface presentation tells you nothing. The useful information is buried. Everything looks fine. Nothing looks exceptional. You’re back to guessing.
I should be fair. Maps didn’t fail me entirely—far from it.
Navigation? Brilliant. Getting from Ao Nang to Koh Lanta, finding ferry times, locating that obscure bungalow down an unmarked road at 10pm?
Maps was indispensable.
The sheer volume of information available—opening hours, photos, menus, general location awareness—saved us countless times.
The problem isn’t that Maps is bad. The problem is that for one specific, crucial use case—“help me choose between similar options”—the rating system has become essentially useless. Everything else works. This one thing, the thing I reach for most often, doesn’t.
The AI Pivot
Here’s something I’ve noticed myself doing more and more, and I suspect I’m not alone.
When I need to actually choose between options—restaurants, hotels, things to do—I’m increasingly turning to AI. Not Maps. Not review aggregators. Direct conversations with Claude, Gemini, Grok. I’ll describe what I’m looking for, the context, the constraints, and get a thoughtful response that weighs trade-offs.
It’s not perfect. AI doesn’t know that the restaurant changed ownership last month or that the chef quit. But it can synthesise available information fast, understand nuance, and give me a recommendation that feels considered rather than algorithmic (or worst expect me to do hours of digging).
I’ve essentially started treating AI as a thoughtful friend who’s read all the reviews and can give me an honest summary. “Given what you’ve told me, I’d probably try this place first, but that one’s worth considering if you want something quieter.”
That’s what I wanted Maps to do. That’s what the rating system was supposed to provide. Instead, I’m routing around it entirely.
I suspect much of what we spent years pushing into maps will now be quickly evaluated by AI and (hopefully) save us the leg work.
The Bottom Line
A month in Thailand. Multiple locations. Family from three countries. Countless meals, several hotels, endless navigation.
Maps helped us get places beautifully. It helped us choose places with significant difficulty.
The ratings told me everything was excellent. My stomach, my eyes, and my wallet told me otherwise. The variance was enormous—between the plastic-chair shack serving life-changing papaya salad and the polished restaurant serving forgettable tourist fare, between the improvised bungalow and the professional hotel—but the ratings flattened it all into the same comfortable band of 4.5 to 4.9.
Perhaps that’s the real lesson: when every rating tells you “excellent,” trust your nose instead. Find the full restaurant, smell something delicious, take a chance.
Some things haven’t changed since 2000.
Maybe the algorithm just hasn’t figured out how to replicate them yet.
*The trip itself?
Wonderful.
Beloved and I, her siblings, partners, all converging on Thai beaches from different corners of the world. Working remotely turned out to be surprisingly manageable. The family time was exactly what we needed.*
I just wish Maps had helped me find dinner as well as it helped me find the ferry terminal.
Things can only get better.
