There are a lot of apps for finding pubs and restaurants near you. And yet whenever my friends and I get together, we spend half an hour trying to think of somewhere, then go to the same substandard pub we always do because effort.
At Gadgette, we were a little skeptical when Twizoo got in touch, because honestly, we’ve seen it before. But Twizoo’s point of difference is interesting: instead of crowdsourced Yelp-style reviews, it uses tweets. Twizoo say there are 7 times more tweets about any given establishment than there are reviews, and that makes sense to me – I never rate places on TripAdvisor, but I’ll throw out a tweet for sure.
You don’t actually have to have Twitter to use Twizoo, but it’s better if you do. That’s because among the tweet reviews, it surfaces ones from your friends if there are any – and I was surprised by how many there were. Man, my friends eat a lot.
To use the extra Twitter functionality, you need to sign in with your Twitter account. Not everyone likes that, so it’s worth bearing in mind. Personally, I’m not sure why you need to authenticate when surely the app could see all the people you follow without actually going into your account, but I’m digressing.
Today, I found myself in Fitzrovia after a last-minute BBC appearance (yes, OK, #humblebrag). I have a friend whose office is round there, so I shot him a text and gave myself the time ’til he appeared outside his office to find somewhere to eat on Twizoo.
At first, I was a bit confused. I went into the app and a bunch of yellow and green bubbles started bouncing in a way that was clearly an error (they don’t seem to animate properly on my phone).
These were the places near my automatically-gleaned location that Twizoo lists. Green bubbles mean a place is well-liked, yellow means not so much, red obviously means avoid. I thought it was a bit strange to be giving me amber-rated places on the front page of the app – why wouldn’t you prioritise the green ones? There were plenty of them as I scrolled down.
I flicked to the next tab, which was my Bookmarks. This seems an odd choice for second tab, as no one who’s new to the app will have any (me included) and it doesn’t seem like a feature many people will use often. Surely once you know you like somewhere, you remember it with your actual brain…?
The third tab is called Feed. This is where Twitter Power comes in. The first option is to see my own tweets about restaurants, to help me remember places I liked. That’s cool. I’ll even forgive the “How Twizoo Are You?” hypercringe headline. But oh, it doesn’t work. At all. Ever. I’ve tried this page many times and get the same error every time.
The next feed is your friends. This was interesting, scrolling through to see which places the people I follow like. OK, it wasn’t excessively useful given that I follow over 2,000 people and don’t know most of them personally – and also that a lot of them were talking about places like Pret (why is Pret even in the app, no one needs to discover Pret) or places in other cities/countries – but I still found some interesting data.
However, oddly, it also includes the negative tweets. Like this one.
I’m not sure if that was an algorithm fail or it just includes all the tweets your friends have made about restaurants (and cafés, and in one case a juice brand), but I enjoyed reading about Haydy’s Nandos woes nonetheless.
The Feed category also includes food critics’ and websites’ tweets, and there’s a #foodporn feed as well. But my favourite bit of the Feed tab by far is Whine & Cheese, which is just a timeline of people complaining at brands about their food. It is fascinating. I lost ages reading people’s tales of lemon cakes with bite marks and undercooked salmon. Equally fascinating is trying to work out how the algorithm decides what to include, because honestly, it’s pretty wonky:
I think the way they’ve tooled this is to buy a list of negative words and just pull in all tweets with those words plus the handle of a food brand. But it doesn’t work in practice. As you can see on the left, “doesn’t disappoint” makes it in, as does, amazingly, “jerk chicken” on the right. I presume “jerk” is what triggered it, because nothing else in that tweet is remotely negative.
Fair to say the algorithms need work, then.
Anyway, none of this was helping me find a place to eat, and my friend was already halfway down the stairs.
I had a quick look at the “Trending” tab, which splits up the most popular places by the type of cuisine, but it wasn’t location-specific. I could have done with knowing what was trending near me.
Instead, I went back to Discover and scrolled until I found something that sounded interesting. The first few were chains (goddamn it I’m not having lunch at Costa), but then I found Mac & Wild, which I’d never heard of, and I was so in the mood for steak.
A tap on Mac and Wild brought up more information, including some tweets from someone I follow and one from someone “notable” (apparently).
I have to say, though, information was disappointingly limited. Rival apps like Zomato include photos of the interior, sample menus and an idea of what you’re getting. Twizoo only told me how much it cost (on a scale of £ to £££), where it was, and what some Twitter people thought. In some cases, the distance indicator was also broken, saying just ” away” instead of “0.27km away.”
Still, I was out of time, so I went for it. And while it was perfectly nice, the lack of extra information became clear immediately. This is not a place I’d have chosen to eat if I’d known enough about it. Yes, it’s a steak restaurant, yes it is where it says it is, but it’s hunting-themed. The doorhandle is a gun. There are framed pictures of skinning deer, and animal hides everywhere. If I’d had interior photos, I’d have chosen somewhere else.
Still, the Veni-Moo (beef and venison) cheeseburger was pretty delicious:
And after a good laugh about the decor, we had a successful lunch. Twizoo’s bright green rating for Mac & Wild obviously correlates well with local opinions of the place, because by 12.45 it was rammed. Or maybe everyone else was using Twizoo too.
We like the premise of Twizoo. While it suffers from the same problem as Yelp or TripAdvisor – that people only tweet when they loved or hated somewhere – showing tweets from people you know is useful and the complaints feed is fascinating. It makes sense to take mini-reviews from Twitter, because you’ll get a lot more of them (from real people rather than semi-pro TripAdvisor addicts) and they’re likely to be more recent than Yelp ratings, which can go back years to when a restaurant was under different ownership, for instance.
Twizoo claims to be more immune to fake reviews than the big sites, but it seems to me it’d actually be easier to spam Twitter with fake love/hate for a restaurant. Their algorithm excludes things like the restaurant’s PR company Twitter account, but it’s very easy to build Twitter bots that look like real people to a machine.
Speaking of which, Twizoo’s algorithms need a lot of work. Quite a lot of the tweets shown aren’t relevant, and including “jerk” as a criticism of a restaurant is bonkers. The app is also seriously lacking in information about the restaurant (which other free rivals offer), and is currently only available in London in the UK.
On the bright side, Twizoo is working on a tweet-to-book system that would be awesome if it went mainstream. Currently only a handful of restaurants have signed up, but if more join, we’d be all over that. Booking websites and phone calls – ain’t nobody got time for those.
As for Twizoo as it stands, we think it’s a great idea for an app – it’s just not quite cooked yet.