To Survive in Tough Times, Restaurants Turn to Data-Mining | Hacker News Reader

To Survive in Tough Times, Restaurants Turn to Data-Mining

nytimes.com

96 pointsmcone61 comments

Cash is king, especially at local restaurants with humans who recognize their human customers, know their preferences and can adjust on the fly to changing needs.

If the siren call of data is followed, restaurant offerings will be influenced by airline yield management techniques. That won't lead to happy customers.

Great point. "Yes, I know you had a reservation, but we sold your table to someone else. I can't do anything here, but if you call the 800 number a robot will send you a $20 voucher for your next meal."

The funny(?) part is that a lot of customers won't mind if they got a meal voucher for the future. When they don't mind it for air travel they will surely mind it even less for food where there are a ton of other options available.

Could restaurants be successful by giving free food (vouchers) to people who make reservations but have no peference for the restaurant, i.e. are happy to go elsewhere? Sounds like Groupon, another bright idea pitched to small restaurants.

I'd say it's closer to priceline.

I could see it go either way - I'd expect people to hate it more, actually. How fun would "oh hey, we don't have anywhere to get dinner, but we have a voucher" be on a date? Especially a first date?

Might be a positive on a first date, it would add some adventure to it and also increases the chances of a second date as you have to use the voucher.

That actually happens and I don't feel it's wrong. For example:

The pre-theater menu, which offers a discount if you eat and leave before a specific time. I could also imagine that tables booked for a less desirable time window are sold cheaper than when you want it on Saturday, 8pm.

Airline yield mangement gets a lot a flack, but I really don't get what's inherently wrong with the process.

Yield management is a form of constrained optimization, and it's tempting sometimes to relax the customer experience constraints to favor a higher (short-term) yield, as we see from time to time with the airlines.

Most of this article has nothing to do with real "data-mining." They're discussing POS kit that identifies regular customers and what they typically order. Top 10 list dashboard tech.

The "data-mining" is occurring on the POS-side (not the restaurant-side). I suspect the idea is that an analytics firm will do the data work and convert it into reports/products/services/api's that are sold to restaurateurs.

As far as whether or not this constitutes "big-data", it might if they can hook into enough POS-systems.

At one point I was working on a POS proof of concept that would offer the software for free to restaurants to use with the caveat that we could sell their anonymized patron data as reports to restaurateurs looking to gather intel on viability of new restaurant locations. I got bored of it and stopped before I validated the market for it, but I think there is space for this. This was before vendors like square had their products for small businesses, however. At the time, my potential competitors were the NCR Aloha's of the world.

Yeah, wherever there's money, there's valuable information.

POS' are a goldmine if you can acquire data from huge numbers of them (unfortunately that's hard).

That's definitely true. I would have had to operate at a loss for a while until I could enough data to sell. I never did the math to see how long that period is, but I'm pretty sure it's worse than my back of the napkin calculations.

I worked at a fast food place that has that capability in 1994. It could correlate patterns for upsells.

Iirc it was an NCR XT based system.

Growing up a popular Chinese restaurant in my hometown had a system like that. Their main business was phone orders for pickup. After a while when I called they would ask if I wanted what I got last time and if I would like to ask anything. Occasionally they would put an extra item in my order and then next time ask if I enjoyed it. Great place, but for whatever reason they eventually fell out of popularity and closed.

It's putting spin on old technology. This all sounds like something that could have been done in the 70s, and was mainstream in the 80s.

The dashboards are probably way prettier now though.

Shoot, I'd bet money it's been done for decades if not even longer by good waitstaffs.

I know that I've been recognized by employees at places I go to frequently, and media depictions of people being asked if they want "the usual" seem to predate widespread computers.

Slightly dystopian in that regard: let's take something that used to be part of being good at your job and building a rapport with your customers, and outsource it to a machine.

Googling for restaurant notes yields a bunch of relevant news stories about such information management.

Here's one in the New Yorker, from 2012: http://www.nytimes.com/2012/09/05/dining/what-restaurants-kn...

Here's another from May of this year: http://nypost.com/2017/03/30/the-incognito-way-this-restaura...

So there has to be a saturation point, right? If every company you interact with is data mining and every data miner sells information to other data miners about you and there's now a couple terabytes of behavioral info on you, when exactly is there so much data as to render it worthless? Supply/Demand would seem to indicate that as more and more companies turn to data mining, the mined data becomes less and less valuable. Or is this just a symptom of everyone in the advertising business getting by, just by pretending that it's still 1950 and they can target advertising perfectly?

I think you're onto something, and I've been surprised that the data mining bubble hasn't burst yet. I do however, think that that comparison to 1950s advertising is a bit off. Mass marketing in that era, and indeed, up until the late mid-2000s was at best, a scattershot approach with the granularity of a market segment. This promise of today's approach is to target the individual, based on their past purchasing behavior, search history, email and conversation history.

That only works after someone is identified. I started getting a lot of adds for sheds when I did a search for them: I have no reason to think of any one advertiser. Even if I buy a shed (which I might not) only one will win, and it might not even be a company that has advertised. The smart money would have been to advertise to me long before I was interested in a shed - the company that advertises for years sticks in my mind as a company that might be worth paying extra for because of their "reputation" (even though they built this reputation in advertising)

This is the part that gets me (you beat me to the reply, heh): So you buy a skateboard; then suddenly, your pages are full of skateboard ads. Then you buy a new set of wheels; ok, now your pages would likely be full of parts for a skateboard. Then you buy a ramp or something: Now you get all kinds of skateboard paraphernalia, safety devices, hell maybe some scooters.

Given these generous assumptions: * That all these purchases are trackable, by the same advertiser, with definite certainty * That you're even buying online in the first place, so you can advertised to * That you'd respond to advertising at all, because thus far their recommendations have been purely reactive, not predictive of what you'd buy * That an eventual purchase could be directly tied to the advertisers work

After all of that investment of time, money, and energy, if they somehow managed to secure a sale of say, knee pads, how much do they actually make? And how much did the knee pad company pay to acquire that lead which led to a sale?

My basic question is: How is any of this profitable? It seems like an absolutely massive ecosystem of enormous firms all trading money back and forth and trading data back and forth, to at best, make marginal bumps in otherwise happening sales.

This boggles my mind and I would love an enlightened HNer to weigh in if they happen to read this.

"Sell more skating stuff to skateboard buyers" isn't the point of targeted advertising. A common misconception is that the targeting is related to what the customer wants, when it's really about creating ways sellers can target specific demographics. The goal isn't about selling some specific product; it's about hitting people at the specific times/situations they are likely to change their buying preferences.

I highly recommend reading this[1] article for a very good explanation about how this works.

[1] http://www.nytimes.com/2012/02/19/magazine/shopping-habits.h...

Ultimately, personal data is a means to an end, and the end is Selling You Stuff For Money.

Increasing wealth inequality in the US means the data collected about most people isn't going to be worth much in absolute dollar terms. For now, the data is being gathered and brokered for the promise of its future value. That is to say, the market is largely speculative.

FB and Google (and all the miserable adtech invertebrates in between) REALLY NEED advertisers to believe that this data is worth something. They need advertisers to believe that the ads are working. There is little to no objective, verifiable auditing of online ads. There is massive, rampant fraud.

There's a reason why tech CEOs like talking about basic income. If nobody has any money, nobody can spend; if nobody can spend, brands won't advertise, and FB/Google go broke. Unfortunately, in the political climate of the United States, BI is a fart in a stiff wind.

Speculation + fraud + shrinking middle class = one major shitstorm on the horizon...

I knew a location where in the 15 years I lived in the area 10 restaurants remodeled the building and then failed. Apparently no one data mined the history of the building. There are lots of reasons why a restaurant go out of business; often decisions made before the first customer even shows up will kill it.

I recollect reading somewhere that opening a restaurant in a location where another restaurant has failed puts your failure chance at 90+%.

That doesn't mean you can't suitably insulate yourself from the failure and make a lot of money running the place into the ground in the meantime.

It's like a college bar that serves to underage people. It closes amid controversy, changes ownership and reopens with a different visual theme and the same business practices every few years.

You can do this with all sorts of businesses. It's more common in places where people have a lot of disposable income and you can run a bad business for a few years before people smarten up, word gets around and the place folds. Tourist areas, college towns, whichever town is seeing an influx of tech industry money or an oil boom, etc are the places where you can run a bad business the longest before folding.

As opposed to like 85% failure rate within the first 2 years for all restaurants! A lot of the time when you get a location you're also getting a liquor license which is required to turn any profit.

As a diner, I don't want to be tracked, catalogued, or profiled. Leave me to eat in peace with good customer service, and a healthy attention to detail. This system could just as easily be used to decide who gets good service, and the best food, and those the restaurant doesn't see as profitable.

While we may desire to escape The Village, where are you drawing a line on "healthy attention to detail"?

The early diners are dawdling, so your 7:30 p.m. reservation looks more like 8. While you wait, the last order of the duck you wanted passes by. Tonight, you’ll be eating something else — without a second bottle of wine, because you can’t find your server in the busy dining room.

I love to eat out and this isn't a common enough problem for me to think automation is the answer. Times are tough for restaurants because competition is intense, rents are high in desirable locations, margins are thin, and the industry is labor-intensive.

Franchise restaurants like Applebees have already industrialized the production and delivery of hot food, and nobody with a choice likes restaurant franchises - hence all the articles for investors about millennials killing things off.

Just another tool in a kit of a mediocre restaurant on its way out of business. What works for a place that flips $10 tables is not what works for/at Jean George.

That's like saying what works for a motel does not work for the Presidential suite at The Ritz-Carlton. True but not insightful and not relevant. There are far more motels serving far more people than presidential suites, so if anything, things they do affect a much wider swath of the population.

The article lumps together $190/head tasting menu places with chipotle-like restaurants in photos.

No they don't, but this advertisement needed a headline.

...and start selling everything at mediocre quality.

The big fast food chains have been doing that for 10-20 years. This is just moving the technology down to the small shops.

If you only do Continuous Improvement when times are bad, you will die.

"Tough Times" is what Lean Manufacturing calls just Draining the Lake - when the lake is drained you can see the rocks on which your business will scrape. The skill is to examine the lake before it gets drained.

do you really need data-mining to tell you that your food sucks?

But how do you know the food sucks? It wasn't until I travelled and tasted food in other places that I figured out when food sucks. Up to then I assumed it was mediocre. Now, I'm more likely to leave feedback with the waiter. I've had bad service and I've had good service and I know which is which.

There are any number of objective and subjective ways to know, the restaurant just needs to care about it. Off the top of my head:

- actually eating the food from time to time

- checking orders before they go out

- checking orders after they are done, to see if food remains

- checking yelp and other reviews

- being a chef and doing chef-like things

I thought to survive in tough times a restaurant just needed to provide great food at a reasonable price. In fact, I've never heard of a restaurant have great (not good) food at a reasonable price die.

It's tough if your food is just average or merely good, as there are plenty of places with good food at more or less the same price. I suppose the tough thing is understanding the threshold to "great" food and what constitutes a "reasonable" price.

It's always tough times in the restaurant business. The margins are razor thin. Many operate more out of passion than profit.

this is clearly not true in all cases; i think the more likely story is that like most industries, 80% of the profits are taken by 20% of the firms.

also, salaries don't count toward margin so i wouldn't be surprised if most owner/managers make a reasonable living (albeit with long hours) even if their businesses aren't very profitable on paper.

What statement is "clearly not true in all cases?"

tautologies?

Local restaurants that have been open for decades have been struggling over the past few years in my area (east side of Indianapolis). From talking to one owner, meat prices in particular took a dramatic jump ~5 years ago, but I don't know how accurate/relevant that is.

One was bought, moved, and closed a few months later. Another eliminated dinner. A third, open 24x7 for 40+ years, reduced their hours and then closed entirely, although that appears to be as much poor management as anything.

Quite depressing. I like supporting local restaurants, but painful to get attached to something that's likely going to die.

One thing I've noticed is the old places that start going out of style all start to become run down and less focus is on cleaning the place. If a place looks dirty people won't trust the food - especially if you're charging enough to actually make a profit.

The old joke is "How do you make a small fortune in the restaurant business? Start with a large one".

I know a restaurant that is fantastic. It seats hundreds, yet they all feel private. The decor is meticulous down to the last detail. Theirs fresh flowers on every table, they serve dishes from a hundred lands. No dish takes longer than 10 minutes from order time, and, no dish costs more than $20. They have the most extensive wine list in North America. They have free valet parking, in fact, if you don't have a car, they'll send a limo to pick you up and, bring you home, free of charge.

They went out of business though.

Their food sucked.

Well it sounds like they were also jetting money out of every artery.

How can a restaurant have such service (limo from- and to- home??) and have bad food?

I mean, I can imagine the food being "just good" (not great). But sucky food? With such a great wine list?

Something sounds terribly off with the.management of that buisness.

Can you share the name? Or is it something like a chain restaurant?

I think the story is meant to be allegorical.

Ah.

It might be. Didn't think about it that way.

Sometimes a bad location really can sink you. A new restaurant with great food at reasonable prices also may not catch on soon enough for a number of reasons and get dealt a kill date.

> I thought to survive in tough times a restaurant just needed to provide great food at a reasonable price.

Right, so tough times means you can either do the great food or the reasonable price, but not both, and it's very hard to choose one of the two.

A restaurant can have their rent doubled or tripled in a market like NYC, making profit for almost any food business impossible.

Lots of beloved spots have closed down here and have been replaced with condos and drug stores.

Even aside from this, and as others have pointed out - restaurants operate on razor thin margins. A large confluence of variables, some more controllable than others, need to come together in just the right way for a restaurant to have a chance of succeeding.

I think the rise in delivery-only places and food trucks makes a lot of sense in that context.

Clearly the rents are too low even in the West Village. Bar Sardine and Fedora still have audacity to insist that one cannot order a glass of wine at the bar without checking in with a pretentious hipster with attitude at the door when the customer is already inside at the bar just wanting to get a twenty dollar glass of wine. It is as if someone forgot to tell the "management" who is a profit and who is the overhead.

The issue with most of restaurants is that they do not operate like businesses: people who own them think there's something special about service that they offer.

Restaurants fail because their product sucks or the restaurants overcharge for the type of product that they have. That's why a $3.50 for a quart of noodle soup Chinese place manages to stay open in the high-rent area of Manhattan while a place with a $22 burger that comes with a sprinkle of attitude from an "artist" who just happened to wait tables flops.

Peoples' definition of "reasonable price" in my area is effectively a price at which the restaurants have to either sell at a loss or lower their quality. I can't believe how many reviews I see where people call a restaurant overpriced because they can make the same dish cheaper at home. Any restaurant who sells you a meal for the same cost you could make it at home is going to be out of business very quickly.

Reply on news.ycombinator.com