Hacker News Reader

Comment by kevindong | Hacker News Reader

Very important note (near the bottom of the post) on why Doordash did what they did:

> Note 1: We found out afterward that was all the result of a “demand test” by Doordash. They have a test period where they scrape the restaurant’s website and don’t charge any fees to anyone, so they can ideally go to the restaurant with positive order data to then get the restaurant signed onto the platform.

View on HN

Can you imagine getting that call "Hi, little Pizza place, we've been fraudulently impersonating your business, dragging your brand through the mud and pissing off all your customers selling your pizza at a loss, clearly we're the sort of people you need to do business with"

I think there's a huge grey area between "hire a 3rd-party person to pick up your food and deliver it to you" and "impersonating your business."

Like DD, GrubHub, or whatever aren't pretending to be the restaurant (shady website bullshit notwithstanding). They're just saying that they can buy and drive the food to you on your behalf.

Absolutely they are. They'll set up phone numbers and answer as though they're your restaurant. They'll set up websites (which I think you allude to? "they aren't pretending to be the restaurant... other than setting up a website pretending to be the restaurant"). They'll pretend to be the end customer if they have to.

It doesn't fully explain why they priced a $24 pizza at $16. I wouldn't be surprised if they're subsidizing purchases, but just skipping fees doesn't explain that.

Web-scraping is hard man, especially with mom-and-pop restaurant websites that are often exported straight from microsoft word or some ancient "platform" that got reconfigured 20 times over.

As article pointed out it picked up full-toppings pizza as plain cheese.

But they still paid full price to the restaurant! Someone in doordash knew the real price at some point then!

The full price was paid because a door dash card was used to pay for the order on pickup.

Which should have triggered some sanity check indicating that their scraped data is way off.

Maybe there are so many triggers that they can't keep up.

In the end you need someone to look at those alerts, maybe they just got missed.

They wanted high orders for the negotiation when getting a new restaurant signed up.

"Hey in the last month we delivered X amount of your food! If we don't list you then you'll lose those sales."

It could be so easy to game this.

Table Column A, menu items, lowest to highest in cost.

Table Column B, prices, highest to lowest in cost.

Naive scraper associates rows as menu item and cost.

You use CSS, etc., to rearrange things correctly. People looking at your site get info as intended, scrapers have problems, and it's only a dark pattern to them.

Surely it's cheaper to just pay someone to do data entry at that point.

Cheaper, less sexy, and doesn't justify the jobs of several dozen engineers, a few managers and a director.

If you're the director, which would you go for?

But that would that get VC funding? It's not their own money they are spending.

"My second thought: I knew Doordash scraped restaurant websites. After we discussed it more, it was clear that the way his menu was set up on his website, Doordash had mistakenly taken the price for a plain cheese pizza and applied it to a 'specialty' pizza with a bunch of toppings."

Lets them inflate the order count for the pitch.

They are taking a loss on every order, but they are hoping to make up for it with volume.

You can't make up for it in volume if each order makes a loss and you get no benefit from having more orders (they still pay each time the full price to the restaurant!)

(that's the joke)

Reply on news.ycombinator.com