Comparing Heroku and Instacart Screw Ups | Hacker News Reader

Comparing Heroku and Instacart Screw Ups

jakeheis.wordpress.com

31 pointsdark_knight17 comments

Wanting companies to be humble, of all possible things, seems like insanity to me. Companies are not people; they do not have emotions; they do not participate in psychosocial guilt/submission demonstrations as a way to signal status transfer.

A (B2B) company is a machine you rely on to run your own business--sort of like an office printer. When the printer stops working, do you expect an apology? The best thing the printer can do is to just start working again--to make you money now. Everything that has happened in the past is a sunk cost. Time spent apologizing is time spent not printing. (Imagine how much time Mr. Wiggins has spent giving people apologies instead of making Heroku work better.)

"Peace of mind" is for consumers. When you run a business, the only thing you should think about the B2Bs higher in your supply chain is "wow they're good at what they do." Whether they whimper believably will not help your own business succeed.

"Companies are not people; they do not have emotions; they do not participate in psychosocial guilt/submission demonstrations as a way to signal status transfer"

That is a grossly incorrect statement. Companies are built by people. As a founder of the start up who was involved in this screw up, I can tell you that we were very emotionally involved. And, we did feel the guilt.

I didn't say that there aren't people in the process--but you shouldn't be letting emotions like righteous igdignation that someone didn't apologize enough dictate the decisions of a business if your goal is to make money. If you are running a company, either you're signing up to be one of those "economic rational actors" game theorists talk about, or you're signing up to fail.

No, but what you are looking for is an indication of future risk. By offering a full refund, Instacart demonstrated a strong commitment to making sure it never happens again by demonstrating that they would incur serious financial penalties upon themselves. Heroku, on the other hand, have made it impossible for me and others to determine whether they are victims of ignorant customers shooting themselves in the foot or whether those customers had a legitimate complaint. Sincere explanation or grin-fucking. The complaints against Heroku seem technically sound, but without me actually using Heroku, I'd never know. I am a potential Heroku customer. After that "apology", I have to add to my pros/cons list a risk that the complaints were valid. Had Heroku taken Instacart's direction, that risk would be mitigated.

" By offering a full refund"

They only singled out the 71 people and offered to refund them if customers reached out to instacart. If they were really committed to making customers whole, they would proactively refund the charges and refund the fee (and then ask the customers if they would like to continue with the service).

> When the printer stops working, do you expect an apology?

Actually, the better thing a printer can do is explain to me _why_ it failed so I know how to avoid, work around it, or understand what to do next. I feel the FAQ is tons better than "sorry". My printer already says "sorry, print job failed". Try again, "sorry, print job failed". Then it is me cursing at it "piece of shit! tell me why it failed, I need to get my stuff printed"..."sorry, PC load letter"..."Arghhhh!"

So actually I like Heroku's response more. Technical info for technical users.

Exactly. This is part of the make me money now idea: you need some information from the printer to get back to using it, so to make you money, it has to first teach you what you need to know to use it. It doesn't have to bow and scrape while doing it; it just needs to get you back to it being useful.

The problem is, you only got the FAQ after a good while and without any decent apology from Heroku.

In your printer example, it's like your printer is not printing a 0 at the end of each of your invoices, and 2 months later tells you it didn't print the 0 because it didn't have enough ink so it skipped the 0 to save on ink.

It doesn't seem like this is a fair comparison because the target audience is so different. Were I a consumer using Instacart, I'd just want the situation to be fixed, and I wouldn't care so much about the technical details. If I were burned by the Heroku issue, I'd likely be technically oriented developer and would want to know exactly what happend, why, and be assured that it wouldn't happen again. It seems like both companies did a fair job of communicating with their target audience.

I feel that the magnitude of money involved should be weighed in this comparison. Instacart was talking about something like 70-odd people with about a $150 per person membership, which adds up to $10k. Whereas Heroku-gate is probably worth hundreds of thousands of dollars, maybe even millions.

The post you cherry-picked doesn't show the whole context. You're ignoring an earlier post where a Heroku GM apologized profusely. Read the 2nd and 4th paragraph:

https://blog.heroku.com/archives/2013/2/15/bamboo_routing_pe...

> Disclaimer: I was not affected by either of these companies’ mistakes; this post it just comparing Heroku’s response and Instacart’s response

> Too bad Heroku. You just lost a customer.

So did you/he or did you/he not use Heroku?

It seems like you/he didn't in the start, so how could heroku have lost a customer?

Sorry, I definitely should have been clearer there. What I meant was that while I was not directly affected by Heroku's problem, their reaction surprised me so much (in a negative way) that I decided to stop using their product.

I am a mere small fry employee at Heroku. I nearly exclusively work on the efforts concerning Postgres.

If you stopped using Heroku's product because you thought it would not pragmatically meet your needs: I'm all for that.

However, by description, it seems though you were more unhappy with the handling of the situation than the quantitative implications to an application. So, I'd like to expand on that line of inquiry:

I'd like to think that trying to expose the implementation problems at Heroku would allow for more well informed decision making (e.g. what is the problem, how severe is it, what barriers exist to fix it, how long might it reasonably take to improve things, how can it be mitigated in the meanwhile) rather than being seen as only apologetics.

I realize the point of Heroku is to abstract a lot of detail, but all abstractions are leaky, and it seems like people might want to know more, especially as ongoing experimentation yields information.

I don't personally feel an opaque mea culpa is the most useful way to go about this. Do you disagree? Is there some other path you would have thought better overall, or in all respects altogether?

In general, I compare this to the many support requests I have responded to over time about things Postgres does not do, or could do better, and (very roughly, unless already committed) how likely it would be that they would receive attention in a semi-near-future release. Generally, people seemed happy to be informed of that, but the dynamic is very different there. Nevertheless, it's the closest thing I've got in my immediate experience.

You must not be a heavy user if the decision was that easy.

You make it sound like an irrevocable addiction.

Hah yeah, the common usage of "user" between software and drugs is a long-running joke.

Reply on news.ycombinator.com