Indeed. If free-text SMS was an ideal way to interact with online delivery and concierge services you'd think someone would have done that already on an existing service. So then, why can't I just order from Seamless or Instacart via SMS?
Oh right, because that experience sucks mightily. I have this lovely full color screen that can display menus, location and delivery times, let me search by cuisine, choose various customizations, etc etc. Then a "magic" computer sends the orders to the restaurant via an already configured and tested system so they are expecting an order and can fill it quickly and efficiently.
Or alternatively, thanks to this innovative new startup, I can SMS to some random number and say "I want food or something do you have any burritos" and then have them text me back, etc, etc.
BUT WAIT says the new startup. IT'S NOT JUST BURRITOS, in fact it's not just food. You could ask for a burrito today and a flight to Cancun tomorrow, and you only have to put one phone number in your address book.
Aha, the real value proposition emerges. This startup uses their proprietary technology to know that Seamless or Instacart or Expedia exist already, and then they use it for me. Except it takes twice as long, without the UI/UX those services have perfected, and costs 30-50% more money.
Step one: Landing page
Step two: Corner the market on taking SMS messages and using them to punch orders into existing apps, using skilled college graduate US based labor.
Virtualized personal assistant services aren't really anything new. This one just had an interesting/novel/awesome (and probably unsustainable) approach to the onboarding, UX, and price model - instead of "hiring" a "virtual assistant" and doing a bunch of sign up forms and paperwork, you just punch an address and card and away you go.
As for why people want a personal assistant to shop for them when they could just use an app, think about it this way:
To get things (even online), I have to:
* Decide what *specifically* I want (brand, item, size, etc.)
* Decide where to get it from
* Obtain their website or app
* (often) Sign up for an account
* Enter a bunch of information to order what I want
* Follow up if it's late or wrong
With this service (or a hired personal assistant) I have to:
* Decide what type of thing I want
* Tell someone to make it happen
Also, I wouldn't call Instacart or Expedia's interface "perfected." Thanks to the way the industry works there's still not a single website where I can say "find me the cheapest flights from point A to point B in date range X" and have it actually do so. With Instacart, I can't just punch in a natural-language grocery list and say "get me decent quality options for each, but make the beef grass-fed please." I have to go through a huge array of brands and potential sources for each item and decide what to get.
Because seamless/instacart/hipmunk/service X aren't individually incentivized to use a generic interface. It's the same reason you have a gazillion logins, each one wants to be the One True Source. (After years of struggle, FB/Google/Twitter managed to hammer that down to "just" 3 choices.) In addition to the 20 apps a user needs, there's new ones they don't know about ("Where do I get flowers again? Car parts?").
The value proposition: People can fire-and-forget an order for anything. Someone else handles the details. People with more money than time -- a great market to serve -- prefer to trade the former for the latter. Cheapo McCouponClipper will sift through the apps individually.
Do you go to the grocery store and tut-tut people in line? Don't they know they can get batteries cheaper at BatteryWorld? Vegetables cheaper at the farmer's market? Windex cheaper at Bob's Industrial Cleaning supplies? Why are they at a general-purpose store with a larger markup?
But it doesn't matter. The incredulity of objectors who ask "Why don't you just install 20 apps and save some money?" will be superceded by people who just want to use the service.
Ignoring cost, and assuming I already know what I want, I would much rather be able to do all my ordering by texting or calling a real person. Even the best, most streamlined ordering UIs are clunky and time consuming compared with "I want a cheese pizza from Johnnie's"
I think what you, and the parent article, are responding to is that it's really hard to actually have an always on-call, fast-acting butler at your service.
I've launched something (once) that went from 0 to breaking in less than 10 hours. I aged at least a couple of years and it took weeks before we managed to get a grip on the situation. We were about as unprepared as could be. This is what all of you starter-uppers are dreaming of, but when the growth is higher than you can handle you're going to have to move really fast if you want to capture your potential. Unfortunately we had no idea what we were doing so I'm pretty sure those first weeks cost us dearly. Fortunately for the magic people their scaling issues are something that can be overcome and I really hope that they'll do well because of this.
Cut them some slack and think about how hard it would be to pull this off with such rapid growth without having any delays or issues during the first phase of rapid growth. Dealing with high levels of traffic and orders is hard, dealing with high levels of growth is a lot harder still.
Like I've said elsewhere, it's not the MVP-nature or the beta that's really the problem. It's that they want to start a service that charges 25% more in exchange for selecting which service to use, with a significantly worse user interface. It took me much longer to type in the order on my phone that it did to select the order from Postmates, the service they likely were using to actually do the delivery.
Even if everything had gone perfectly and I hadn't spent over an hour waiting for responses, I still wouldn't use this service. To pay more for a worse interface? No thanks.
At the very least they should've set up an auto-reply that explains they've been flooded with unexpected high load, with an average reply duration of half an hour or whatever it was at the time.
But in this case, waiting an hour and a half for a middle man to do what it would have taken me 2 minutes to do wasn't worth the extra $9 they were asking for.
Why'd you rely on a very beta web service being offered to the public for the very first time to do what you could have done in 2 minutes?
I didn't make an order, but I fail to see how you could have been surprised by this. They have no idea how much demand they were going to have at launch, and it's not a service that can scale easily.
I didn't rely on it. I tried it out to see how well it would work.
When I first signed up, they wait-listed me. Then, a few hours later they let me (and apparently everyone else) in. They didn't need to do that, but they did.
Your objection is probably more powerfully expressed as "waiting for a middleman to hire another middleman".
Of course, if you're intimately familiar with all the various services available to you in SF then this is not going to be a useful service. I mean that perfectly straight, not sarcastically; they basically tried to charge you $9 to "know Postmates existed". But there will be a lot of people who may still be willing to pay for not knowing all those various services. I'm personally skeptical but it's at least worth a try.
(I can't speak to the complexity of keeping track of the SF services, since I don't live there, but my impression is that it's challenging.)
I bet a major cause of the traffic surge are just curious people like you that aren't even interested in this service, or jealous people that are intentionally pushing the limits and trying to make them fail. In fact, the fact that you're slandering them with an HN post might suggest you are more the latter. I am not in the market at all for this so I obviously haven't used it, but hopefully for them they can identify and keep their legitimate customers (not you) happy.
Note that the "Magic is a beta product released less than a week ago" excuse is irrelevant.
If Magic wants to be competitive, it needs to have services equal to or better than existing options. (The Postmates comparison mentioned in the article is valid). That's the way it is with all startups and "they're just figuring themselves out" is not a sympathy granted by the marketplace.
I disagree. They launched a beta to see if anyone would use it. If no one used it, then they could cut bait and move on.
But people did use it, enough that it caused them to be horrendously slow. This means they should put more money into making it bigger. If there are a few hundred "innovators" (in the Everett Rogers/Crossing the Chasm sense) who were so alienated by bad service during the beta that they will never ever use it again, even when it becomes a good service...well, that sucks but has no bearing on the ultimate success or failure of the service.
If you're going to beta test something, you have to expect it to have a few flaws. That's the point.
> Note that the "Magic is a beta product released less than a week ago" excuse is irrelevant.
Maybe in disneyland, but that's mostly how things go down in the real world. It's a startup, first week, got hammerred by HN heavily, probably others, that's as much as they can handle. Things will go fine, they just need the time to adjust.
The story is about a very disappointing first user experience and how it led me to never want to use the service again. The burrito is just a side plot.
I think that the 'magic' people were simply not prepared for a runaway success. Wait a while to give them a chance to get their act together and try again.
First user experience during the hockey stick phase being 'poor' is to be expected for anything with this degree of growth. Try scaling anything through three orders of magnitude in 24 hours and you'll get why.
The amazing thing to me is that your order got delivered at all.
> I think that the 'magic' people were simply not prepared for a runaway success.
But they were. Because they were using an invite-only signup to limit the amount of new users they can take in, they could make a reasonable assumption about how busy they could get at rush times.
Besides, the people on the other side aren't robots - at the very least they could've apologised for the late replies and explained about how busy they were. Because they're humans - or so the website claims - and that means they should show a bit of humanity and empathy and whatnot.
TBF I don't really think they hire actual humans, that just doesn't scale. Especially if 90% of their users just want to order food - reasonably predictable input, that.
First, they had an invite wall when I signed up. I was ~4000th in line, but they let me in anyway. This was their first mistake.
The bigger problem is that their service is just a simple middle man. They don't provide delivery services, they just choose which service to use. For the examples they gave (food delivery, grocery delivery, airline tickets), it's never going to be worth it to me to have them charge me an extra 25% to pick between the three services that will do that for me (Postmates, Instacart/Postmates, Hipmunk). What's more, SMS makes it more difficult to order. Postmates' interface is much better than trying to text an order in.
It's not that this is just an MVP. It's that this service really only takes away the decision of which other middle man to choose, and there's just not enough value there.
On one hand, I get where you're coming from, but come on:
"Wow, everyone! This is truly amazing. We released Magic under 48 hours ago as an experiment and a side project, and now the traffic and requests are streaming in faster than we can handle."
I think the MVP advice to startups is wrong. It is probably given by investors who would like to see many startups that fail fast and pick winners who survive the first round.
It should be MVP + metrics + quiet A/B testing and iteration for months before launch.
No, this is exactly how early-stage startups often work. It's a great lesson to anybody who hasn't experienced it.
Planned launches rarely go as well as planned, and big visibility happens when you are the least prepared. Been there, experienced that.
According to the team, this was a side project, they had put a crappy landing page out there (with Times Roman as a font and got some flak for that) and somebody else submitted it to Product Hunt and it took off. They quickly improved the landing page and started handling orders.
Now almost everybody reading HN knows them. They chose a brand name that is easy to remember. If they get their shit together in 6 months and relaunch, they will be totally fine. Most early stage startups can only dream of the kind of visibility they got.
The bigger problem is that this kind of surprise success doesn't necessarily validate the idea, but at least now they can see and start to measure retention.
Despite the humour and, on the surface, being a story about ordering a burrito, I think this is actually a very useful 'what not to do' cautionary tale for startups. They've clearly got a concept which, on paper, sounds like a good idea. They've then rushed development and gone to market without actually testing enough to know that they can deliver (literally) what the customer wants. That kind of early0stages growing needs to hapen before the company can gain a public reputation.
That's simply a function of growth. You can not prepare for runaway growth like this, it's easier said than done and from the sidelines it looks like a piece of cake.
At this level of growth any action that you did not automate away will bite you in a terrible way. And that's what all the handbooks say right: wait with automation until you've found your product:market fit. But what if that outruns you?
I agree, but in this context I think that it's a valid criticism as their business model is a sort of 'all things under one roof' delivery system that should be prepared to offer a simple delivery like this one ate a competetive price/timeframe. I'm definitely not saying that people starting a business, particularly one in a service capacity, shouldn't expect to have to adapt with growth. I suppose the main point I was making is that prior to launching you should ensure that you are fulfilling your core functionality to at least a market-comparable level.
I guess I am wondering how Magic is supposed to succeed. Most sites I need to go to tend to have very custom means to make ordering efficient, in the few cases where you might not know where to order something from a quick search is all that takes?
Who is their target market? Let alone if I saw another service hitting up my site like this I would likely try to find out why customer's weren't finding it on their own or using it on their own, two things which would be at the top of the list to improve all the time.
Perhaps to give people a separate billing point to reduce exposure of personal information? Look I just paid thirty bucks for a pizza and Pizza Place will never know it was really for me!!! Seriously the only convienence I can see if reducing the number of entities with my CC info
The latency is certainly understandable for a product that just went live to quite a bit of promotion... but considering that they included an invite barrier to keep things from outpacing their infrastructure, there's obviously a pretty severe point of failure here.
For some contrast, a first-time user could fire up the Domino's app and have a pizza delivered at the regular cost (+ tip and regular delivery charge) in 30 minutes.
Oh right, because that experience sucks mightily. I have this lovely full color screen that can display menus, location and delivery times, let me search by cuisine, choose various customizations, etc etc. Then a "magic" computer sends the orders to the restaurant via an already configured and tested system so they are expecting an order and can fill it quickly and efficiently.
Or alternatively, thanks to this innovative new startup, I can SMS to some random number and say "I want food or something do you have any burritos" and then have them text me back, etc, etc.
BUT WAIT says the new startup. IT'S NOT JUST BURRITOS, in fact it's not just food. You could ask for a burrito today and a flight to Cancun tomorrow, and you only have to put one phone number in your address book.
Aha, the real value proposition emerges. This startup uses their proprietary technology to know that Seamless or Instacart or Expedia exist already, and then they use it for me. Except it takes twice as long, without the UI/UX those services have perfected, and costs 30-50% more money.
Step one: Landing page
Step two: Corner the market on taking SMS messages and using them to punch orders into existing apps, using skilled college graduate US based labor.
Step three: VC/YC EXPLOSION
Step four: Profit