Office Hours with Kevin & Qasar at Startup School SV 2014

All right, so my name is Kevin Hail.

Uh, my name is Cas Unice, and we're Partners at Y Combinator.

What that means really, Billy, is that especially when we're not in batch, we're out there trying to recruit and talk to as many founders as possible.

In addition, we read tons of applications, work with the startups as they go through the batch, and then help alumni afterwards.

Most of what we do is do tons of office hours, which is what we're going to do here today.

There are four different types of office hours that we do.

The kind of first one is group office hours.

We split the batch into subgroups, and we meet two partners at a time with like five or six startups.

Those startups kind of describe, and it's a weekly meeting.

They describe kind of what's going on in their company.

Group office hours are really good in terms of helping you as a founder understand as you go through YC, kind of what tempo your company's growing at, and also see that the problems that you encounter as a startup are actually pretty similar across very different types of companies.

The second type of office hours that we do, which we'll try to emulate here, are individual office hours.

That is kind of when you hear the term office hour vast majority of your time, that's what we're talking about.

The third type of office hour is company office hours.

We'll have folks like Apple or Android or whoever come in, Rackspace, AWS, come in and help companies with technical problems.

And then the last type of office hours that we do are investor office hours.

So folks like Sequoia and Andreessen, etc., will come over and they will talk to companies and that's right before demo day, kind of warm them up and getting ready for fundraising and for demo day.

So with that being said, there's some other things I think that's worth noting about office hours and how to do them effectively.

Well, the first thing is obviously what we're going to do here on stage is not normal office hours.

So we only have 10 minutes with each company.

Usually, we don't do this with 1,500 people inside of the room, so, 1,700.

And then, I think the other thing is we usually have a lot of context with the company, so we have already read their applications, we've got to meet them a couple of times, and so usually we're diving deep on a problem and it's going to be a lot more tactical.

So usually we won't get to do that here.

What we said to all these companies is afterwards we'll go and like talk to them a lot more in depth.

Yeah, we're going to try to avoid creating like a pitch and response scenario and much more along, okay what are problems that you face and then trying to kind of dissect them.

And hopefully you guys will get some value out of that.

The other thing is we make everyone sort of understand really clearly is that in office hours our job is not to tell you what to do.

A lot of people come to us and kind of ask us like what would you do in this instance and stuff.

And a lot of times we will tell you our opinion for sure, but it's usually what I tell a lot of the startups is you're trying to figure out your own heuristic so that you ultimately sort of run the company.

You don't need to come back to us anymore.

I think the thing we like to tell people is like you need to start earning your own equity.

Eventually, over time there's two things that we like to do when we have office hours to make them as efficient as possible.

The first one is we love companies that know their numbers, right?

You'll see a little bit of this today.

We'll ask tons of questions, like diving into the company to try to figure out really clearly, okay this is where you are and then start getting a sense of like this is where you need to be.

Your metrics are kind of the heartbeat of your company.

It's always, I think, off-putting if you don't know very basic things, because otherwise what are we talking about?

And then the last thing is we always prefer that office hours as you're coming to us with a specific problem at hand that you want to have solved.

So if you come to us and you just want to sort of shoot the BS or just like catch up on stuff, like I'm going to be really sort of pissed because like that's a really inefficient use of time for us and for you.

Because ultimately you should be like probably working on your shop and working on whatever growth numbers or KPIs that we want you to be focused on.

And then there's a clear sign that if you're having multiple office hours with us that we're going to be worried if you're repeating yourself.

So if we had talked about something and I talk to you again like a week later or two weeks later, it seems like we're hearing the exact same story or same right, that is not a good sign for you.

The companies that do the best are usually ones that like everything that we had talked about they went and tried and they're either coming back to me to say like this is what worked, or like hey Kevin, like this is the stuff that totally didn't work, here's the reason why.

And then we tried something else.

Okay, with that, let's bring our first company on.

I think it's Divide.

Oh, we had some slides, obviously they were not that relevant.

All right, yeah, give them a round of applause.

You guys are brave for coming up here!

Thank you.

All right, so why don't you guys introduce yourselves and tell us what you guys do.

Um, I'm Aon Shiman.

I'm William Webb.

Uh, Divide is an open-source backend-as-a-service.

So we help applications easily connect to a backend server.

We handle things such as user registration and management, data storage, and querying.

Okay, so how far are you guys live?

Are you in beta, what's your...?

We just launched in a private beta data, or well, public beta technically, on Monday.

And so far we only have an Android SDK.

So we just launched it to the Android Dev subreddit.

And yeah, so far it's going pretty well.

You said applications, so is that web applications or mobile applications?

Right now it's only Android, and we're going to start working on iOS SDK soon.

And we want to do HTML5 and JavaScript as well.

Look, open-source pars, right?

Exactly.

Okay, okay.

Well, just starting there, like how are you guys similar or different from Parse?

Well, the whole reason that we started developing this is because we were using Parse for a few of the different applications that we were working on.

But we kept running into issues of things that we couldn't do with Parse, because you can only do what it allows you with its APIs.

All right, so what are some things that you can do that Parse couldn't?

Well, the main reason, or one of the main key points of what we do with Divide IO is it's completely made to be extendable.

So it's open source.

So as a developer goes, they use it.

We try to provide all the basic APIs that they would need.

But if they need to do anything extending or based off it...

Can you give me something really specific?

It's still like, sure.

Like if you're, say you have a user sends in data for whatever reason and you want to do events based off that.

Or if you want to query like another website from your web server, Parse doesn't give you a really good way to do that.

So, one of the big things is they can't do it or they don't give you a good way to do it.

Um, they do give you a way to do it.

You know more about that, what, on the JavaScript side?

So before, maybe a year and a half ago, they released Cloud Code, which is basically you could create JavaScript functions that you could customize and do things.

So every single time you hit one of those, it counts as an API hit, and then you go from there.

And so I guess the biggest thing I'm trying to figure out is, like, do you guys have users right now?

Some people trying it out and using it?

Yes.

So we don't know exactly how many because it's open source and we just launched on Monday, so...

How many requests are you serving?

That's the thing, like, we don't have... oh, they install it themselves, the whole thing, so it's open source, which is kind of the other, I think, part of the open-source Parse description.

We should probably discuss, what kind of business do you see this becoming?

Because there are other open-source products that become large businesses, so what do you see happening for you guys?

So we're going to try to do the Red Hat, MongoDB model, kind of do consulting services.

So, but Linux is pretty complicated and used by large institutions.

Is your thing super complicated and so hard to use that people need support and pay for it?

Yeah, like literally, that's the business model.

It's so crappy, we will help you figure out how to make it well.

MongoDB is pretty good.

I mean, from our research, we think that companies will pay for support.

Like, they want the option to be able to just pick up the phone and call.

Oh, are they paying for support right now for you guys?

Not for us, no.

But we know that you... I think because we looked up support actually on your site, and it said like go to Stack Overflow.

Yeah, yeah.

So like literally that says like go to Stack Overflow.

And yeah, right?

Yeah, so we do provide like free support because right now we're just trying to build the community.

Yeah, so I think that's the first thing.

Like I think for anything like this, especially if you're trying to build a platform that you get people excited about, you're desperate to just get people to sign up and grow.

And you've got to figure out how you're going to sort of detect that.

But your biggest thing is like I would not put any barriers to that growth and I would be as open as possible to say we'll help anyone right now doing this, especially since your guys' thing is in beta currently right now, right?

Yeah.

Yeah, because it has beta like all over.

So the reason why it's in beta...

Yeah, well, we wanted to wait until we had an iOS SDK to really push the full launch.

But I mean now we kind of...

I'm just saying like it's a backend as a service, right?

And so the last thing as someone that's working on something, I want to see a back end as a service I'm going to build my app on top of is beta, right?

And so like that's going to limit growth in lots of different ways.

So there's all this different kind of stuff that like you've got to overcome.

And I think like you should forgo because you're seeking growth no matter what.

And then once you get a big enough community then you can start saying like, all right, there's going to be some people who are going to be willing to pay for dedicated expertise.

I mean, but that comes way down the line but that's scary because if it's so far down the line...

I mean the kind of general mechanics of a company are you need money to support your company, and then you need to extract value from whatever ecosystem or extract money from the ecosystem.

And that would require for you guys a large community who's dedicated to Divide, and then institutions or folks that are willing to pay, small or large companies that are willing to pay.

That's pretty scary because that could take a long time to get to.

Have you guys thought of any other alternative other than the Red Hat model?

Well eventually we would like to do some type of hosted solution to directly compete with Parse and the others, but we want to make it to where it doesn't have the same limitations.

So we're already like running close to time.

Oh wow, for this?

Yeah, so just kind of end on here, like your biggest thing is like you've got to build the community no matter what.

And that's like the biggest evidence that I as an investor is going to look at for something like this, right?

Is like getting a sense for how many people are using it and are you talking tons of users that are trying it?

Like has anybody even asked you guys about like support or help for this stuff?

Yes.

So how are people using it right now?

From what we could tell, like I said, we just launched on Monday just with one Reddit post.

I see.

So we have like 12ch.

So we have like 12 people who are part of like the beta testing, and we ask for all their feedback.

And so we've already gotten feedback on our licensing.

We already changed that because we use GPL, and we overlooked that you're supposed to open source everything when you use GPL library.

So we changed the license.

We got feedback on our website.

That's good.

I mean, that is definitely the right attitude.

I mean, I think the analogy is kind of when you're building an early-stage company, you're kind of like in this dark cave, and every piece of feedback, whether structured or unstructured, you can get from your users or community, it like informs where you should go within that rather than just kind of headlong running in.

So I think that's a great start.

But certainly a really... yeah, this is just getting as many people signed up as possible and figure out like who are the first 100 people that are going to be super dedicated to this and have built something cool with it.

Because those are going to be the examples that inspire all the rest of your...

And also, I really would push on the support as a business model assumption by trying, even at the beginning, to offer some support and see if someone...

I mean, it's certainly, you know, it's early, but right, you should test that assumption.

All right, guys, I go [Applause].

Thanks.

Have I already lost the clicker?

All right, all right, the next company is Kona.

Kona, and if you forget, it's on their shirts.

All right, similarly can you guys introduce yourselves and then what do you do?

Yes.

Uh, I'm CH.

We're from Monterrey, Mexico, and...

And it's in the same.

So what we do is a mobile app that it's for building your shopping list.

And what we do is we leverage all the information on prices, deals, and indoor location of products.

Let's be a bit more specific.

When you say it's a mobile app for building your shopping list.

So when I open it up, do I just see... what do I see?

You have to build your shopping list or you can copy-paste it from, for instance, your iPhone Notes app or why wouldn't I just use my like post-it or iPhone or notepad?

Yeah, yeah, the difference is that we have all the information on the prices, location of products inside the store and deals.

So it's like actually like any regular shopping list.

So we help people save time and money when grocery shopping.

I'm trying to think here.

So I put my things as a list, and then the next thing you spit out to me is a bunch of stores of the prices of all those items?

Like, yeah, like you have your pre-selected store and wherever you get into another store.

Oh, so the first thing I see is I have to store your store.

Yeah, but, but where, for example, if you get into another store, it will automatically switch to the store you're at with your shopping list.

Okay, so it'll adjust, like, location-based?

Yes, it's location-based and immediately, you will show you deals for you based on that.

Okay, so I first have to download the app.

Then I put on all this information, where are you just kind of peel back what you guys are doing?

Where are you getting all that inventories and pricing information?

So we, depending on the retailer, like we have some agreements with some retailers back in Mexico with some other, because they have the power in the industry.

Although they want it for themselves, like they like okay, I love it, but I don't want my competitors in.

So we kind of bypass that by using bots and updating information daily.

So you're scraping their websites?

Some of them, some of the retailers, yeah, and APIs that are free.

Okay, so some scraping, some APIs, and but the information, the prices are different from the web to in the case.

Like in the stores we working, like depending even the store, each store is different in the same street and depending which is the store that delivers you the product, you have the price of that.

Do you guys have some users right now?

Yes.

All right, so how many users?

And so we have over 11,000 users, and yeah, we started initially... where are you guys launched?

Are you in Monterrey, Mexico City?

Monterrey.

And so basically, yeah, we've learned a lot.

Like for example, the first months, like we bled users, and then we started to see like people started to love more the prices and the deals, and our cohort curve started to stabilize much higher.

So like I'm just going to make an example of you.

So like, um...

I apologize ahead of time.

So you said like, you know I get this all the time; a lot of startup founders, they come and talk to me and they'll say, like, you know, how's your sort of user growth or what have you?

And they'll say like one specific number, like we have 11,000 users, and for us, especially at YC, I'm interested more in like acceleration versus like a specific velocity.

So what I immediately want to know is like what is the rate of growth right now?

So, so like yeah, for the past six months actually our rate has been 9.5% week, every week.

If we, if we kind of like...

Where are users hearing about this app?

So initially, we do it through our network in Mexico, and then we have a lot of friends.

11,000?

No, no.

And then we started to do some advertising, like in Facebook ads and testing which kind of message work better.

How much have you spent on acquisition?

For instance, it goes between 0.1 to 4 per user.

Yeah, okay, so in total you spend like 10.

And then how, for every dollar you spend, you can get up to 10 users?

Oh, okay, sorry.

How engaged are they?

Like how often are they making lists?

And so our active users use us 2.8 times a month.

Yeah, but what is percentage of users that are active?

That's 36%.

Okay, and for instance, that the top 20 of, out of those 36%, the top 25 use over five times a month, so it's more than once a week.

So one question I have is do you actually get anywhere to switch what products they buy or do you like change anyone's behavior?

Because that's the goal, right?

If you have something that shows people prices for stuff or showing them some kind of thing, this is ultimately going to be this recommendation data play.

And so I'm wondering like do you have any evidence that like, oh, someone decides to shop somewhere else or buys a different product as a result of looking at your app?

Yes, so our end goal in the end is to empower people with information to make wise decisions when grocery shopping and save money.

So are they your ultimate customer, like your people buying stuff?

How are you as a business making money?

Yeah, so we're actually making money from brands because they're like we initially tried to get from retailers but they're very long cycles.

And, yeah, they have the power.

But brands love us, because we allow them to reach the right customer at the right time in purchase intentions, leveraging the location of the store that they go in.

And so, yeah, that's what we've been doing.

I do want to go back to the data question only because that’s so critical on this app growing.

So you're scraping stores, and what happens if they don't have any online presence?

No, we don't have those stores.

Yeah, we only have like, for example, in Mexico we have Walmart of Mexico, sure, all the big retailers already there.

There's only...

Yeah, so it's not a high when you use the example of down the street.

There's a lot of retailers.

You're actually not in a lot of retailers, you're just in one.

In no, we're like five of the big retail stores and like all throughout Monterrey, Mexico City.

And we actually started with small stores because the small stores wanted to, but we found out that it's different.

Like, difficult to scale, so we plan to have all the big stores first and then...

Have you found that users have a hard time remembering that your app exists?

I know you have this 36% who are active, but what about the rest?

Yes, actually, like for every, like whenever it happens two months that they haven't used it, we send an email to why haven't you used us?

So yeah, and they've been saying like, okay, you don't have my store yet at my store.

And whenever we add a new retail chain, usage grows and also like, okay, I'm not the guy that goes grocery shopping in my house.

So that's also...

So it's targeting...

At the beginning we target everybody, because we want to see who sticks and now we...

So, ultimate customers are your brands.

So how's that going with them?

Like how are you attracting the first brands, and how are you like growing sort of the brands who are your real customers?

So we have like, through our personal network, we have experience with consumer products and we have connections there.

But then actually when we started, we send a newsletter with the benefits to the brands.

Some stores have come out to us.

How effective is it?

Is it like... I'm sorry.

How effective is like advertising on your platform?

Yeah, so what do you exactly do?

I mean, if so, let's say I'm Nestle in Mexico City, what am I buying?

Yeah, so it's basically two things.

One is the advertising, like targeted advertising, and the other is the analytics.

Because one we give analytics on the top of mind.

Like even for the person who hasn't seen the ad, they see okay next, for instance, this Coke is over Pepsi, and it's over in our product searches.

So we give this right now, like we're surprised, like they like this brand awareness information and the effectiveness of the advertising we still need to prove that, right?

So what are your biggest problems right now?

So the... like, well we have several problems right, like challenges ahead.

One of them is like, uh, so we've been focusing on stickiness right now.

We also want to even grow more viral.

That's one issue, and we've been doing it by improving the product, having a better experience.

So when you say that right, so you want to be really clear like why you're doing the thing that you're doing, why you're focusing whatever attention on.

So what's the underlying reason?

Is it because is this what your customers, the brands and stuff?

Yes, the brands like they want us to have more users like okay, we love it.

Okay, so we kind of went to brands to kind of prove a concept to kind of clear out some assumptions.

But now our focus is go huge with the users in order to have a better offer.

So the biggest issue is definitely post 11,000 users you can get to the next 100,000?

Yes, that's our goal.

Okay, and then on the retailer side, what are you guys going to do about growing from beyond Mexico City, Monterrey?

How do you expand the kind of data input?

So we have a network of, as we said, so we're going to continue tapping them.

We actually have some new investors that are very good in that network.

But after that, we have to build a marketplace where...

Why would you need... no, I mean you guys are just you?

Oh, just to talk to them to get access to their APIs.

Is that why?

No, in order to... well to connect us to tri us, right?

And then after that, in order to grow we have to build like a marketplace where the brands can do their campaigns without us.

Like right now, and we're doing it intentionally.

We're learning what is what the brands like.

Yeah, yeah, I understand that, but the thing is you don't build the marketplace to grow, right?

What you got to do is grow all the users and stuff, and then you build the marketplace to make it more efficient.

Yeah, right?

So like the only focus is like how do I get as many users as possible because the brands will just come to like whoever has all the users.

That's true, right?

So like you just figure out the supply side, right?

And then you can worry about demand, which like obviously seems like that's latent, which is good.

Yeah, so for that, what I'm trying to think of is what is the mechanics for growth right now?

Like how are people hearing about your app and like downloading it?

That's clearly the biggest issue.

You said some Facebook and stuff like that.

Is it cost-effective?

Like for acquiring users at 40s?

Yeah, with what we have been paying, like the value that... the reason that doesn't, that formula really doesn't make sense, the 10 users for a dollar is because not all 10 users end up staying, right?

So you're really paying for...

Yeah, exactly.

So it's like a third of that efficiency?

Yes.

Um, and that doesn't really scale in the long term because then if you equate that to the amount of brand dollars that will be spent for those limited users, you're just going to be throwing a lot of money out.

And you hope you get to the point where you don't have to advertise, but there's nothing fundamentally in the product which would get you away from that.

Another interesting, it's almost like you're structurally setting up the product to continue to lose money, which isn't really...

Oh, that's not the plan.

Yeah, and for instance, like we, our target, like our pricing is based on what they currently pay for in their marketing.

Yeah, no, we're not like... so we're comparing to inter, like to internet standards of advertising, we're much higher, like much more high, very cheaper.

We’re totally out of time.

Yeah, but it is great that you're going for brands.

I mean, I think that makes sense.

I think the biggest thing is you've got to figure out like what is this trigger that gets people to say like, oh, when I need to have a shopping list or I've got to figure out what's the best prices for stuff, like how do I think of Kona to do that?

Or how do I get my current users to tell other people about it?

That seems like the only thing that you should be like focused on right now because paid acquisition is one of those things where it's like everything else already works.

And when we add paid acquisition to it, it's like throwing gasoline on a... if all this other stuff doesn't work, then you're throwing like water in.

Yes, and we do have organic growth!

Awesome, thank you guys!

Thank you!

Thanks for coming up here.