State of the Seed Market in 2017
tomtunguz.com
tomtunguz.com
The second seed round, makes me even more suspicious of this: are we simply renaming A and B rounds to seed 1 and seed 2 rounds? The increased size of these rounds would seem to imply yes.
Furthermore, I wonder about the lack of stigma in a second seed round. It seems like this would happen a lot if the founders didn't raise as much as they needed, and investors felt they could get favorable terms, perhaps more favorable than the first round of investors - which could mean the exact same terms, but operating on additional months of performance data.
To me, it seems like most of the changes in the State of the Seed Industry are about definitions and semantics.
The other element in the landscape is the huge increase in accelerators, AngelList and the like. There are now a lot of options to get that first $100-200k to prove out a business idea.
Yes, but I don't think it's so much that they wait as that there's more competition. People are, in a sense, working longer on their application for funding and since they are further along (and cost have risen in general) they need more money.
Small sample size and all, but if this is the way things are going, it'd be a good thing and force more founders to focus on metrics that matter early.
E.g. this scenario is entirely plausible:
1) Friends and family
2) Accelerator
3) Some combination of pre-seed / angel (depending on whether it's individual or institutional money)
4) Seed
Seed is basically just the new A.
The amounts offered at the seed stage aren't even enough to cover a year's salary/RSUs/other comp for someone with a bit of skill.
The time it takes to raise the seed round isn't time spent building a business. And with that time reallocated, someone technical can probably just save the money at a job or do consulting on the side while doing their startup.
I'd love to raise money for my startup, but every time I look at the numbers, the numbers just don't add up. If I go spend time fundraising, it may or may not work out. I'll definitely be out a bunch of time, and if I succeed I'll be out a bunch of equity. But the amount I get paid for that equity is problematic, especially considering I'll be expected to work below market for a while on my startup.
So instead investors simply don't get the opportunity to invest in what what, in my obviously biased opinion, is one of the few startups that could make their portfolio work.
They have a really good idea for a startup, but it's a lot of work to get it to the point where they can start charging for it, after investing 6 months of time they have a prototype and they go to raise a seed (last time one of them did was in 2011) and are basically told that even though the idea is great and the team is strong, a pre-revenue seed is way harder these days.
So they quietly raise $100k to get them forward calling it an "Angel" round, with the idea to raise a $1m to $1.5m seed later in the year, which you can call a second seed if you want to. If they succeed, the average seed amount goes up, if they fail their smaller seed never happens.
I think investors have learned that revenue is worth twice the valuation and family / angel / incubator money is filling in the earlier stage.
Overall I think it's probably for the better, provided that everyone understands that it's happening.
Aka 2015 seed round = 90's A round, as many have observed already.
I've also got about 30 capsicum seedlings and two lychees, which are coming along nicely. That was a substantial seed round, Series B. Recently I added to my seed empire by planting three avocado pips - they will take some time to grow. I think that makes them Series C.
These are all heirloom and organic (actually, veganic) and will reduce my food bill by a tangible amount, which is great because my investors are demanding and hungry for more. Where I live, organic produce is exorbitant.
Of course should the seed market tank, I've got a hedge in that my day job is primarily focused on designing distributed data pipelines and data-related architecture.
>> planting three avocado pips“Ungrafted” trees (like those grown inside from seeds) rarely produce fruit. In order to have an avocado tree that produces fruit, one must graft the seedling."
https://www.willsavocados.com/index.php/grow-avocado-treeI'm taking a portfolio approach, whereby I'll average in over time until I'm holding about 20 or so plants. Worst case I'll have some lovely shade and beautiful green trees.
Has that changed? If you're raising a million from your friends, that's rather a well-heeled bunch you hang with.
So if a new technology is introduced to make small teams or single founders very productive, then the size of seed rounds will go up.
(Assuming that the team needs money at some point for something.)
Let me guess: The usual criteria are all the first technical work done, traction significant, and traction growing rapidly. That's not planting or watering seeds; instead, that's a sapling or small tree.
Indeed, for a Web site, costs can be so low that by the time there is significant traction, there is plenty of ad revenue for rapid organic growth and no equity investment needed.
A useful seed round could be $50,000 or even $20,000 except for the legal and BoD expenses!
But that doesn't really square with how others use the words.
But that little bit of arithmetic is well beyond nearly all equity investors.
https://en.wikipedia.org/wiki/Directors_and_officers_liabili...
But really what I meant was that, when there is a BoD meeting, say, monthly or quarterly, the members need funds for travel, food, and lodging, and usually the company pays for those. Just one BoD member flying first class and staying at a high end hotel once a month could eat into a $20,000 seed round in a hurry.
Or, one or a few co-founders, living cheaply, on their own savings, each with a $1000 computer, typing in software can be cheap and where a seed round of $20,000 to $40,000 can do well for Internet fees, publicity, maybe cloud fees. That's what I had in mind.
For those small seed rounds you describe, you're better off self funding and saving up the money yourself.
Of course things change once you need to hire outsiders. But that's what the A round is for.
My guess is that the tune is called by the limited partners who are much more comfortable investing in growth capital.
Now, if you mean that the actual costs to keeping up a website can be covered by ads, then yeah, I guess. But founders have to pay rent and eat, so you're now doing this nights and weekends, and the economics of ad revenue are still the same, except now you can't focus on it much so it takes even longer.
And calling a startup profitable without accounting for the cost of founder time is really misleading.
Other than passion-fuels content creators, which are their own category, my observation is that there are very few ad-revenue-driven startups out there that have raised under six figures and reached any kind of success.
Conclusion: raise a very small round or bootstrap, do the work yourself, and charge actual money for your product. Forget ads. Once truly profitable, raise money if terms make sense.
Nearly every effort at powered human flight failed until the Wright brothers did some good engineering.
Actually, in practice, the probability of success means next to nothing, even if it is tiny. Instead, what does matter is the conditional probability of success given additional information. A great example is the probability of an airplane surviving after flying over Baghdad early in Gulf War I. The probability was low. But the conditional probability given the engineering of the Lockheed F-117 stealth bomber was quite high, and, indeed, the F-117s did nearly all the early bombing of Baghdad and never got even a scratch.
I've carefully planned my startup.
On ad revenue, the Mary Meeker reports from Silicon Valley venture capital firm KPCB have long claimed that for ads on Web sites (not merely for mobile devices), can expect revenue of about $2 per 1000 ads displayed. Sure, may get paid by the click, but can still get an average of $2 per 1000 displayed.
Each of my Web pages is just 800 pixels wide with everything positioned to the last pixel with tables. The pages are all just dirt simple; the fonts are large; the contrast is high. There are no icons, rollovers, pull-downs, pop-ups, or overlays. The HTML and CSS are dirt simple and need only a standard Web browser up to date as of about 10, maybe 15, years ago. There is nearly no use of JavaScript, and the pages load nearly instantly and never jump around.
There are both horizontal and vertical scroll bars, and the pages are still usable in a window as narrow as 300 pixels. So, the pages should be fine on any client device from a high end workstation to the cheapest smartphone.
The UI/UX is simple and intuitive, and a child in, say, rural Thailand who knows no English should be able to figure out how to use the English language version of the site from about 15 minutes of experimentation or about 3 minutes of instruction. The Internet data rate a user needs is usually meager.
For the ads, there is a banner at the top of the page in the standard size of 728 pixels wide and 90 pixels high. Down the right side of the page are some ads, at least four, sometimes more, say, an average of five, in the standard size of 300 pixels wide and 250 pixels high.
The site is intended to be of high interest to everyone on the Internet -- children, adults, in the US, and around the world. The results are all not just safe for work but just squeaky clean, proper, uplifting, educational, informative, artistic, engaging, entertaining, especially entertaining, etc.
The site should be especially popular with relatively good user demographics. E.g., the NYC MOMA crowd will love it! But so will the girlfriend of Joe Sixpack.
So, let's see, assume:
Get, on average 24 by 7, one user a second and a peak of about twice that.
Due to the UI/UX, each user sees on average 10 Web pages (it's an interactive UI/UX, fun, awash in variety, something like an adventure, for some users addictive).
Uh, there's no login and no use of cookies, and the site still works fine with JavaScript disabled in the user's browser. The site is one of the best on user privacy: After a user's session state times out, nothing is kept on the user. No background data is used on a user. Still from the UI/UX, there is some good data for ad targeting; I have the math worked out but have not written the code yet; I don't intend that ad targeting yet.
People will want to come back for more and also tell their friends and send results to their friends -- right, virality.
Each page contains one banner ad and on average five of the other ads for an average of six ads per page (the banner ads may yield more revenue per 1000, but I am neglecting that).
So, for assumptions, we're talking on average one user a second; each user sees 10 Web pages; each Web page has six ads; and the revenue is $2 per 1000 ads displayed.
So, in one month the revenue would be
1 * 3600 * 24 * 30 * 10 * 6 *
2 / 1000 = 311,040
With my software timings, one good server computer for $1000 (I just did some careful shopping) say, AMD FX-8350 with eight cores at 4.0 GHz and 32 GB of ECC main memory, on average half busy 24 by 7, could send the average of 10 Web pages a second.
For the Internet upload data rate, each Web page sends for about 400,000 bits (the ad networks send the ads; my server merely sends the URLs of the ads at the ad servers). So, for a peak of, say, twice the average, that would be a peak data rate of 8 Mbps, now trivial for an Internet connection, say, just from next to my left knee, over coaxial cable, where the big deal is just a little extra for a static IP address, which my ISP is happy with -- they do encourage their customers to run Web servers.
So, maybe I'm off by a factor or 2, 4, 5 or so in places. E.g., maybe I'll need more than one server: So, as the usage and, thus, revenue from one server starts to grow to put too much load on my first server, then, quickly, in, say, a week, there will be enough revenue for a second server, then 10 servers, etc. However fast the usage grows, if the usage does grow, it will throw off plenty of revenue for more servers.
Sure, the main issue is just, will the users come. I'm betting they will.
For supporting me and my kitty cat, say, food, clothing, shelter, etc., I should be able to get that out of the $311,040 a month or early on the $200,000, $100,000, $50,000, $20,000 a month.
And with traction enough to interest equity investors, I will be nicely profitable in every sense with plenty of retained earnings for capacity growth, a new Corvette, etc. So, net, I see no possibility of equity funding -- by the time the equity guys are ready to write me a check will be too late for me to accept it.
For example, you're using $2 CPM as a baseline, then just multiplying by your 6 ad slots. There probably aren't very many broad audience, high traffic websites out there getting consistent $12 CPMs. That seems really high for any kind of entertainment, viral site. I think $1-2 would probably be more realistic, especially since you're using ad networks instead of selling inventory directly.
Also, I suspect 10 pages per visitor is high, but maybe not.
Finally, it's easy to say "one visitor per second", but that seems kind of arbitrary. Why not ten per second or one per minute? It might be harder than you think to build up to 25 million pageviews per month.
I think you've thought this out well in some regards, but this could very easily turn into something that gets a tens of thousands of visitors per month and ends up making a few hundred dollars, not a few hundred thousand dollars.
Good luck though, I really do hope it works out!
It's necessary project planning. Some of the planning is some scenarios. I used the Internet before the Web and didn't do such scenarios; I should have; big mistake. At that time, I had a terrific background in computing and applied math and could have done any of the really successful Internet projects. I don't want to miss out this time.
The $2 per 1000 from KPCB's Mary Meeker was, IIRC, for ads displayed, not pages displayed. I agree that having 6 ads on a page might not yield $12 per page, but I don't know how much less it might yield. I did mention that I admit I might have to divide by 2, 4 or some such.
> especially since you're using ad networks instead of selling inventory directly.
If that difference is costing me seriously, then later I will change over. My plans do call for my own ad targeting which, BTW, should be able to work with the ads from an ad network. As I understand it, the ad network business is competitive. So, going to Ford, Pepsi, Game of Thrones, etc. directly myself should not be necessary. Or, apparently lots of Web sites are using ad networks; maybe they are not hopelessly bad.
There is, sure, the medium and longer term, but I want to get to profitability, even just $20,000 a month fairly quickly. If the ad networks get a big bite out of the ad revenue, initially okay by me. Since I'm a solo founder, 100% owner, and the only employee, just $20,000 a month would give me a nicely profitable lifestyle business, and that would not be a project failure. Getting the ads myself now would be premature optimization. More broadly, nearly no business is fully vertically integrated and, instead, makes use of good vendors. Ad networks should be by now close to such good vendors.
If all the people in the world like the site as much as I intend and believe, then I'll have the first business with market value of $1T+. So, for possible results somewhere between $20,000 a month and a $1T+ business, I should go for it.
The probability of success is irrelevant and what is very relevant is the conditional probability of success given the details of my project. The crucial, core details are nice, with some shockingly powerful math well beyond essentially anything out of Silicon Valley. There's essentially no Web site out of Silicon Valley where the powerful, valuable, crucial core is some original applied math with some advanced math prerequisites. E.g., the Stanford, Berkeley, Davis, CMU, MIT, etc. computer science profs didn't take the right courses in grad school to have the prerequisites.
E.g., IIRC the Plenty of Fish guy was getting all his ads just from Google and getting revenue of $1 million a month still as a solo founder and only employee -- serve farm from two old Dell servers. Sure, later he grew his staff to 70 people and then sold out for his $500+ million.
> Also, I suspect 10 pages per visitor is high, but maybe not.
The site is interactive. To get from the site what is intended, it's about 10 pages per visit. People will look at a page for ballpark 30 seconds each. So, the 10 pages will be 300 seconds or 5 minutes. So, I will get their eyeballs to look at my content and then also the ads for ballpark 5 minutes. That 5 minutes figure will vary quite a lot depending on what the user is after in the addictive adventure that is the UI/UX. That is, a user might flip through the pages quickly or on some pages stay a while. I will be eager to look at the site log file and see how long users stay. For now, I'm planning that my session state server time out after 30 minutes -- there is good reason to believe some users will get highly involved and stay that long (really longer since the time out is for inactivity time, not total session time -- total session time can be an hour or two).
> Finally, it's easy to say "one visitor per second", but that seems kind of arbitrary.
Yes, that arrival rate is in this planning, of course, nearly fully "arbitrary" but does have a good purpose in the planning:
Can the arrival rate be this high? Yes, if believe the one necessary basic assumption that people will like the service. If people don't like the service and don't grow to, then the whole project is a flop -- that's possible for nearly any startup.
To be more clear, the main, one crucial, assumption is just that the target audience, i.e., all the Internet users in the world, will like the service. With that main assumption, all the other assumptions are quite believable and small potatoes. Without that main assumption, there is no hope for the project.
Can the arrival rate be higher? Sure, but the point here is to understand what would an arrival rate of 1 per second, if do get it, would represent in both monthly revenue and server farm size -- good to know this for the planning. The answer is ballpark, back of the envelope, simple assumptions, $300,000 or so a month. Also from my software timings, the arrival rate of 1 per second doesn't take a 1 million square foot server farm, need to know that, and, instead, just one mid tower case with one processor, whole system for $1000 in parts -- just did some quite careful shopping for such parts.
Or, for a scenario, go live with one server. If people fill that server, then there will be plenty of free cash for more servers.
Or, if I go live with just one server and people like the service, then the usage will generate enough cash for more servers; that is, the project is ready for organic growth if and when users like the service.
Again, the arrival rate of 1 per second is relevant, not so arbitrary, because it is about what one server with $1000 in parts could handle.
If the site does get to an arrival rate of 1 per second in, say, only a few months of being live, then just from common sense and the fact that the service is intended to have very broad, not just niche, appeal, the arrival rate should keep growing, a lot. Then I will have a quite successful, significant new Internet Web site and business.
Sometimes Web sites do become popular. Plenty of Fish did. One shocking example is Drudge Report: Look at the data he shows on his site: He runs only one or a few ads per page, but with the page views he's getting he has a heck of a business. He may be getting more revenue than, say, the NYT. By now, Drudge stands to be a quite wealthy guy.
It's the Internet! A LOT of people like a LOT on the Internet a LOT. Sure, Google, and Facebook are big. But so is Instagram and Snapchat, Twitter, Drudge, and more. The Internet in total costs super big bucks, and the bucks are there -- shocking but true. The appeal of Google is nearly universal; Facebook is somewhat less universal (e.g., I'm a good Google user but basically hate and avoid Facebook). The rest, Snapchat, Instagram, etc. have relatively narrow appeal but still are very successful as businesses. E.g., Drudge is not really broad but to the right of center. Same for Breitbart, which also gets one heck of a river of traffic.
I remember using the Internet -- downloaded from a friend the source code for the old IBM Scientific Subroutine Package -- back before HTTP, HTML, and the Web. I didn't see the potential of Google, Facebook, Snapchat, Drudge, etc. just because I would not have assumed that so many people were willing, one way and another, to spend the money to add up to the big bucks. IBM didn't see it either: At one time, they ran the whole Internet. E.g., I noticed Amazon right away -- books and records. Well, now the guess is that they are nailing Sears, K-Mart, Macys, a big fraction of all the shopping malls, and threatening much of Walmart. How? Sure, Internet data rate high enough for the many excellent still images for the good product descriptions on the Amazon Web site. Those images need the now much faster Internet download speeds, and who would have predicted those? I should have but didn't.
My service is to solve a problem pressing for essentially every user on the Internet. I have a really good solution, and no one else does. Due to that, and the success of Amazon, Google, Facebook, Snapchat, and the Internet as a whole, no way will I assume away an opportunity for success for my work. Again, if any reasonable fraction of the people in my target audience like the service to any significant extent, then I'm on the way to the world's first $1T+ business (right, the whole ad business may not be that large!).
The service I'm providing solves a pressing problem for nearly everyone on the Internet. I believe I have a good shot. At this point, there's only one good way to know -- use the TIFO method, try it and find out. So, to try it, go live, do alpha test and beta test, get some publicity, pay attention to additional feedback, and see if people like the service.
If people don't like the service, then at least I got some good experience on the software development side of bringing up a Web site. If people do like the service, then I'm wealthy. Drudge is wealthy; so is the founder of Plenty of Fish. Maybe I will be, too. This effort doesn't have to get to $1T+ to be successful for me. The $1T+ is scary: That's not just money but power, more than I want.