In aviation circles, the span of time between the two World Wars is remembered as the Golden Age. A global race to speed, range and performance records pushed the limits of technology and pegged countries and manufactures against each other.  It was a great run for Italian engineering too, culminating with the 1934 Schneider Trophy speed record that Francesco Agello established on a Macchi MC 72. His 709 km/h (440 mph) remains the fastest speed ever attained by a piston engine seaplane.

As an an aeronautical engineer, I was very excited to learn that, on June 12, 2009, Maurizio Cheli established another speed record and aviation milestone on an Italian built craft. He pushed SkySpark, a 100% electrically powered plane, to a top speed of 250 km/h (155 mph). The news from Turin echoed into  Silicon Valley.  SkySpark is a 100% Italian consortium of university, private and public resources.

And while dreams of electrically powered sub-sonic intercontinental flights are premature to say the least (energy density of Jet A-1 fule = 40MJ/Kg; energy density of hydrogen fuel cell = 1.5 MJ/Kg) this achievement demonstrates that, with little resources and brilliant engineering, you can spark true progress.

I have been in looking into crowdsourcing business models for the last couple of months. Personal and professional interest. As often the case, when a subject is on your mind, you stumble upon it. So, when ads like this one started popping up on my Facebook page I clicked through.

The ads send you to Logo tournament (LT), a crowdsourcing play on logo design services. It’s quite straightforward. Company needs a new logo, it submits a “contest” to LT and sets a price. Designers send in one or more designs. The best design wins and the designer gets paid.

I tried to figure out how a business model like LT’s would look like.

Revenue

LT charges designers a 15% transaction fee. Companies set the price they are willing to pay in advance, with a minimum allowed of $250. Paypal transaction fees are passed on to customers.

Note that LT will take the money from companies in “escrow” and select and pay a winner at the end of the contest even if a company does not select a winning design. That means that if a contest is listed, the money will be paid and the commission earned.

LT lists all open and closed contests at any given time. I counted about ~165 open concurrent contests with an average age of 5.1 days and ~1,800 closed contests. I sampled 100 contests at random and calculated an average prize amount of ~$315. Assuming a steady flow of contests, that means about ~1000 contests a month (i.e. 30/5.1*165=971).

With commissions at 15%, that adds up to a monthly revenue of close to $50,000. The estimate is obviously highly sensitive to the monthly contest count estimate. Here’s a back of the envelope revenue estimate (with lower and higher estimates to test for sensitivity).

Operating Expenses

To estimate how much LT is spending to run its crowdsourcing marketplace, let’s look at staffing, infrastructure, customer acquisition marketing and payment processing. For the sake of this exercise, I’ll bundle the rest (e.g. legal, accounting, office rent, insurance, travel, etc.) as a miscellaneous overhead charge. WARNING: broad, sweeping assumptions :)

Staffing

I do not personally know the LT team, but I’d suspect it to be pretty lean. Let’s assume a team of 3 full-time employees (marketing/support person, developer, designer). The company being Canada-based, depending on seniority and benefits, average cash-out monthly cost per head would be in the, say, $5,000 to $7,000 (US$) range. Let’s assume also temporary and specialized staffing needs to be met with contractors or offshore resources adding anything between 10 to 20% to the cost of payroll.

Marketing

Let’s assume that LT is primarily investing its marketing budget on CPC campaigns on Facebook and similar CPC networks. Assuming an average CPC of $1 and a conversion rate from landing page to paying contest customer of 3% (Averaging out typical eCommerce conversion rates), LT must invest $33 to acquire one customer.

Not all customers are acquired this way, though. To account for repeat business (i.e. customers running more than one contest) and the fact that some will become customers through word-of-mouth and other “pull” channels, let’s assume a word-of-mouth/viral/returning-customer factor of 1 to 1, i.e. one “free” customer every 1 customer acquired though direct-response campaigns. That reduces the effective unit acquisition cost to $17. To get to the total monthly acquisition cost, one can multiply the unit cost times the estimated number of contests per month.

Finally, let’s assume that other marketing programs and channels (e.g. social media, events, schwag…) account for an incremental 15 to 25% in marketing spend. This is a measure of the activity that is necessary to generate some pull demand for the service and promote it to the design community. With marketing programs overhead at 20%, LT should invest in marketing ~$19,400 per month.

It is easy to see how rapidly customer acquisition costs can get out of control if LT does a poor job of (a) converting customers after they click on an ad, (b) getting repeat business and (c) attracting customers through word-of-mouth.

Infrastructure

Let’s assume an average storage capacity requirement of 2MB per design submission (Submission file + JPG previews and thumbnails). Of a random sample of 100 completed contests, each contest averaged 275 submissions. Let’s assume LT archives past contest graphics for an unlimited amount of time, in which case its storage capacity needs grow over time. Assuming that LT list of ~2000 contests includes all contests ever run, its cumulative required storage capacity today would be about ~1.1TB. At an average estimated cost of $0.150/GB (Amazon AWS pricing) the total monthly storage running cost estimate is ~$165.

To estimate monthly data transfer costs, let’s look at the site traffic. Data from Compete.com estimates about 16,000 monthly uniques. Assuming 10 page views per unique and average page load of ~0.4MB, the total monthly data served adds up to ~64GB. Assuming a price of $0.17 per GB transferred per month (Amazon AWS pricing), the monthly data transfer costs are a whopping ~$11.

Even after adding another estimated $2,000 a month in hosting, database and other infrastructure costs, it seems infrastructure expenses account for a small share of overall operating costs.

Payment processing

LT passes Paypal charges on payments received to the company that submits a contest. It however pays Paypal fees on its payments to the winning designer. Let’s take the basic Paypal fee of 2.9% + $0.30 as a base. The processing fees add up to a hefty ~$7,800 per month.

Other Operating costs

To estimate additional operating costs, including rent, insurance, travel, IT, etc.), let’s assume a percent overhead related to the size of the team, ranging from 10% to 20% of payroll costs.

Bottomline

Internet businesses are hard to make work and a crowdsourcing marketplace is no exception. Below an estimated operating earnings summary.

Some considerations:

• Marketplace size is key. All things equal, if LT were able to grow its contest intake by a factor of 10, it would turn a 15% loss into a 35% profit. The trick is to be able to scale the marketplace size while at the same time keeping customer acquisition and marketing costs down.
• Marketing is key. Poor customer acquisition performance can kill a business like LT’s. Specifically, it is essential to keep conversion rates high through well targeted ads and flawless user experience when a prospect enters the conversion path. “Free” marketing is also critical. LT must be able to leverage WOM from existing customers and repeat business to keep effective customer acquisition costs in check.
• Cash from Customers. Where does LT gets its working capital from? Well, since a costumer pays into LT’s Paypal account when their contest starts, but the prize money is only paid to the winning designer after the contest is closed, one to three weeks later, customers are effectively financing LT’s business. In fact, the faster customer acquisition grows, the more cash can LT float to finance growth. Brilliant!
• Transaction costs are not trivial. Paypal fees of almost 3% on a 15% brokering commission amounts to 20% of net revenue after the designer is paid!

You can find the calculation sheet here.

The recently published IMD World Competitiveness Yearbook 2009 shows some significant changes over last years overall rankings, though not much at the top:

Of the 57 economies ranked by IMD, the US still ranks No. 1 in 2009. Hong Kong has switched places with Singapore to gain the 2nd place and is swiftly “closing the gap” with the US. Switzerland maintains its 4th rank from last year.

Italy lost another few positions, from 46th in 2008 to 50th in 2009. Big and unexpected movements in the middle of the range, though possibly due to the different recession timing in different geographies. It will be interesting how the global recession plays into 2010 rankings:

The most spectacular movements are seen for Indonesia, rising from 51st place to 42nd and Estonia, falling 12 ranks to 35th place. Some countries suffered important reversals: Colombia (51st), Greece (52nd) and Taiwan (23rd) fell 10 places each, followed by Romania (from 45th to 54th). Other important declines include: Luxembourg (from 5th to 12th), Hungary (from 38th to 45th), Spain (from 33rd to 39th) and Ireland (from 12th to 19th). And to think that Ireland was ranked 5th in 2000!

Particularly noteworthy this year is the aptly-named, newly-reported “Stress Test on Competitiveness” ranking:

Denmark finishes in first place in the “Stress Test” rankings, an analysis of which countries are better equipped to fare through the financial crisis and improve their competitiveness in the near future. In other words, the test is future oriented – it focuses on exposure, readiness and resilience in a period of world recession.

[...] Despite finishing first in the overall 2009 World Competitiveness Yearbook rankings, the United States comes in 28th position in the Stress Test, underlining the concern of the market with the depth of the crisis and the time that it will take to solve it.

Interesting. Here’s the 2009 scoreboard (PDF file)

Out of 10 people that can do a job, 1 is an A-player, 3 are B-players and the rest are C-players, or so the story goes… Simple enough, right? So, I ran a couple of public polls on Linkedin. Just for fun. They probably have no statistical relevance, but the results are interesting  :).

First, I asked respondents to self select in the A, B or C player category. 78% put themselves in the A-Player list…

Then I asked respondents to assess their coworkers. 42% said less than 1 in 100 coworkers are A-players, 77% said less than 1 in 10 are. Obviously B- and C-players don’t like to answer polls ;)

In a recent post, Auren Hoffman argues that, in this economy, hiring has become harder… and he is rather convincing at it:

That’s right … hiring in tough economic times can actually be much harder than when times are good.   In a downturn, the amount of resumes from C-Players massively increases while the amount of resumes from A-Players probably remains the same.

I generally agree with Auren’s premise and conclusions, though I believe it is important to recognize that most companies will never have access to the A-Player pool in the first place, and explained why in my comment.

… Second, about the analysis. The article rests on a fundamental assumption that ANY company has access to A-players or at least all the available A-players in a specific domain. That is simply not true. For example, let’s consider the subset of all A-player graduates from top colleges that are interested in Management Consulting. Now, a leading firm, such as McKinsey for example, is almost guaranteed that close to 100% of those A-players’ resumes will end up in its funnel. Sure they will be buried in tons of not so good candidates. For McKinsey is then only a matter of putting in place a good process. That is simply not true of Mom&Pop Consulting Group. In fact, M&PCG is almost guaranteed that none of the A-players will apply. The difference here is the buyer’s (i.e. the company’s) brand. Now, some will say, M&PCG should simply actively seek the A-players in the hidden market. Nice, but I doubt it to be economically feasible to go find out all the A-players. Even if that were possible, most A-players would have no interest in the company.

Every CEO likes to think about his/her company as the best place to work on earth. The hard truth, that every startup founder is all to happy to ignore, is that for the vast majority of companies, the pool of realistically “reachable” A-players is close to nil. No matter the market.

In other words, while it is obviously wise to do a thorough job at selecting employees, a manager will always compromise. “We only hire the best” is ALWAYS qualified. You only have a chance to hire the best person available to you at that moment, i.e. the best person that knows about you, is willing to work for you, is on the market for a job, etc. etc. And if someone is available to you, it does not mean he/she is available to me.

If Foreign Direct Investment is the single most important indicator of how attractive to business a country is, this data from the Economist must be troubling.

We are drowning in irrelevant advertising. Some are inspired by it, most ignore it, some would like to get more relevant ad messages.

I have been convinced for a while now that selling user data is one of the most direct and potentially lucrative ways social networks can achieve the kind of revenues that would justify their valuations. A recent article on Facebook’s intention to “cash in” on user data reminded me of a recent conversation.

Last November, I sat on a “User Contributed Data” panel with Richard Cooperstein of Facebook and other experts at the last Monaco Media Forum. While Richard found himself in the position to have to defend Facebook’s privacy practices, I humbly pointed out to the audience that other, more powerful and deep databases store much more private and sensitive information which is routinely packaged and sold to businesses and government.

In fact, Facebook SHOULD leverage the vast amount of personal data it stores to allow marketers to target their campaigns more effectively, on-line and off-line. Here’s why it makes sense:

1. Economics. Advertising from Facebook impressions alone (Social Ads, Engagement Ads, Display, etc..), will hardly unlock but a fraction of the true economic value of all that targeting data.
2. User value. Behavioral and audience networks have proven the viability of targeting users beyond the confines of a single domain and reach them when a specific ad message is more valuable to them.
3. Proven model. Credit card companies and data aggregators, such asChoicepoint, Lexis-Nexis and Acxiom, already own far more private and sensitive data about individual purchasing behavior, health and personal finances than Facebook will ever have and have built huge businesses of trading that data and selling it to corporations and the government.
4. Privacy controls. With the appropriate opt-in/opt-out and consumer disclosures, social networks can find a balance between users’ right to privacy and ownership of personal data and the ability to effectively target them.

Would I mind if my credit card processor knew to offer me deals on backpacking gear (Facebook “knows” I am into that) instead of Alaskan cruises I will never be interested in? Not a bit.

In just two years (the 2008 edition is its third one), the Monaco Media Forum has established itself as a major appointment in the digital media conference calendar. The 2008 edition takes place on Nov 12-14 and the Taglift crew is on its way. This year the attendees list and the roster of speakers is as impressive as it gets. You can follow the event on a special socialmedian site.

I have been invited to participate in a panel on “user-generated data” moderated by Esther Dyson with Richard Cooperstein of Facebook, Auren Hoffman of Rapleaf, Mattias Miksche of Stardoll and Lisa Suonio of Dopplr.

Looking forward to it.

Photo by rsepulveda

So I finally did it. With the weather forecast promising three straight 80s-and-fog-clear coastal days, I headed to King Range National Conservation Area to descend north to south the 25.6-mile stretch of the Lost Coast trail from Mattole to Shelter Cove. Here’s the log.

Friday, Oct 10, 2008

3:15 am: Left San Francisco to drive north on H101. Deserted Golden Gate bridge has a surreal quality to it. I am doing good time all along the way, though encountering multiple slower men-at-work areas along the freeway.

7:50 am: After stopping in Petrolia to confirm directions, I arrive at Black Sand Beach trailhead parking lot. Air is cold and dry. At 8:00 am sharp Sherri, my shuttle ride, pulls up in the lot, exactly as planned. The party of three + dog I am sharing the ride with joins us soon after. They had spent the night in a nearby camp.

8:30 am: Quick stop at the NCA Ranger Station for maps and bear canisters. I ask the ranger on duty if he thinks I can make it to Randall Creek before the afternoon high tide renders access along the trail impassable. He is skeptical, but suggests to get going quick if I want to try.

10:00 am: Mattole trailhead. I pay and thank Sherri, waive goodbye to my shuttle partners and get going. Determined to keep a good pace. Glad the gusty wind is blowing from northwest, at my back.

11:50 am: Punta Gorda Lighthouse. 3 miles into day one, I reach the abandoned lighthouse, stop for a dry meal and head down to the beach to explore elaborate driftwood shelters.

1:20 pm: Bluffs, sand and high grass give way to boulders as the beach passage narrows considerably. I can sense the tide rising as the day goes by. Negotiating the narrowest stretches of the trail, I am now stepping in and out of the water and watching for 10-foot waves, keeping a good pace to avoid being squeezed back out by the tide.

3:30 pm: Randall Creek suddenly opens up on my left as I clear a narrow beach passage. A stream emerges from the vegetation at the bottom of the creek and expands in a beach pond just above the high tide line. I setup camp on the bluff just above the it.

6:30 pm: Randall Creek. Just a memorable sunset.

Saturday, Oct 11, 2008

6:40 am: I wake up before sunrise. Hot coffee, milk and granola, while I savour the light flooding the bluffs, turning the ocean to blue and giving it its morning sparkle.

10:00 am: Spanish Flat. The broad grassy headlands expand under the bright morning sun.

12:40 pm: Big Creek. At about 14 miles from the start, the beautiful, wide creek opens up covered in dense vegetation. Sand dunes, gravel and boulders slow me down as I start to feel the strain in my heels and knees.

1:45 pm: Big Flat. As I reach the widest flattest section of the landscape along the trail, I reach the northernmost tip of a unpaved landing strip. The isolated house of the small airport residents is the largest of a handful of residences scattered along the route. Big Flat is busy. I meet other backpacking parties as well as surfers that have setup camp there and are enjoying the long, steady breaks. I also meet the ranger I had talked to at the ranger station the day before. He is surprised I made there ahead of the tide and that I am backpacking alone.

3:20 pm: Shipman Creek. After a tiring mile-long stretch of boulder hopping, I reach the beautiful campsites at Shipman Creek, where I decide to pitch tent. I am now 6 miles away from Shelter Cove.

8:30 pm: Shipman Creek. While I am tending a beach fire, two hikers with their dogs arrive at the camp. The dogs end up chasing a skunk into my site which in turns causes the skunk to spray all over my tent and gear. After attempting to get used to the stench in the tent, I decide to take my bag out and sleep under the rising moon.

11:30 pm: Shipman Creek. It is getting colder and I wake up. The moon has now set and I open my eyes to the most incredible, saturated, star-studded black sky I have ever seen. Truly awesome.

Sunday, Oct 12, 2008

8:50 am: Shipman Creek. Sore feet and heels and a light headed make me feel happy I only have little more than 6 miles to reach the destination.

10:30 am: Gitchell Creek. It is a beautifully warm and bright day. I meet multiple parties of surfers on their way north, lugging their boards and gears along the beach. Feet hurt and hiking in sand is slow and exhausting.

12:30 pm: Black Sand Beach. Gated by big, tall rocks the beach widens up, stretching for over a mile down to the trail head. As I get to the end of the trail I take a moment to enjoy the breeze, the sun, the sand and to soak my feet in the freezing Pacific Ocean water.

Until the next one.

An interesting coincidence this weekend in San Francisco. Two big events, two celebrations of the fruit of human ingenuity, creativity and technology. Both are the result of the vision of Italian minds.

After 8 years or renovation the new California Academy of Sciences, designed by Italian architect Renzo Piano, opened its door to awe thousands of visitors with its innovative architecture and its spectacular exhibits.

At 2 pm on Saturday, the 289-foot Maltese Falcon, the largest private sail-propelled yacht in the world, designed by Italian shipyard Perini Navi, made its entrance into the San Francisco Bay through the Golden Gate, sporting its beautiful computer-controlled sail system.

Things to be proud of.