This week’s elevator pitch comes from Valu Valu, an online marketplace for video games whose prices are based on dynamic scientific pricing, creating the optimal price for both the seller and the buyer. The pitch was concise and outlined the service the the site is delivering well, but didn’t tell us how Valu Valu will make money. After doing a little bit of research, we discovered that Valu Valu charges a 5 percent transaction fee on the total purchases (there’s no transaction fee charged to “local” transactions). The site currently features video games but plans to expand to other markets in the future.
Founded by ex-Microsoft techies Emmanuel Marot and Bruno Botvinik, Valu Valu uses a proprietary scientific pricing algorithm that continuously optimizes prices based on market conditions, a.k.a. supply and demand, so that buyers are happy with prices of goods (and thus will make purchases) and sellers make more money. The seller’s price is determined automatically, giving sellers limited control of the price of their goods. Other online marketplaces, like Ebay or Craigslist, allow the seller to determine the price of the item being sold. But Valu Valu’s method saves sellers’ time by establishing a set price, and cuts out haggling or auctioning time.
Of course, Valu Valu will face competition in game sales from popular online retailers like Amazon.com and GameStop’s EB Games. I did a side by side price comparison of the “James Bond 007: Quantum of Solace” game for Playstation 3 between Valu Valu and Amazon. Amazon’s price for a new game came in $10 lower than Valu Valu’s estimate for a brand-new game. The Valu Valu’s product that was actually being sold was “just like new” (which sounds like a nice way of saying used), but even Amazon’s used Bond game were selling more than $10 lower once again.
Valu Valu just launched the beta version in February, so hopefully the start-up will be able to attract more users in the future. I think Valu Valu may be on to something. It seems like an innovative technology that needs to be tinkered with a bit more.
Here’s a screenshot:
It happens to be late at night but after watching the pitch, I didn’t understand most of what he said….’subtitles’ sort of helped.
I just double-checked to make sure I wasn’t having a hearing problem…but I had the same issue. His accent is too strong.
I’m having a flash back to Steve Marin and the Pink Panther 2
http://www.sony...hepinkpanther2/
Hamburger, ammmm burger…
Well – I could understand him perfectly. It just takes a ‘little bit of effort’.
“I think Valu Valu may be on to something.” – well, considering that this statement followed directly after “… but even Amazon’s used Bond game were selling more than $10 lower once again.” – I am somewhat less enthusiastic about their business model.
You know what’s really funny. You clone valu valu’s algorithm, start a website that prices every single item they sell 10% cheaper.
But you’d have to be really smart to do that. NOT.
This technology already exists on the net. Not only does this exist but *shock* technology that prices items depending on users and their past purchases exists as well. “add to basket to see price”. Yes, your cookies sometimes determines your online price. Life isn’t fair.
good
This does not require any complex algorithms. It is not calculus. Actually it is not even algebra. It is arithmetic.
Just take lowest price on amazon * discount factor for condition of product * weighting factor based on prices of similar products on ebay and amazon marketplace – 15% and you have it.
Even someone who used to work at Microsoft should be able to figure that out
From India
Anjali Sen
Sure, from the smart [not] smelly spammer From India…
Shut up!
@courtney,
I don’t think you take bath daily.
Ok, I do not get what he is speaking.
“…..After doing a little bit of research….”
Their web site show clearly thay charge a 5% fee for each transition.
Did not have a problem understanding every word. Revenue management isn’t new, I remember a company online years ago moving thousands of computers using the method, forgot their name. Most of the work as with most other companies is within marketing. Algorithms, tech can evolve. From experience, it makes no sense to judge a company’s future nor potential based on a quick 2 minutes presentation.
Wishing them the best,
Cheers
Sahar Sarid
Bido.com – Social auctions
I suspect (an educated guess) what they’re using is MCMC (Markov Chain Monte Carlo). MCMC is a popular algorithm/technique in science. Its use spreads from particle physics to fluid dynamics.
The Google PageRank algorithm is derived via Markov Chain. This was found out later by other researchers, in which Brin & Page didn’t realize that their derivation of PageRank was indeed a Markov Chain process.
I have seen a few publications of the use of MCMC in online auction pricings such as described in the following publication:
An Agent-based Platform for Online Auctions
but of course, there are various other methods that have been applied to online/e-commerce environment dynamic pricings, using dynamic programming such as described in this paper, with the abstract shown below:
Title:
—-
“Dynamic pricing: ecommerce – oriented price setting algorithm”
Abstract:
——-
Pricing in electronic commerce is based on bargaining. Pricing models that can fast change prices during transaction on the consequence of the buyer’s needs is beneficial to electronic commerce. Demand sensitive model is one of the pricing models that can be used for fast changes of prices in electronic commerce. Price setting algorithm for demand sensitive model helps sellers to get decision variables, price per unit that maximizes profit for the quantity ordered by buyers. In this paper we analyze the price setting algorithms of demand sensitive model. We use a simple example to explain how the changes of price elasticity of demand changes price per unit, gross margin and quantity demanded. Also we show how the changes of quantity demanded changes the unit price and marginal cost. The investigation shows that an increase in demand ordered decreases price per unit of a good, at the same time increasing profit margin to seller and decreasing production cost. The seller extracts some of the buyer’s surplus value as profits with residual surplus remaining with the buyer over and above the actual price paid. Buyers do not pay the same amount of total price for the good ordered within the same group of order, because of the difference in the net browsing cost.
I think that in the future, ecommerce will deploy dynamic pricing engines for competitive advantages, since one who is using a fixed price engine can’t afford to lose customers to competitors who have variable prices depending on certain conditions, which is more flexible.
Yes but there is no need to use Markov model to make predictions when the market prices are already revealed in places like Amazon Prime, Ebay and Amazon. These prices auto adjust in real time as each sale is made and as a result, you avoid the state vs revealed preferences problem of building an elasticty model.
I am willing to bet a bottle of champagne that simply taking lowest price on amazon * discount factor for condition of product * weighting factor based on prices of similar products on ebay and amazon marketplace – 15% will yield a much more market reflective price than the algorithm they are using right now.
From India
Anjali Sen
Sorry, but your comment does not make much sense. It reminds me of the language the students used with other during the second semester at B-School. It seems that you don’t even know what you are writing about.
Oops, the link to the article with the abstract shown in my previous message is shown below:
Dynamic pricing: ecommerce – oriented price setting algorithm
@Falafulu Fisi
Genius comment. I’m wondering if you can apply algorithms for service oriented industries? I see that you have a good understanding of semantic principles, very interesting. Could you contact me? I have been looking for a smart fellow like yourself.
James (of) adventurelink.com
You know what would be so much better?
Over a year ago I interviewed with the Amazon.com page landing optimization team for a C++/SQL job doing SEO page landing optimization for when somebody lands on an Amazon.com page from Google or Yahoo.
Amazon’s page will change based on the referrer that the user agent sends to the server from the browser upon landing, and based upon cookies that person may have from Amazon.
Why not use that page landing data to see how much X person is will to pay for Y item when navigating from Z page.
Say the person was on a porn site, and they Google’d sex toy or flesh light because they are horny. They find some on Amazon in the health care items.
That person may be in an agitated state that would incite them to buy at a higher price than somebody who is navigating through the site looking for other items.
So this is an innovative idea that I just came up with. These ideas have yet to be explored. Somebody should explore them.
Of course with all the software on the internet that runs shopping cart and ecommerce, it would be a mighty job to plug into all of those, so you would have to create this type of page landing optimization as a SaaS service.
The server PHP, ASPX whatever would connect to the SOAP api when a request comes in and sends the user agent info, search engine key words, and previous purchase info in a standardized format. Based on that the proprietary server side technology returns a number from 1 to 10 of how likely that person is to buy the item. Based on that number you set the price of the item.
Google has double click and adsense so they can use cookies and data from a number of other websites when a user lands on their page. They can tell if somebody is addicted to buying X item from that data, say gadgetry, and raise or lower the price based on that.
Congrats on the launch, Emmanuel! Keep those Mario DS games stocked.
Paraphased: “Buyers will get cheaper pricing, …Sellers will make more money (7% or so).”
Anyone else find this hilarious?
Thanks for the positive comments… and the other ones as well!
I wish pricing would be as simple as getting a price on Amazon and discounting it. Unfortunately, it’d be very hard to convince sellers to put their product for sale on a marketplace which is systematically the cheapest. And, as much I love Amazon, their prices may not bethe most efficient ones everytime! Also video games is a ‘test’ market for us, our ambition is much broader. In the longer term we aim to price may kinds of goods or even services.
We evaluated some Monte-Carlo techniques, but decided to pursue other fundamentals. Good guess, though, it’s all related to Markov Decision Chains, in a way…
At last, I know it may sound funny to promise both cheaper prices for buyers and more revenue for sellers. But Dynamic Pricing is, oddly enough, not a zero-sum game. E.g: for airline tickets, it allows both Coach and Business travellers to get cheaper prices (yes, both of them!), and Airlines to make more revenue…
Emmanuel
I do appreciate the hard work you have put into this, but how in the world do you get buyers trust if your prices are HIGHER than on amazon.com? They offer great shipping, great service, years of trust, etc. etc. … how can you justify a premium to amazon.com?
At the very least the price has to be lower than on amazon.com, discounted for the lack of reputation/services/delivery/trust that buying from an unknown person provides. So this discounted price has to be the ceilng price that you can propose.
Then the trick is figuring ut the optimal market price below that ceiling. But once again, there IS a market for these services already, so market information is already there in places like Ebay and Amazon Marketplace transaction data.
I think you have made the problem and the solution too complex. For commodity like goods such as video games, the problem and the solution are actually very simple.
From India
Anjai Sen
Again, you continue making absurd noise. Please stop the spam!
I agree with Anjai.
You are the troll here Ted
Emmanuel,
What about the inputs? Just using Leena’s example, although it is very simplistic, it sounds like your algos either need some tweaking OR you lack inputs (at the risk of adding complexity to your algos).
What is coming to mind for me is that algos depend on inputs to create outputs, right? (Simple.) So where does your system obtain inputs from? Maybe you can’t speak on that, as it is part of your value proposition and a trade secret (or you plan to patent).
I totally get your argument for price optimization, but it has been done before. The argument that you’re going to get the cheapest buyer price and highest seller price is a bit…off…to my ears. What happens when you apply your algos to other products? What happens when you compare your algos results to other markets, such as Amazon, eBay and Craigslist? Are the results comparable in your tests? Did Leena make some mistake or misuse, which maybe you can work with her to correct?
Something is missing here. I just can’t tell what it is as yet. Please enlighten.
Leena’s example is a bit of an edge case, unfortunately. Most of the products we got are in fact a tad cheaper than on other marketplace. E.g http://www.goog...ng=p#ps-sellers.
While we’re still fine-tuning our algorithm (a new version will be put online in the next couple of weeks), the way it’s designed it can find the ‘equilibrium price’ by itself. However, we made the mistake of taking very approximative starting prices in the algorithm. Hence the discrepancy one can observe. We’re working on fixing this. After all, as noted by Anjai, ‘market information is already there’…
Also, our engine is not product-sensitive, but we definitely wanted to ensure it gives top notch results before applying it to expensive goods. Hence our exclusive focus on video games at start…
Best of luck to them. However, they’re going to lose potential customers really, really fast if they keep pricing things above Amazon. I don’t need an MBA for that statement.
Hi Emmanuel,
Congratulations on the launch. I can understand how scientific pricing helps airlines and the like who have limited inventory (eg. seats on a plane) and that inventory is perishable (the plane is taking off whether the seats are filled or not). But how does this sort of optimization help sell used video games, where the inventory isn’t really limited, and the commodity isn’t perishable. Even more interestingly, how would it apply to services?
I think the idea is great and obviously nobody is expecting you to publish anything that might invalidate potential patents etc, but it would be a lot easier to get excited about the pricing model if we had a simple CommonCraft style example of how it worked…
Thanks for the advice, we’ll try to explain some of the assumptions of our engine.
As for the ‘non-perishability’ of games or services, one may consider any good to perish over time (we do not tackle collector items or art pieces). A ‘Gear of War 2 for Xbox 360′ or a ‘text editor for Windows Vista’ won’t be worth much in a couple of years from now. It’s also financially better to sell something for $20 today than for $20 next week (a value of ‘now versus tomorrow’ which is by itself entirely perishable). So, there is a drop, albeit not as brutal as for plane seats at take-off or rental cars remaining in the parking lot on Sunday morning.
I like the concept. I think the scientific/data mining approach is the most efficient way. As long as you keep it focused on one sector/genre I think it’ll work. This is a big potential market. First I’d evaluate the risks though at a digital security site like justaskgemalto.