This is not about analyzing things to find out what’s going on. It’s about the relative ineffectiveness of applying analytics for the purpose
What’s more effective? Analyzing petabytes of (often stolen) data ($$$) (often incorrectly), or mimicking in real time what is being proven to work in real time?
“Big data can be analyzed for insights that lead to better decisions and strategic business moves.” Castpoints allows others to just copy what works. If Lisa’s leadership role of marketing for lawn mowers, and all the associated tasks, has a 90% ROI rank, she can make it available for a price. Now Anne, who’s current role is only doing 30% ROI can buy Lisa’s expertise. Both profit. And they can see tomorrow’s probable decisions because there are other high ROI roles further along on the same pattern. Simple pattern matching with first semester statistics.
A mock supply and demand chart from Castpoints is below. This would all be real time free info. An amazing amount of info is generated from things people already know. A start / stop time; min or max price. We don’t have to analyze anything and guess about what people are doing or intending, or what would be of value to them. They already made it public, yet their privacy is protected.
Transportation > SF > SUV (context)
23 @ Mission and 1st st. to Seaclif (item)
15 USD avg (for he next few hours)
(This is a design I am playing with for cell phones. The big screen version will be nicer.)
- The left scale
- The black 36 is the average price of supply.
- The red 34 is the most recent transaction.
- The thin red line is the historical average.
- The black 23 is the average price of demand.
- The red circles are transactions. This is where supply and demand overlap looking forward.
- The purple line is my order for a ride from Mission and 1st street to Seaclif to buy groceries sometime today. It says:
- I am willing to be picked up in an hour, but no later than 12 hours from now.
- I’m willing to pay near zero, up to $34. I want to make sure I am willing to pay more than average because I don’t have any food in the house. But my schedule is flexible enough that I might get the ride a bit cheaper.
- Right now it looks like supply will overlap my demand about 6 hours from now denoted by the short red line going though the long purple order line.
- The purple boxes are for changing the order quickly. There are a couple of arrows (not shown) just click on those and the 4 order parameters change, say 3%, as denoted buy the box edges.
- The green circles are current opportunities. Either supply and demand are relatively very far apart, or relatively imbalanced.
- If you are a potential driver, you could get an alert about times when you can earn about 20% more RIO than history would suggest.
I can always link orders together to get a ride to the market, then a bookstore, then lunch, then back home. Or, I can put in an order in the category:
Transportation > SF > SUV (context)
8 @ Mission and 1st st. to Seaclif store for an hour then return (item)
35 USD avg (for he next few hours)
Of course there are filters for the driver’s reputation rank and performance bond level.
This chart template can be for any item in the world. In other words, anyone can find the supply and demand of anything easily, and be notified about arbitrage opportunities.
Knowing the supply, demand, price, and ROI’s of any item or option makes decisions much simpler, and defendable.
Update Dec 2015
Founders Dating: Best pricing optimization software?
There are a bunch of fantastic services in the space – we’ve worked with www.wiser.comand www.360pi.com… however, these services are often really expensive (thousands per month). A quick and dirty way … is to pull down pricing from Amazon API – and stay within a % of the lowest price…
Instead of spending resources basing prices relative to others, consider using prices that automatically adjust to all variables.
Restaurants can choose their level of commission rate, at or above the Company’s base rates, to affect their relative priority in its sorting algorithms, with restaurants paying higher commission rates generally appearing higher in the search order than restaurants paying lower commission rates. Commissions are generally based on a fixed percentage of the value of the order.
Could this be why fake restaurants have started popping up on Seamless in New York City? Just last week, it was discovered that some restaurants have joined the website under a fake second (or third, or fourth) name, but when investigated the Seamless accounts all direct orders to the same restaurant.
Because the pricing is indirect, people try to game it. Then it takes resources to manage that. Which is why their apparent fee to match supply and demand is at 13% when it could be at < 1%.
Many task and services should have 2 prices curves. One for people who have more time then money, and another for people who have more money than time.
Update January 2016
Google Flu Trends is no longer good at predicting flu, scientists find Google has a lot of resources, yet made basic mistakes like not accounting for auto suggestions. Again, it’s doing things indirectly. What’s the pedigree of the CDC data? How often do they change their definition of “flu”? How does Google deal with it’s daily search algorithm changes which probably effects auto suggest? The supposed 45 search terms and weightings — how are those selected and determined?
With CP, the real and actual raw data would be available to everyone in real time. (CDC data seems to be weekly.) This massively reduces many uncertainties. Then entities would compete to generate the best prediction and monetize that.
Google Flu Trends and Google Dengue Trends are no longer publishing current estimates of Flu and Dengue fever based on search patterns.