Why Can’t We Apply the Google Algorithm to Every Day Life?
Published: May 15, 2008
Author: David Rodnitzky
Yesterday I went to the pharmacy to pick up a prescription. I had called the prescription in the day before and all I needed to do was tell the pharmacist my name, pay my co-pay, and leave. In front of me was a man who wasn’t quite as prepared for his pharmacy visit as I was. He first asked to pick up a prescription, only to be told that he had shown up too late and the prescription had been canceled. He asked for a refill, and then was told that he would have to call his doctor first. He then asked if he could just pay for the prescription and had to wait to get the full price. Once he heard the full price, he had to ponder whether he really wanted to pay $125. Finally he decided to just come back later. The whole process took about 20 minutes.
One of my biggest pet peeves is sitting in lines. Did you know that the average American will spend five full years of his life waiting in lines? That’s one year less than what we spend eating, and 10 times longer than we spend waiting at stoplights! And amazingly, a lot of this waiting is time spent waiting to buy something – to give our money away! As Paco Underhill recounts in his classic book on shopping psychology – Why We Buy – retailers spend millions of dollars a year trying to figure out the exact right placement of products in the store, how to create promotions that attract people to shop, and even what music will drive the most purchases, and yet after all this effort, they leave consumers to wait in long lines at the end of the shopping experience. Indeed, I suspect I am not the only American consumer who has spent a lot of time shopping in a store, chosen some items and then seen the long checkout line and dropped my potential purchases on the spot and headed to the exit.
Waiting in lines is just one of many inefficiencies we have to deal with in everyday life. There’s the waiting on-hold when you call customer service (or worse, having to spend five minutes trying to interact with a voice-recognition program and then being transferred to an agent who doesn’t have any of the information you just inputted); the oblivious driver going 45 in the left lane of a 65 MPH highway; the disorganization that greets you when boarding an airplane, since the frequent flyers all get on first, sit at the front of the plane, and then crowd the aisles, preventing anyone from sitting down; the fast food restaurant with 15 people working and no one but you in the establishment, but no employees taking any initiative to take your order (yes, I am talking about you – employees of the KFC located on El Camino in South San Francisco!).
The same annoyances once occurred in search engine marketing. In the old days, he who bid the most showed up in first position – even if his product was not relevant to consumers and his ad text gave consumers no reason to click through. Like everyone else, I gamed this system. I used to buy ads for FindLaw that lacked any semblance of a call to action like: “FindLaw: The #1 Online Legal Resource.” I figured it was free branding and I definitely drove a lot of impressions on the cheap. But Google changed all that with it’s ‘yield management’ algorithm. The technical (er, Wikipedia) definition of yield management is: “the process of understanding, anticipating and reacting to consumer behaviour in order to maximize revenue or profits from a fixed, perishable resource.” In other words, by combining maximum CPC with click-through-rate (CTR) and eventually Quality Score (a somewhat amorphous calculation of relevancy), Google changed the bidding process from ‘he who bids the most wins’ to ‘he who makes the most money for Google wins’ (a CPM model that maximizes revenue per 1000 impressions). Add in quality score and what Google is really saying is that the ad with the highest monetization for Google that also meets Google’s minimum standards for consumer relevancy wins.
The current model applied to most of our waking life is not one of yield management. Waiting in line at the pharmacy, or on-hold on the phone, or behind a slow person in the left lane – these are all examples of a “first come, first serve” model. There is no consideration of the value each consumer places on their time, nor is there any thought put into the amount of time required to meet each consumers’ needs. If you happen to get to the airline counter a split second too late, you may need to wait for 30 minutes behind the confused tourists who want to talk to the manager about why their tickets aren’t refundable.
So while Google tries in vain to apply yield management to TV, radio, and print advertising, I wish someone would try to take Google’s success in AdWords and apply it to my everyday life. Here’s an example of my modest proposal for everyday yield management: When you get to the airport to check-in, you are given an estimated wait time and estimated transaction time – for example, your wait time is 25 minutes and the average transaction time is 5 minutes. If you accept these two variables, you pay nothing for this transaction. If, however, you want to shorten your wait time, you can bid money for this service – you bid the maximum dollar amount you are willing to pay to shorten the wait time; the airline then uses a bid auction to rank order bidders. So if I bid $20 to shorten my time in line, and you bid $15, I pay $15.01 to move in front of you.
But wait – there’s more! The next twist to this model is that you also have to bid on the length of your transaction. If you need to book a complex four city flight with long stop-overs and use frequent flyer miles for part of the trip, you need to increase the estimated length of your transaction. So instead of a five minute transaction, you need to bid for a 30 minute transaction. If you under-estimate your transaction by more than 10%, you have to pay overage charges by the minute – this is an incentive to prevent people from gaming the system.
Now you factor in your max bid and your estimated time and you derive a “cost per transaction minute” bid. In other words, if I bid $20 for a 30 minute transaction and you bid $15 for a 5 minute transaction, my cost per transaction is $.66 a minute and your cost per transaction is $3 a minute. Because the airline will make more from serving you first (and because you have placed a higher value on your time), you now win the auction.
The final element of this system designates a “quality score” for each counter agent. Counter agents are measured by the amount of time over or under the estimated transaction times they complete transactions. So if you bid 30 minutes but happen to end up with a customer service agent who is 75% slower than the average agent, you are ‘comped’ an extra 25% of time. At the end of the month, agents receive bonuses and promotions based on their ranking versus the average time.
The result of this system is that both the airline and the customer are incentivized to serve the most needy customers first, and to serve them in an efficient manner. Customers who are willing to wait a little longer or need a little more help are pushed to the back of the line, while customers in a hurry or who simply can’t stand lines pay a little extra to get through more quickly.
By the way, I recognize that such a system would be a regressive tax which inordinately penalizes poorer people more than richer ones, and I suppose if you really wanted to get sophisticated, you could find a way to factor in income to the equation, but for the time being, this is just an example, so give me a break!
My point, though, is that Google has taught me that there has got to be a better way to create efficiency than the ‘first come first serve’ world we currently live in. Happy customers, happy retailers, better customer service – let’s Googlize the world now!