7. $800
8 megapixels
Camera A
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
8. $800 $1,000
8 megapixels 10 megapixels
Camera A Camera B
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
9. $800 PRICE $1,000
8 megapixels QUALITY 10 megapixels
Camera A Camera B
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
10. Assume identical on all other dimensions
(Or, that these two dimensions predict most of your choice)
$800 PRICE $1,000
8 megapixels QUALITY 10 megapixels
Camera A Camera B
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
11. Sensitive to quality
Sensitive to price
(megapixels)
$800 PRICE $1,000
8 megapixels QUALITY 10 megapixels
Camera A Camera B
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
12. Sensitive to quality
Sensitive to price
(megapixels)
50% 50%
$800 PRICE $1,000
8 megapixels QUALITY 10 megapixels
Camera A Camera B
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
13. $800 $1,000 $1,200
8 megapixels 10 megapixels 9 megapixels
Camera A Camera B Camera C
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
14. “Camera B is “Wait, this is a
way better than pretty bad
camera C…” option…”
$800 $1,000 $1,200
8 megapixels 10 megapixels 9 megapixels
Camera A Camera B Camera C
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
15. “Camera B is “Wait, this is a
way better than pretty bad
camera C…” option…”
40% 58% 2%
$800 $1,000 $1,200
8 megapixels 10 megapixels 9 megapixels
Camera A Camera B Camera C
Huber, Payne, & Puto, 1982; Dhar & Simonson, 2003
29. But I don’t
That’s not want that
enough for much…
me…
Tall Grande Venti Trenta Quadragrane
Simonson, 1989
30. Seems about But I don’t
That’s not right want that
enough for much…
me…
Tall Grande Venti Trenta Quadragrane
Simonson, 1989
31. Seems about But I don’t
That’s not right want that
enough for much…
me…
Tall Grande Venti Trenta Quadragrane
preference
Simonson, 1989
32. Seems about But I don’t
That’s not right want that
enough for much…
me…
Gravitate towards the center
“compromise” option
Tall Grande Venti Trenta Quadragrane
preference
Simonson, 1989
33. Is the Amazon river
longer or shorter than
5,000 miles?
34. Is the Amazon river
longer or shorter than
5,000 miles?
How long is the
Amazon river?
35. Is the Amazon river
longer or shorter than
3,000 miles?
36. Is the Amazon river
longer or shorter than
3,000 miles?
How long is the
Amazon river?
37. Is the Amazon river Is the Amazon river
longer or shorter longer or shorter
than 5,000 miles? than 3,000 miles?
Tversky & Kahneman, 1974
38. Is the Amazon river Is the Amazon river
longer or shorter longer or shorter
than 5,000 miles? than 3,000 miles?
4,500 3,500
Tversky & Kahneman, 1974
51. Choice Overload
We want lots of
options….
But we’re happier
with fewer options
52. Context effects
• Bar, cameras, and Economist.com –
asymmetric dominance
• Coffee & popcorn – compromise
53. Context effects
• Bar, cameras, and Economist.com –
asymmetric dominance
• Coffee & popcorn – compromise
Anchoring
• Amazon river
• Social security number and bidding
54. Context effects
• Bar, cameras, and Economist.com –
asymmetric dominance
• Coffee & popcorn – compromise
Anchoring
• Amazon river
• Social security number and bidding
Choice Overload
• Jam
These stories that I’m about to tell are similar to what Dan Ariely talks about in his book Predictably Irrational. There’s another great book on the science of decisions and shopping called Why We Buy by Paco Underhill, that helps you get in the mind of the consumer. But before I jump into the effects, I want to start out by discussing why all of the effects that I am about to describe happen in the first place
These stories that I’m about to tell are similar to what Dan Ariely talks about in his book Predictably Irrational. There’s another great book on the science of decisions and shopping called Why We Buy by Paco Underhill, that helps you get in the mind of the consumer. But before I jump into the effects, I want to start out by discussing why all of the effects that I am about to describe happen in the first place
So because our brain has a lot of inputs, we use heuristics. Rules of thumbs that simplifies all of the data our brains have to deal with. Heuristics exist throughout our cognitive processes - heuristics in vision, for example, allow us to see the world in three dimensions from the two dimensional image projected on our retina. Our mind uses these shortcuts to process and understand information just like we use them to figure things out in other domains as well. So in statistics, we use the rule of 72 to estimate the number of years it takes for an investment to double at a given interest rate. So if we have a 9 percent interest rate per annum and we want to know how many years it will take for an investment to double, we divide 72 by 9, and we get 8 years. The exact number is 8.04 years, but the heuristic is generally good enough. But, as we know, sometimes these shortcuts don’t work. The rule of 72, for example, breaks down at high interest rates. And optical illusions are great examples of where heuristics in our vision system break down.
And because heuristics don’t always work, people are *sometimes*irrational. Particularly when it comes to decisions, we make mistakes in very, very systematic ways – ways that the field of consumer psychology and behavioral economics is designed to understand. So let’s dive into this field with a basic anecdote that illustrates one of the most important effects in behavioral economics. Here’s how it goes.
When I walk into a bar, I have a strategy. I bring a friend who looks a lot like me, but is slightly less attractive. Why do I do this? It’s because by bringing a friend who is similar but looks slightly less attractive, I appear more attractive. It shifts the comparison from me versus other guys… to me versus my friend, because we’re similar and easy to compare. And that makes it more likely for people to pick me in the bar. Now this is sort of a silly example, but you can imagine it within the context of products.
So let’s take a step back and think about how people buy products.
So let’s take a step back and think about how people buy products.
So let’s take a step back and think about how people buy products.
So let’s take a step back and think about how people buy products.
So let’s take a step back and think about how people buy products.
So let’s take a step back and think about how people buy products.
Increase sales of a desired product –imagine that product has a higher profit margin – just by changing the options that are presented. So this is clearly very powerful, and you can imagine that this basic effect – called the asymmetric dominance effect – can be applied to a wide range of situations in e-commerce. Not only can it get people to buy particular products, but it can also get people to buy who otherwise didn’t. In the academic literature, just adding Camera C increased sales by 100-260%.
Increase sales of a desired product –imagine that product has a higher profit margin – just by changing the options that are presented. So this is clearly very powerful, and you can imagine that this basic effect – called the asymmetric dominance effect – can be applied to a wide range of situations in e-commerce. Not only can it get people to buy particular products, but it can also get people to buy who otherwise didn’t. In the academic literature, just adding Camera C increased sales by 100-260%.
Increase sales of a desired product –imagine that product has a higher profit margin – just by changing the options that are presented. So this is clearly very powerful, and you can imagine that this basic effect – called the asymmetric dominance effect – can be applied to a wide range of situations in e-commerce. Not only can it get people to buy particular products, but it can also get people to buy who otherwise didn’t. In the academic literature, just adding Camera C increased sales by 100-260%.
And finally here’s a classic example from pricing that illustrates how these effects can be used. If you haven’t seen this before, basically, the (click) Economist.com wanted to sell more of the expensive subscription, so they added in a really bad deal in the middle to make the more expensive option more attractive (click). And sales of the $125 package increased significantly. They didn’t change their products, they just changed the context of the choices that they presented.
And finally here’s a classic example from pricing that illustrates how these effects can be used. If you haven’t seen this before, basically, the (click) Economist.com wanted to sell more of the expensive subscription, so they added in a really bad deal in the middle to make the more expensive option more attractive (click). And sales of the $125 package increased significantly. They didn’t change their products, they just changed the context of the choices that they presented.
And finally here’s a classic example from pricing that illustrates how these effects can be used. If you haven’t seen this before, basically, the (click) Economist.com wanted to sell more of the expensive subscription, so they added in a really bad deal in the middle to make the more expensive option more attractive (click). And sales of the $125 package increased significantly. They didn’t change their products, they just changed the context of the choices that they presented.
And it’s part of a much larger family of effects called context effects. (click) It’s called a context effect because it happens when people pay attention to the context of their choices. You may have seen this example before. Clearly, this says “The Cat,” but since the A and the H are actually identical, you only know what they mean based on the context of the letters before and after the ambiguous character. This isn’t just true for visual perception, it’s true for our choices – the context and content of the options around us makes a really big difference in how we understand products.
And it’s part of a much larger family of effects called context effects. (click) It’s called a context effect because it happens when people pay attention to the context of their choices. You may have seen this example before. Clearly, this says “The Cat,” but since the A and the H are actually identical, you only know what they mean based on the context of the letters before and after the ambiguous character. This isn’t just true for visual perception, it’s true for our choices – the context and content of the options around us makes a really big difference in how we understand products.
And it’s part of a much larger family of effects called context effects. (click) It’s called a context effect because it happens when people pay attention to the context of their choices. You may have seen this example before. Clearly, this says “The Cat,” but since the A and the H are actually identical, you only know what they mean based on the context of the letters before and after the ambiguous character. This isn’t just true for visual perception, it’s true for our choices – the context and content of the options around us makes a really big difference in how we understand products.
Here’s another one
Here’s another one
Here’s another one
Here’s another one
Here’s another one
But now imagine that we add some extra options. And this happens everywhere.
But now imagine that we add some extra options. And this happens everywhere.
But now imagine that we add some extra options. And this happens everywhere.
But now imagine that we add some extra options. And this happens everywhere.
But now imagine that we add some extra options. And this happens everywhere.
But now imagine that we add some extra options. And this happens everywhere.
So here’s another one, and this one involves participation, so put on your thinking caps, Lenovo. Here’s the question: Is the Amazon river longer or shorter than 5,000 miles?
Great, now estimate…how long is the Amazon river?
Now what if I had flipped the question around and asked:Is the Amazon river longer or shorter than3,000 miles? Most people know that it’s probably longer than 3,000 miles, but if I then ask you
Now how long is the Amazon river…your best estimates will be significantly lower than your estimates the first time around.
So when I prompt you with a higher number, you guess a higher number, but if I prompt you with a lower number, you guess a lower number. And it’s because you were anchored (click) to the higher number.
So when I prompt you with a higher number, you guess a higher number, but if I prompt you with a lower number, you guess a lower number. And it’s because you were anchored (click) to the higher number.
And this is called an anchoring effect because people become anchored to specific pieces of data.
And this is called an anchoring effect because people become anchored to specific pieces of data.
Now the Amazon river example is interesting, but here’s an even crazier and very popular example. Researchers ran an experiment in which people came to an auction and wrote down the last two digits of their social security card as their bidding number. (click)
The last two digits of some people’s social security cards was lower, and some were higher. (click) The crazy part, which you can probably predict now, was that people who had higher bidding numbers actually bid more money – 60% to 120% more – for the same products. And of course, that’s crazy. People can also become anchored to non-numerical attributes as well.
The last two digits of some people’s social security cards was lower, and some were higher. (click) The crazy part, which you can probably predict now, was that people who had higher bidding numbers actually bid more money – 60% to 120% more – for the same products. And of course, that’s crazy. People can also become anchored to non-numerical attributes as well.
Okay, let’s jump into our third and final category of behavioral economics effects that are relevant to e-commerce. Imagine you are a store manager and you’re setting up one of those displays in a grocery store to feature a new brand or product line. As a store manager, you have options. You can either show 6 types of jam or 24 types of jam. Now, which option do you think attracts more attention – which display would get more people to stop and look? Now, which display do you think generated more sales? The display with fewer choices actually generated ten times the number of sales. Lots of options sounds good, but you get paralyzed with choice. Analysis paralysis. How do you know that you got the right jam? They’re hard to compare and you just don’t know what to do, so you leave. And this is what we cause choice overload.
Okay, let’s jump into our third and final category of behavioral economics effects that are relevant to e-commerce. Imagine you are a store manager and you’re setting up one of those displays in a grocery store to feature a new brand or product line. As a store manager, you have options. You can either show 6 types of jam or 24 types of jam. Now, which option do you think attracts more attention – which display would get more people to stop and look? Now, which display do you think generated more sales? The display with fewer choices actually generated ten times the number of sales. Lots of options sounds good, but you get paralyzed with choice. Analysis paralysis. How do you know that you got the right jam? They’re hard to compare and you just don’t know what to do, so you leave. And this is what we cause choice overload.
Okay, let’s jump into our third and final category of behavioral economics effects that are relevant to e-commerce. Imagine you are a store manager and you’re setting up one of those displays in a grocery store to feature a new brand or product line. As a store manager, you have options. You can either show 6 types of jam or 24 types of jam. Now, which option do you think attracts more attention – which display would get more people to stop and look? Now, which display do you think generated more sales? The display with fewer choices actually generated ten times the number of sales. Lots of options sounds good, but you get paralyzed with choice. Analysis paralysis. How do you know that you got the right jam? They’re hard to compare and you just don’t know what to do, so you leave. And this is what we cause choice overload.
Okay, let’s jump into our third and final category of behavioral economics effects that are relevant to e-commerce. Imagine you are a store manager and you’re setting up one of those displays in a grocery store to feature a new brand or product line. As a store manager, you have options. You can either show 6 types of jam or 24 types of jam. Now, which option do you think attracts more attention – which display would get more people to stop and look? Now, which display do you think generated more sales? The display with fewer choices actually generated ten times the number of sales. Lots of options sounds good, but you get paralyzed with choice. Analysis paralysis. How do you know that you got the right jam? They’re hard to compare and you just don’t know what to do, so you leave. And this is what we cause choice overload.
Okay, let’s jump into our third and final category of behavioral economics effects that are relevant to e-commerce. Imagine you are a store manager and you’re setting up one of those displays in a grocery store to feature a new brand or product line. As a store manager, you have options. You can either show 6 types of jam or 24 types of jam. Now, which option do you think attracts more attention – which display would get more people to stop and look? Now, which display do you think generated more sales? The display with fewer choices actually generated ten times the number of sales. Lots of options sounds good, but you get paralyzed with choice. Analysis paralysis. How do you know that you got the right jam? They’re hard to compare and you just don’t know what to do, so you leave. And this is what we cause choice overload.
And we buy more when we’re happier
And we buy more when we’re happier
And we buy more when we’re happier
So just to recap, we reviewed (above). Any questions on any of these effects?
So just to recap, we reviewed (above). Any questions on any of these effects?
So just to recap, we reviewed (above). Any questions on any of these effects?
So now let’s transition into what Cognection does. (click) And our core product is cognitive comparisons.
Our core technology essentially converts a number of effects from behavioral economics into algorithms. This has a wide range of applications Since we’re filing for a provisional patent, I can’t go into much detail, but let me show you how we apply this technology in the real world. And I’ll start with an example of cameras.
Our data engine starts out with those geek tables. (click) They include all of the parameters that anyone could ever be interested in. They are way too complicated and not suited for comparisons. We then algorithmically tease out the most important attributes to consumers both in aggregate and individually.
Our data engine starts out with those geek tables. (click) They include all of the parameters that anyone could ever be interested in. They are way too complicated and not suited for comparisons. We then algorithmically tease out the most important attributes to consumers both in aggregate and individually.
Like a salesman in a store, we watch how people browse and run microexperiments to extract a user’s preferences. That’s key to understanding consumers. We present them with recommendations that are designed to test their preferences and understand what features they’re most interested in and the price point they’re looking for. Given those preferences, we find the best product out of the hundreds or thousands of options out there, and explicitly highlight it (click). So earlier I said that too many options is bad, but it’s really not. Lots of options are good in general, but you need to show the right options to the right people, and we can do that since we understand preferences. We simplify choices by showing only the relevant products, like in the (click) jam example of choice overload. You do this manually through the segmentation of laptops, for example. I have a T410s for business, but if I spent twenty minutes looking at the Essential line, you’d probably guess that I’m looking for a basic home laptop. So, anyway, given the best and relevant products for each user, (click) weadd in those similar products that are designed to make the better deal look even more attractive – the (click) products on the right are like my slightly less attractive friend that make the better deal more attractive. (click) So you see us using that man in the bar effect – asymmetric dominance – as well as the (click) coffee effect – the compromise effect – to help consumers buy the better deal. And, like the river length example – (click) the anchoring effect – we include a higher priced product on the right to get them thinking about spending more money and shifting towards a higher end product that might interest them. That forms our core product, Cognitive Comparisons.
Like a salesman in a store, we watch how people browse and run microexperiments to extract a user’s preferences. That’s key to understanding consumers. We present them with recommendations that are designed to test their preferences and understand what features they’re most interested in and the price point they’re looking for. Given those preferences, we find the best product out of the hundreds or thousands of options out there, and explicitly highlight it (click). So earlier I said that too many options is bad, but it’s really not. Lots of options are good in general, but you need to show the right options to the right people, and we can do that since we understand preferences. We simplify choices by showing only the relevant products, like in the (click) jam example of choice overload. You do this manually through the segmentation of laptops, for example. I have a T410s for business, but if I spent twenty minutes looking at the Essential line, you’d probably guess that I’m looking for a basic home laptop. So, anyway, given the best and relevant products for each user, (click) weadd in those similar products that are designed to make the better deal look even more attractive – the (click) products on the right are like my slightly less attractive friend that make the better deal more attractive. (click) So you see us using that man in the bar effect – asymmetric dominance – as well as the (click) coffee effect – the compromise effect – to help consumers buy the better deal. And, like the river length example – (click) the anchoring effect – we include a higher priced product on the right to get them thinking about spending more money and shifting towards a higher end product that might interest them. That forms our core product, Cognitive Comparisons.
Like a salesman in a store, we watch how people browse and run microexperiments to extract a user’s preferences. That’s key to understanding consumers. We present them with recommendations that are designed to test their preferences and understand what features they’re most interested in and the price point they’re looking for. Given those preferences, we find the best product out of the hundreds or thousands of options out there, and explicitly highlight it (click). So earlier I said that too many options is bad, but it’s really not. Lots of options are good in general, but you need to show the right options to the right people, and we can do that since we understand preferences. We simplify choices by showing only the relevant products, like in the (click) jam example of choice overload. You do this manually through the segmentation of laptops, for example. I have a T410s for business, but if I spent twenty minutes looking at the Essential line, you’d probably guess that I’m looking for a basic home laptop. So, anyway, given the best and relevant products for each user, (click) weadd in those similar products that are designed to make the better deal look even more attractive – the (click) products on the right are like my slightly less attractive friend that make the better deal more attractive. (click) So you see us using that man in the bar effect – asymmetric dominance – as well as the (click) coffee effect – the compromise effect – to help consumers buy the better deal. And, like the river length example – (click) the anchoring effect – we include a higher priced product on the right to get them thinking about spending more money and shifting towards a higher end product that might interest them. That forms our core product, Cognitive Comparisons.
Like a salesman in a store, we watch how people browse and run microexperiments to extract a user’s preferences. That’s key to understanding consumers. We present them with recommendations that are designed to test their preferences and understand what features they’re most interested in and the price point they’re looking for. Given those preferences, we find the best product out of the hundreds or thousands of options out there, and explicitly highlight it (click). So earlier I said that too many options is bad, but it’s really not. Lots of options are good in general, but you need to show the right options to the right people, and we can do that since we understand preferences. We simplify choices by showing only the relevant products, like in the (click) jam example of choice overload. You do this manually through the segmentation of laptops, for example. I have a T410s for business, but if I spent twenty minutes looking at the Essential line, you’d probably guess that I’m looking for a basic home laptop. So, anyway, given the best and relevant products for each user, (click) weadd in those similar products that are designed to make the better deal look even more attractive – the (click) products on the right are like my slightly less attractive friend that make the better deal more attractive. (click) So you see us using that man in the bar effect – asymmetric dominance – as well as the (click) coffee effect – the compromise effect – to help consumers buy the better deal. And, like the river length example – (click) the anchoring effect – we include a higher priced product on the right to get them thinking about spending more money and shifting towards a higher end product that might interest them. That forms our core product, Cognitive Comparisons.
Like a salesman in a store, we watch how people browse and run microexperiments to extract a user’s preferences. That’s key to understanding consumers. We present them with recommendations that are designed to test their preferences and understand what features they’re most interested in and the price point they’re looking for. Given those preferences, we find the best product out of the hundreds or thousands of options out there, and explicitly highlight it (click). So earlier I said that too many options is bad, but it’s really not. Lots of options are good in general, but you need to show the right options to the right people, and we can do that since we understand preferences. We simplify choices by showing only the relevant products, like in the (click) jam example of choice overload. You do this manually through the segmentation of laptops, for example. I have a T410s for business, but if I spent twenty minutes looking at the Essential line, you’d probably guess that I’m looking for a basic home laptop. So, anyway, given the best and relevant products for each user, (click) weadd in those similar products that are designed to make the better deal look even more attractive – the (click) products on the right are like my slightly less attractive friend that make the better deal more attractive. (click) So you see us using that man in the bar effect – asymmetric dominance – as well as the (click) coffee effect – the compromise effect – to help consumers buy the better deal. And, like the river length example – (click) the anchoring effect – we include a higher priced product on the right to get them thinking about spending more money and shifting towards a higher end product that might interest them. That forms our core product, Cognitive Comparisons.
We can then also add in other attributes like product rating or (click) even discount to improve the consumer experience and enable customers to find the best deal.
We can then also add in other attributes like product rating or (click) even discount to improve the consumer experience and enable customers to find the best deal.
And fundamentally that’s what we’re trying to do when we buy a product – justify our choices and have confidence that not only did we just get a good deal, but the best deal for me.
And, if there are two products that are (click) basically identical…but one is overstocked or has a higher profit margin, we automatically display the one that is better for the retailer, but nearly identical to the consumer.
And, if there are two products that are (click) basically identical…but one is overstocked or has a higher profit margin, we automatically display the one that is better for the retailer, but nearly identical to the consumer.
And, if there are two products that are (click) basically identical…but one is overstocked or has a higher profit margin, we automatically display the one that is better for the retailer, but nearly identical to the consumer.
And, if there are two products that are (click) basically identical…but one is overstocked or has a higher profit margin, we automatically display the one that is better for the retailer, but nearly identical to the consumer.
We typically embed on the product pages of retail websites and serve up recommendations customized for each individual shopper dynamically as they browse. We’ve run several tests of Cognitive Comparisons with our other beta customers. On the website of Electronic Express, a large electronics retailer in the southern USA with 19 brick and mortar stores and a pretty decent online presence, we had the following results.
First, (click) 24% of site visitors were actively engaged and clicking through our recommendations. We have yet to do any formal user testing, so this is huge for us.
Second, and most importantly, conversions improved by 12.8% for people who clicked on our recommendations. Not only was this a huge increase, but it is statistically significant at a 95% confidence interval as well. Simply put, more people are buying. That’s another huge value prop for retailers, and a great source of product validation. We’re confident that with new optimizations, we can push those numbers even higher. As you can tell, these effects are very powerful, and recommendations are just the start.