Discover more from The Reputation Algorithm
What Is A Happiness Algorithm?
How to measure that most crucial of customer sentiments.
This CMSWire article by Alan J. Porter got me thinking of what consumer behavior signals might go into a mythical “happiness algorithm.”
The Reputation Algorithm is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
It’s a fascinating piece about measuring customer satisfaction. (If you watched the video, you can skip to the next section.)
Of course, everyone wants to know how satisfied their customers are. It’s a standard business metric.
There are all kinds of things that people do to try and measure customer satisfaction. Among them are really long, in-depth surveys that nobody takes. The problem with those is either volume—you don't get enough data—or self-selection; only a certain demographic, a certain type of person is going to spend a lot of time filling out surveys.
But there's this company cited in this article called HappyOrNot. It's a Finnish startup that measures customer satisfaction but they do it in a very particular way.
They have these customer-operated devices that look like Fisher Price toys. They're battery-powered terminals with four big push buttons.
One button is dark green and it has a smiley face on it.
Another one is light green and it has a little less smiley face on it.
Another is light red and it has sort of a frowny face.
And the last is dark red and has a very frowny frowny face.
You see where this is going.
So these are installed in stores. And as customers leave the store, a small sign asks them to rate their experience.
All they have to do is make a simple choice: Just are you happy or not? Basically, slap a button. And that records that interaction, that customer sentiment, and a timestamp.
HappyOrNot was hired by a European gas station chain to measure customer satisfaction at 150+ outlets. One gas station rapidly emerged as the leader in customer sentiment. And then there was one that just fell away behind; it was the worst performer.
So the executives swapped the managers for those locations. And within a short period of time, the worst performer quickly rose to the top while the best performer plummeted to the bottom.
So you know what the problem was, right?
So the problem with surveys or polling is just gathering the responses. These HappyOrNot kiosks gather enough data because they make it easy for people to give you their opinion. And, assuming your location has enough foot traffic, you enough volume so that you have legitimate data.
They can register thousands of sentiments a day. They're self-explanatory, right? Customers know how to use them. You don't have to explain them.
Customers can use them without breaking a stride as make their way out the door. Simple and easy. The responses are anonymous.
So there's another company that was cited in the article, a Swedish sofa retailer hired them to help understand a sales problem.
Revenues were high during the late afternoon and evening but low during the morning and early afternoon. The customers felt most satisfied during the hours when sales were low and least satisfied during the hours when sales were high.
The executives realized they'd be looking at the problem the wrong way for years. They hadn't considered the possibility that the day should be even higher. So they added more salespeople in the afternoon and evening and the sales improved.
So what are the metrics HappyOrNot is collecting? What are the signals that help us measure happiness?
We’ve got the data of where the HappyOrNot kiosks are located so we can measure customer sentiment by geography.
We’ve got our four data points:
Kinda Unhappy, and
And we’ve got out timestamp data so we can measure during what daypart customers are most happy versus when they are most unhappy.
Through that all that data from 150-plus locations together in a data warehouse and you’ve got the foundation for an algorithm that can visualize customer happiness.
While HappyOrNot poses the specific question of whether or not you are happy and provides direct instructions for how to express yourself, other metrics intended to measure happiness are not so straightforward.
Net Promoter Scores
The Net Promoter Score that asks you to rate on a 10-point scale (or sometimes five point) if you would refer a business or product to a friend or colleague kind of measures happiness. But that’s more of a satisfaction metric.
The CMSWire article I cited at the outset has a few ideas.
If people are using keywords that indicate happiness, that can be a source of sentiment data. People will express their emotions in their search queries. The search phrase “I love Starbucks” and variations thereof average 390 queries a month at Google.
Conversely, people will use keywords in their queries that express unhappiness. Seven years ago I wrote a blog post about my unhappiness with the city of Minneapolis after getting towed due to their deliberately confusing parking signs. To this day, that post has consistently earned traffic from people searching Google using the phrase “Mineapolis sucks.”
Measuring that type of search volume over time can give you trending data on how happy people are with your organization, brand, or product. Or city.
The star ratings at online review sites are an obvious source of signals that imply or suggest happiness but that, too, like Net Promoter Scores, is more of a satisfaction metric.
The language used within online reviews, however, is often a direct signal of happiness or unhappiness.
Some examples of words associated with happiness that you’d likely find used in online reviews:
Conversely, these are some words you might find in negative reviews:
You can mine social media for conversations about your organization, brand or product that include any of the above words as another way of tracking consumer happiness.
The Happiness Algorithm Infrastructure
While HappyOrNot may come closest to directly measuring happiness in the moment, all of these methods are nevertheless fairly imprecise.
If we were to build a happiness algorithm, what infrastructure would we need, what components would be required, and who would be best positioned to build such an infrastructure?
My money would be on Apple.
Think about all the technology Apple offers consumers.
I own the MacBook Air on which I write The Reputation Algorithm.
I have an iPhone and an iPad Mini, which I use to communicate and to consume content.
I use my AirPods to listen to music and podcasts.
I have AppleTV+ to watch entertainment.
The only Apple product I don’t currently own is an Apple Watch (though I’m trying to convince myself I should probably buy one).
All of these devices know everything I’m doing and pretty much everything I’m consuming, so there’s an abundant data set of stimuli that can potentially prompt feelings of happiness.
I read a lot of content on my MacBook: Websites and blogs and PDFs and email. I watch YouTube videos and listen to music on Spotify and podcasts on Google Podcasts or Apple Podcasts. I interact on social media via my browser.
My MacBook tracks my keyboard and my cursor movements so it can measure those interactions mapped to the content I consume. It also has a camera and a microphone, so theoretically it can listen to and see my reactions to content I consume on the device.
My iPhone and iPad also have those capabilities in addition to GPS data harvested from my wifi and celluar connections, so it can track my movements through space and time.
It also has a gyroscope that tracks the devices’ own orientation in space. It can tell if the front camera is facing down or up or if the user is holding the device in a manner indicating it is being actively used.
Use of my AirPods are indicitive of active listening and because the content is streaming from my Apple device, Apple knows the nature of that content as well.
Throw an Apple Watch into the mix, and you’ve got a monitoring system that can theoretically collect direct signals—stimuli and reaction—that could much more precisely measure happiness than the metrics we’ve previously discussed.
The camera on these devices could track facial expressions and through facial coding, map expressions of happiness to exact moments of consumed content. Likewise, the microphones on these devices could theoretically hear squeals of delight, for example, and again, map that reaction precisely to content experienced at the moment of delight.
Finally, the Apple Watch’s biometrics capabilities could theoretically measure our bodies’ unconscious physical reactions to moments of happiness.
The Apple Watch can measure your heart rate, has an electrocardiogram to measure your heart’s electrical pulses, and can measure your rate of breathing.
As this heatmap from Psyblog illustrates, your whole body lights up when you experience happiness.
You’ve heard the expression “jumping for joy”? If you’re holding your iPhone or iPad, your device can measure that.
When you’re happy, your face flushes. Theoretically, your device’s camera could capture that. If your heart races, your Apple Watch would know.
This is, of course, is all speculation but it’s certainly plausible.
Am I missing something? What do you think? Let me know in the comments.
My Media Diet
I’m changing up the format for this section. I think it makes it easier to read by linking to the stories only at the end of the quote. Whaddya think? Better?
What I’m Reading
CNBC - In 1938, Harvard researchers embarked on a study that continues to this day to find out: What makes us happy in life?
The researchers gathered health records from 724 people from all over the world, asking detailed questions about their lives at two-year intervals. Read the full article.
“I don’t care too much for money. Money can’t buy me love.” The Beatles get to the heart of the matter: The key to happiness is relationships.
- Twitter’s decline is paving the way for other platforms to build next-generation replacements. And now the biggest player in the game is getting involved: Meta is in the early stages of building a dedicated app for people to post text-based updates…
The most remarkable aspect of the project is that Meta plans for the network to be decentralized. While the company would not elaborate beyond its statement, in a decentralized network individual users are typically able to set up their own, independent servers and set server-specific rules for how content is moderated. Read the full article.
The question is if Meta/Facebook’s bureaucracy will prevent it from being nimble enough to exploit the moment. Elon Musk’s lunacy at Twitter opened the floodgates for competitors, with Mastodon, Post.news (which I’ll write about soon), and others already providing safe haven for Twitter refugees. By the time Meta/Facebook enters the fray, will it be too late? Not that Meta/Facbook couldn’t forcibly own the market through its sheer size and power. But by the time they get around to it, will people have already settled into their new communities? Will the market have already fragmented? This fragmentation and atomization of social media was bound to happen sooner or later. With the significant exception of LinkedIn, during the past several years the primary social networks have been increasingly dominated by trolling racists, cabals of conspiracy theorists, human traffickers harvesting victims, adversarial governments sowing division, insurrectionists planning to overthrow democracy, and run-of-the-mill scam artists pilfering pockets. Of course people will flee when they can find a saner alterative. Twitter was just the precipitating event.
- If you click through the thread above from media researcher Roland Meyer, you’ll see a bunch of absolutely fascinating examples of a generalized AI art aesthetic. God, I am so into this idea. In fact, last month, I wondered if this exact thing might happen. Whether the human brain would start to notice all the hallmarks of generative-AI art and start to recognize it as a specific style…
I’m also a big believer in the idea that to name something is effectively kill it. Or, at the very least, by naming it you sort of lock it in place. Which makes me wonder if generative-AI art is going to be looked back on as a revolution in art for a very different reason than AI evangelists think it will be. It’s basically going to make everyone want to make stuff that DOESN’T look like this. Read the full article.
This is fascinating. It is in line with what I’ve thinking of how we humans may react to the generative-AI revolution: That once our culture is saturated with AI-generated content, it will be demonstrably human-created content that people will seek out. Conversational podcasts are an example. Live performances. Here’s the Twitter threadmentions:
Vice - Between August and September of last year, some users between the ages of 18 and 25 on platforms including Facebook, Discord, Tumblr, and Telegram who posted one of more than a thousand keywords ranging from “depression” to “sewer-slide” suggesting they were at risk of harming themselves were, without warning, directed to a chatbot.
This was part of an experiment carried out by the founder of a controversial mental-health nonprofit called Koko and a Stony Brook University professor who also runs a suicide intervention consultancy…Any experiment like the one carried out last year would necessarily raise questions about the ethics of researching anti-suicide interventions in social media, and the role of chatbots and scripted health interventions in suicide prevention.
The Koko experiment, though, raises further questions, in part because it was carried out as “nonhuman subjects research,” meaning that participants were deprived of a number of protections related to their safety and privacy. Experts consulted by Motherboard described the study using terms ranging from “distasteful” to “inexcusable.” Koko founder Rob Morris, though, defended the study’s design by pointing out that social media companies aren’t doing enough for at-risk users and that seeking informed consent from participants might have led them to not participate. Read the full article.
Well, yeah, of course it would have lead to fewer participants. But laziness is no excuse for unethical research. Especially on vulnerable participants. Not that trying to find a more effective way of getting people at risk of suicide they help they need is a bad idea; it’s just that informed consent is pretty damn important.
Politico - [TikTok’s] plan — a bid to stave off more restrictions on its app’s use — centers around keeping more European users’ data on servers in Europe and allowing a European security company far-reaching access to audit cybersecurity and data protection controls. Read the full article.
Here’s the thing with the whole “keeping users’ data on in-country servers” argument: You still have no insight into nor control over the algorithm that manipulates TikTok users. Privacy is one thing but radicalization is something entirely different.
Axios - The whistleblower's allegations, which have not been independently seen or verified by Axios, suggest that TikTok overstates its separation from its China-based owner ByteDance, relies on proprietary Chinese software that could have backdoors, and uses tools that allow employees to easily toggle between U.S. and Chinese user data. Read the full article.
An uber-popular social media app among Americans (and American youth, specifically) whose predictive algorithms of your content tastes are so creepy-good that they generate data about what kind of content is most influential with you and which is controlled by a hostile foreign government. What could possibly go wrong?
Phys.org - There are some advantages that come from using an unconventionally spelled brand name, which may explain why companies often select them. However, little is known about how this strategy impacts consumers' beliefs about the brand, and ultimately, their willingness to support it.
New research from the University of Notre Dame finds that in general, consumers are less likely to support uniquely spelled unfamiliar brands, compared with those that use the conventional spelling of the same word…However, the study finds there is no backfire effect when a company's motive for selecting a uniquely spelled name is perceived as sincere. "When a brand name is crowdsourced by consumers rather than chosen by the company, the backfire effect is eliminated," [John] Costello said. Read the rest of the article.
Which explains why I ultimately abandoned the idea of calling my book: Th3 R3put@ti0n @lg0rythm.
The Verge - McNamara touted the streamer’s [Spotify] top position with young listeners and sees video as a big part of that. “Video podcasting is one of the fastest growing areas of podcasting, and we expect that growth to continue,” she said. “Right now, we have 70,000 video creators on the platform and with a lot more on the way.” Video podcasts are also better suited to Spotify’s new homepage, which pulls from TikTok’s playbook. Users will be able to scroll through clips of music, podcasts, and audiobooks. While audio-only podcasts function in this new feed, video podcasts appear to be a better fit. Read the full article.
I’ve been telling clients this for years: Video is the gift that keeps giving. You get a video content asset, yes, but you can strip out the audio and repurpose it as a podcast. You can transcribe it and you’ve got a blog post. You can embed the video on your site, optimze it to earn search traffic, hijack social media’s video-friendly algorithms, and grab a still to use in your newsletter to drive click-through metrics. Plus, done right, it’s more trustworthy if it’s humans talking and more emotionally compelling if it’s a storytelling video.
The New Yorker - For many students, the humanities already are the little bird. Tiffany Harmanian, a senior at A.S.U., is premed, with a neuroscience major (“I come from a family of doctors—I’m Middle Eastern!” she told me), but minors in English and founded a student organization called the Medical Humanities Society. Growing up, she lived in novels and poetry. But it hadn’t occurred to her to go all in as an English major while being premed. “People involved in the humanities may not even need to go to school for what they’re wanting to do,” she said; she didn’t see what studying “The Waste Land” had to do with making it as a poet. “Also, because of the world we’re living in, there’s this desperation for being able to make money at a young age and retire at a young age,” she added.
I asked her what she meant.
“A lot of it has to do with us seeing—they call them ‘influencers’ online,” Harmanian said, pronouncing the word slowly for my benefit. “I’m twenty-one. People my age have crypto. People have agents working on their banking and trading. Instead of working nine to five for your fifteen-dollar minimum wage, you can value your time.” She and her peers had grown up in an age that saw the lie in working for the Man, so they were charging out on their own terms. “It’s because our generation is a lot more progressive in our thinking,” she told me. Read the full article.
I linked to this article last time but this is another passage that struck me because back when social media was taking off, I was impressing upon people that the next generation—what we’re calling Gen Z (a stupid name, by the way)—is the first in human history to grow up with the notion that they, themselves, have an audience for which they must perform. And monetize. That ethos is defining how this generation measures success.
Slow Boring - But even though I personally enjoy my walker’s paradise neighborhood and obviously support things like missing middle housing, dropping parking requirements, and pedestrian-friendly street design, I’ve never liked the frame of the “15-minute city” as a policy goal.
This has been hovering around the periphery of my consciousness for years now, and I’ve generally thought of it as a harmless but slightly misguided framing concept. But I worry now about polarization dynamics. If conservatives respond to the 15MC with conspiracy theories that then get debunked in mainstream media sources, it could lead people who’ve never heard of any of this prior to this year to decide that they need to become super invested in the concept. [Emphasis mine.] Read the full article.
I found this passage sadly revealing of the state our country finds itself in. The fact that it even occurs tothat “polarization dynamics” (a phrase I’m going to steal, by the way) is something to worry about with regard to community planning is quite damning.
StarTribune - [Minnesota Vikings wide receiver, K.J.] Osborn shared the story on his verified Twitter account about how he helped carry a man away from the flaming wreckage of a car accident. During an appearance on NFL Network, Osborn said first responders told the group, "we saved this man's life," after Osborn, the driver of his car, and two bystanders acted quickly upon seeing the accident under a bridge. The 25-year-old said he couldn't sleep Sunday night, and commended the efforts of the three others with him. Read the full article.
Lifelong Vikings fan here! For me, then, this was a heartwarming story. Osborn was drafted by the Vikings in 2020 and was strictly a special teamer, playing only as a kick/punt returner. And for all the world, that appeared what he was destined to be. But he broke out as an excellent receiver the following year and this past year as well. Most of the non-football stories about athletes tend to be controversies, and usually bad ones <COUGH COUGH>Deshaun Watson</COUGH COUGH>. Vikings players used to get in trouble so frequently, I ended up maintaining a list of players who’d been arrested. It’s nice to see such an inspiring story for a change, particularly since Osborn’s a Viking! Aaaannnnd, from inspiring to sad:
StarTribune - Bud Grant, the most iconic sports figure in Minnesota history, died Saturday. He was 95.
Grant was a standout athlete for the Gophers, played professional basketball for the Minneapolis Lakers and starred in the NFL, but made his mark in the state as coach of the Minnesota Vikings, leading them to four Super Bowls in his 18 seasons and earning a spot in the Pro Football Hall of Fame.
"He never embarrassed a player, never criticized or chewed a player out in front of anybody else," former Vikings running back Dave Osborn said. "If he had something he didn't like about you, he'd take you aside and talk to you.
"Bud just had a knack for leading, handling people. He was a great football coach, but Bud could have coached a sport he didn't know anything about because he just knew how to handle people." Read the full article.
A great Minnesotan and a great leader. Thanks for the many, many fond memories, Bud.
What I’m Listening To
- In this episode, the hosts discuss Substack, why it's turning into a platform, and why e-mail may or may not be its future. Listen to the episode.
This is a good take on what Substack really is. I must say, I’ve been really impressed since I started using this platform. I will plan on giving it its due in a forthcoming full post.
The Reputation Algorithm is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.