AI and the Gartner Hype Cycle
Brian Morrissey recently mentioned that we should view the reaction to AI in the context of the Gartner Hype Cycle – which was new to me.
It’s a graphical representation of the phases of adoption of and reaction to new technologies.
Phase 1 is the Technology Trigger. This is when a new technology attracts attention and there’s a lot of excitement about it – positive or negative. We might imagine a connected world with the Internet of Things, new immersive experiences with virtual reality, or a decentralized digital currency with blockchain. In the case of AI, we worry about Skynet and murderous robots.
Phase 2 is the Peak of Inflated Expectations. This is where we get the exaggerated, almost messianic claims about the technology. There are lots of unrealistic projections and too much enthusiasm. The Segway was going to change cities. Virtual reality would allow the disabled to participate in more of life. AI will eliminate everyone’s job.
Phase 3 is the Trough of Disillusionment, which reminds me of the slough of despond in Pilgrim’s Progress. Following on that theme, the proverb says “hope deferred makes the heart sick.” Once we realize the technology won’t live up to its exaggerated expectations, we feel let down. There are security or regulatory problems. People don’t like it as much as they thought they would. It doesn’t integrate with other systems very well.
Phase 4 is the Slope of Enlightenment. The technology matures, and we see there are some practical applications and benefits. The technology starts to gain more realistic and sustainable uses, and organizations start to understand its value. People figure out new ways to use the technology, and it works well in particular niches.
Phase 5 is the Plateau of Productivity. At this point the technology reaches mainstream adoption and becomes part of everyday life or business operations. It’s mature and stable, with well-established practices and a wide user base.
When it comes to AI, we’ve had the technology trigger. I think we’re still climbing towards the peak of inflated expectations. But the trough of disillusionment is coming, which means we’ll start to find problems.
It lacks generalized intelligence.
It hallucinates too much.
It has way too much bias.
It has very limited contextual understanding.
It’s not nearly as creative as we had hoped.
It doesn’t have a good grasp of basic ethics.
It’s too much of a black box. We don’t understand how it works, so we can’t trust it.
It gets boring.
But then we’ll start to find reasonable tasks for AI. It won’t be as scary, and we won’t expect too much of it. But it will be useful.
That, my friends, seems like the probable path for AI.
12
views
6 tips for short videos
I forgot the cold open, plus 5 more tips for short videos
I like to model best practices when I can, so that I set a positive example. I do that pretty well with my print newsletter, but I forgot a few things when I started this podcast.
1 -- you have to capture the viewer's attention in the first few seconds, so it was a mistake to start this podcast with music.
You need to do a quick opening that tells the viewer what the show is about.
2 -- if you want to be on some services, like TikTok or Instagram, you should stay below 60 seconds. I'm not sure that's possible for my subject matter.
3 -- ask questions and try to get some engagement, comments and such from listeners.
4 -- explicitly ask people to subscribe, like, share, and leave a review.
5 -- be clear and direct.
6 -- include captions so people can read the text and keep the volume off.
1
view
The Associated Press creates a test kitchen
Brian Morrissey linked to an article by Mark Stenberg on Adweek about how the Associated Press is trying to diversify its revenue stream. Like every other publisher. But in AP’s case, the’re coming from a different starting point.
They get 80% of their revenue from licensing – which is crazy high – and they only get 5% from advertising. They want to double their ad revenue.
Since they focus so heavily on licensing, you can understand why their current model doesn’t give them much first-party data. In a way they’re writing for their licensees, not for their readers.
They want to fix that, because first-party data is very important in today’s market.
A few things stand out in this story.
The AP focuses on hard news, but Morrissey says advertisers often avoid hard news out of brand safety concerns. But they can still use news to drive traffic. They’ll be live-blog breaking news for that reason.
They want to emphasize direct media buys. Currently they get roughly 85% of their ad inventory from programmatic.
They’re going to do more with photography and especially video, which has higher CPMs than photos.
They want to capture more first-party data.
They want to highlight lifestyle content, like sports and entertainment.
They’re partnering with Sovrn, Nativo, and Taboola. I’ll provide links below.
All of this is to create their own test kitchen to find out what content engages a digital audience. Obviously that will help them in their licensing work.
I used the way back machine to compare their older site with their current site. The major change I noticed is that they moved links to video content higher on the page, included more video content, and called out sports and entertainment content a little more obviously.
I expected to see more promotions for e-newsletters.
But I also expect this will be an on-going mission, and they’ll adjust as they go.
Links
The AP Debuts New Website Designed to Attract More Ad Spend
https://www.adweek.com/media/ap-debuts-website-ad-spend/
Sovrn
https://www.sovrn.com/
Nativo
https://www.nativo.com/
Taboola
https://www.taboola.com/
Matt Bailey’s podcast
https://podcasts.apple.com/us/podcast/endless-coffee-cup-digital-marketing-caffeinated/id1130111127
12
views
What comes after email? A reply to Brian Morrisey's Rebooting Show
On The Rebooting Show, Brian Morrisey and his guests wondered what will replace email.
They made a few salient observations.
1. Email was never meant to be a content delivery platform.
2. Using it that way is a bit of a hack.
3. You can’t embed a video player in an email.
4. You can’t even control fonts very well.
I’d add to that …
5. People can clog your inbox without your consent.
6. Content you want and value is sitting there next to the latest nonsense.
7. It’s not always easy to find things.
So what comes next?
But before I get into that, I’d like to acknowledge that people have been proclaiming the death of email for a long time, and it continues to defy expectations. I’m not going to argue that email is dead or dying. But I do believe there is a better solution for content delivery.
And that’s a generic reader app.
By “generic” I mean that it’s not tied to a brand or to a walled garden, like Facebook.
By “reader app” I mean that it would be where you’d go for everything you want to read.
Here are some of the key ideas.
* If you see an article on the web that you want to read later, a browser plugin would allow you to send the article to your app. Somewhat like Evernote.
* Your app would have an email address, and all your free e-newsletters would be delivered to the app.
* All your subscription content – newsletters and magazines – would also be delivered to the app.
* Ideally, you could also read books through the app.
* The only content in your app would be stuff you want to be there.
* You can take notes in the app.
* You can comment on things you read in the app, and make them public or keep them private.
* You can organize things in folders, or by keywords, however you want to organize it.
How would the app owner make any money off this?
I can see two ways.
First, users pay a monthly fee for the app.
Second, all paid content delivered through the app would pay a surcharge – some percentage of the price of the content.
I pitched an idea like this to an exec at Evernote a few years ago, and he loved it, but he didn’t follow up because he thought it would have ruined his relationship with Apple.
This app has to be premised on a disdain for all walled gardens. There are no exclusive deals. No content is prohibited. This app is designed for the reader first.
Links
The Rebooting Show
https://podcasts.apple.com/us/podcast/the-rebooting-show/id1595625177
1
view
When it comes to book publishing, can personalization beat quality?
I don’t often comment on book publishing, but Bo Sacks recently distributed an article called “The Imitation Game” by Ken Liu, and it caught my attention. The article is about whether AI will start writing all our bestsellers, and Ken’s somewhat tentative point is that an algorithm looking for the most likely “token” given a pre-existing set of tokens (which is approximately the way large language models work) must plateau at some point short of true intelligence.
My reaction to that is … maybe.
In the midst of his struggles with AI, Ken said this.
'Authors know that the idea that “good books” will prevail in the marketplace is a cruel joke. Publishers are very, very bad at connecting readers to books—that is why they are struggling.'
That got me thinking about the concept of good books and best sellers, and so on, so come along with me on a thought experiment by imagining the following lists.
List one is all the books that were written in 2022. That would include everything that was published – including self-published books – but it would also include manuscripts that never got out of the slush pile, for whatever reason.
List two contains good books from list one. Of course we’d have to come up with a standard for what constitutes a good book, but let’s pretend we had some reasonably objective measure for that.
List three is books that were picked up by publishers. That’s a very small percentage of all the books that were written, and, according to Ken, doesn’t overlap very well with the list of good books.
List four is books that became popular.
Let’s go back to the books that were picked up by publishers. Why does a publisher pick any particular book?
The Adam Smith answer is “because they think it will sell,” and I’ll grant you that most of the time, but it’s not the whole story. Sometimes they publish famous people’s books because it looks good for their brand, to curry favor, or for some other reason. And sometimes they publish books to meet diversity quotas, or to push some fashionable idea.
They’re not necessarily choosing “the best” books, nor the ones people most want to read, nor the ones most likely to sell.
Ken imagines a novel-writing AI called “ORWELL” that lives outside of the traditional publishing structure and tries to sell directly to the public. Its books aren’t necessarily better than human-written books, but they’re more targeted to individual readers. The book is exactly what you want to read, with all your foibles and strange tastes.
This could be an example where sales in the long tail – these one-off, personalized books – end up selling more copies than the best sellers.
So … why do we care? The general take-a-way here is that customization has the potential to beat quality, and also, it has the potential to beat all the marketing gimmicks, influencers, slick trailers, great advertising campaigns, etc.
Links
The Imitation Game, by Ken Liu
https://slate.com/technology/2023/06/book-publishing-artificial-intelligence-copyright.html
12
views
The case against personalized advertising
On a plane ride this past weekend I read Bob Hoffman’s Advertising for Skeptics, which is worth your time. Bob’s books are nice and short, which I appreciate because almost all business books could easily be one fifth as long.
He has a lot of interesting things to say about advertising, but I’m going to focus on one point that he makes, which is the idea that personalized, targeted advertising defeats the purpose of advertising.
The case for targeted advertising is that you find the right person at the right time. Why advertise cat food to someone who doesn’t have a cat?
You can see this same general approach in campaigns to get targeted leads. The campaign is trying to find out if you’re the right kind of prospect who has the budget and the authority to buy and you’re ready to buy within the next six months? If not, I have no interest in you.
That is one legitimate side to advertising, and Bob admits that targeted advertising may be appropriate for B2B marketing.
But general advertising relies on common knowledge. If I’m viewing an advertisement, it’s not enough that I see the ad and get the message. For the ad to be effective, I need to know that you see the ad and get the same message.
For example, when I was a kid Gillette had a campaign that the wet head is dead. You don’t have to use oils or creams to control your hair. You can use “the dry look.”
Part of the point of this message is to make you feel out of touch if you are using oils or creams. The wet head is dead. Get with it.
If I was the only person seeing this ad, I might agree or disagree. But if I know that everybody in my high school is seeing this ad, that might affect my behavior.
Not mine, because I’m an ornery cuss and I don’t care too much about those sorts of things. I’ll continue to wear cargo shorts – which are the greatest thing ever – no matter what the fashion moguls say.
But that’s just me. Most people are a little more sensitive to what other people think about a product.
The point of advertising is to link a brand to an image. If you smoke these cigarettes or ride this motorcycle, you’re a tough guy. But if all the sudden smoking those cigarettes associates you with some undesirable image, that’s no good.
It all depends on what everybody else is seeing. And that’s why personalized advertising isn’t always a good idea. Advertising is more effective when I know that everybody else is seeing what I’m seeing.
Sources
=======
https://www.amazon.com/Advertising-Skeptics-Bob-Hoffman/dp/0999230735
14
views
Sneaking past a paywall is stealing
If I steal your car, you don't have a car anymore. It's a physical object. We can't both have it at the same time.
Everybody realizes that's wrong.
But if I make a copy of some digital music you've purchased, you still have it. So have I wronged you?
I may or may not have wronged you, but here are some of the things that I have definitely done.
1. I've violated the rights of the copyright owner. It's up to him to determine who has access to that content.
2. I've deprived the copyright owner of the income he should have received from my ownership of that material.
3. I've undermined the whole social contract and legal framework surrounding intellectual property. Intellectual property laws protect creators precisely so that they're incentivized to innovate and create new content. By stealing a digital object, I'm taking away some of that incentive and I’m making it less likely there will be new, innovative things in the future.
I recently saw a tweet where someone complained that some artist can't sing for crap. The artist replied, "you're right. I sing for money."
Some people will argue that nobody can claim rights to something that has no production cost. It costs essentially nothing to make a copy of a digital recording.
That's true, but it doesn't change any of the things I said previously.
Others will argue that the current system protects corporations and the rich against the rights of the public. But since when does the public have some inherent right to someone else's property?
Some will even say that they're doing the copyright owner a favor by getting a free sample so they can decide if they want to purchase it. Further, in the case of music they're giving the artist more market exposure. Isn't that good?
That’s not a decision for the consumer to make. It's up to the owner to decide when and if and under what terms they want to offer a free sample.
Here's the kicker. All the things that I've said about copying music apply equally well to content behind a paywall.
Sneaking past a paywall is stealing, and publishers need to make that case.
10
views
Publishers need to train their own LLMs for their readers
I’m assuming you’ve used ChatGPT. If so, you’ve probably run into its very annoying super ego. For example, you’re doing some research on nuclear power and it hectors you with grade school morality about the dangers of this and that.
It’s quite annoying.
But there’s a value in that, which is that it’s possible to layer a point of view on top of AI responses.
That’s actually a good thing.
As I’ve mentioned a few times before, tools like ChatGPT will undermine the current search engine model, where I ask a question and Google gives me some homework to do.
“Here, read all these articles and you might find what you’re looking for.”
That was very useful in its time, but now it’s rather quaint. When I ask a question, I want an answer.
The next crucial step is that I might want an answer from a particular point of view.
Let’s say I’m a Lutheran seminary student. I might want to have an answer that’s consistent with Lutheran doctrine. I might even want to get more precise and ask for Philip Melanchthon’s opinion on a particular idea. (Philip was a disciple of Luther.)
You can do that with ChatGPT. I just tried it, in fact, and it did a remarkably good job. It also bypassed that annoying super ego problem. It didn’t keep reminding me that there are other views and blah blah blah. I wasn’t asking about other views. I was asking about Philip’s views, and that’s what it gave me.
Here’s the point for publishers. Let’s say you run an investing newsletter. There are all sorts of different strategies and mindsets and approaches to investing. If you hook up AI to your site, you want it fine-tuned to represent your brand with your voice and your approach.
That’s what publishers need to start doing. Not only do you need to feed your content into your own instance of a large language model. You also need to tinker with the super ego of the AI so that it learns to respond in your voice, with your perspective.
10
views
How to do quadrant analysis
I’m a big advocate of practical tools. Checklists. Forms. That sort of thing. And when I was a VP of operations at a large association, I learned a lot about frameworks.
Today I’m going to talk about a simple one. Quadrant analysis.
You’ve seen these before. It’s a 2x2 matrix where some value on the x axis goes from low to high and a value on the y axis goes from low to high.
If you have a hard time visualizing this, go to my website and see the article about it.
I’m going to start off with two classic examples of quadrant analysis, and then I’ll do one specifically for publishers.
* Urgent vs Important
* Effort vs Impact
* Market Share vs. Advertiser Interest
If you'd like to see the graphics, visit this page.
https://krehbielgroup.com/2023/06/01/quadrant-analysis-for-publishers/
6
views
The Krehbiel "brand involvement" matrix
Bob Hoffman thinks it’s a marketer’s delusion to believe that anybody wants to interact with your brand. I tend to agree with him. But my friends Lev Kaye and Leslie Laredo believe there are some situations where interaction is appropriate.
How do you decide when?
That’s the point of The Krehbiel Brand Involvement Matrix. It’s a first attempt to create a framework for deciding when and how much customer pestering is appropriate.
Please leave your suggestions and comments.
2
views
Best practices are for people who are bad at math
What’s the best length for an email subject line?
What’s the best length for an article on the web?
How many free article views should a reader get before you show the paywall?
There are a lot of so-called “best practices” answers to these sorts of questions, and they’re usually nonsense.
Bo Sacks distributed an article by Andrew Butle called “Publishers: What is the ideal word count for articles and other content?” in which he highlighted the contradictory and confusing answers on the ideal length of an article.
It’s 1,500 words. No, it’s more than 2,000 words. No, actually it should be longer than 7,000 words. And so on.
The same confusion applies at the more strategic level, when we ask questions like “do readers prefer print or digital?”
What are they reading? Is it a book or a magazine or a newspaper?
The problem with this sort of answer is that it comes from taking averages across huge swaths of data. It’s meaningless because it mixes things that don’t fit together. As Andrew points out in the article, different objectives call for different word counts.
And when it comes to email, if you’re mixing B2B and B2C results, or marketing and content emails, your results are meaningless.
Asking for the best length of an article is like asking for the best length of a spoon. Is it for feeding a baby, eating cereal, stirring soup, making chili, or brewing beer?
When asking whether people prefer to read a magazine in print or digital format, what are they comparing? Some magazine publishers still think that posting a PDF on a website is a “digital edition.” Print will obviously win in that contest.
The answer will also vary by the type of content. People might be happy to read their news magazine on their smart phone on the metro, but read about boats at home in their easy chair.
Don’t get me wrong, there’s a place for inferring best practices from averages, and strategies from survey results. We’d be foolish to ignore those things. But you have to make sure the person has done their homework.
When you see these sorts of things, you need to wonder if the average is across a data set that combines disparate things. Maybe there are several different averages hiding in the data. You need to learn to ask the right questions.
Along those lines, Andrew’s article does break down the stats a bit to try to make some sense of the question of the ideal length of an article.
So when you read these industry articles, or listen to the keynote at the marketing conference, keep that skepticism in the back of your mind. Are they accounting for the possibility that the data is a lot more clumpy and complicated than they think, or are they taking a simple and useless average across a complicated landscape?
Sources
https://whatsnewinpublishing.com/publishers-what-is-the-ideal-word-count-for-articles-and-other-content/
4
views
Thoughts on the Apple Vision Pro
We have another “this changes everything” technology to deal with. At some point people will get over that, but it hasn’t happened yet.
The latest thing is the Apple Vision Pro. I haven’t used it. My comments are based on what I’ve been able to see online.
The new device is like a virtual reality headset, but it’s more “augmented reality.” Apple calls it “spatial computing.” It allows you to create a “desktop” out in space, rather than on a dedicated, dimensionally limited monitor.
My first reaction is that nobody is going to want to wear that thing for very long. It’s heavy, and while I’m sure they try to make it comfortable, it’s just unnatural.
You’re not going to wear this thing out in the world.
For starters, the battery only lasts for two hours. Along those lines, I hope they have the sense to allow you to have multiple batteries.
It’s very operating system centric. It doesn’t seem to have a generic focus on augmenting the experience of everyday life.
Wearing one of these things is essentially the same as pointing a camera at everyone. Humans aren’t used to the idea of being filmed all the time – whether they are or aren’t.
It’s the sort of thing you wear on your couch in your living room.
My biggest question about the device was whether it has a new mode of input. It does, to some extent. It tracks your eye movements, so you select things simply by looking at them and pinching your fingers. You can input text by speaking, but that’s not really the greatest. Imagine being on a train where half the people are talking to their headsets. Not practical.
It can integrate with a keyboard, which is good, because as of this moment, nothing beats a keyboard.
There are no controllers, which tells me that Apple is still caught up in the arrogance of designing an “intuitive” interface. “You will use this device the way we say,” which is … arrogant, as I said, but also just plain stupid. People are different, and they need to be able to customize things to work best for them.
“Wait a minute, I have to put on my headset.”
Apple Vision Pro is an interesting device, but as I foreshadowed in my book, The Intruder, the real change is going to be when the tech is in your eye, and you can input data without speaking. That’s where it’s all going. The Apple Vision Pro is a decent step towards that goal, but … I’m not sure I’d use the thing even if they gave me one for free.
But this podcast is for publishers, so what should publishers make of this.
First, you’re going to be bombarded with nonsense about how this changes everything, and you need to start doing special content and marketing for the Vision Pro. Sure. Just like we had to do special stuff for Pokemon Go, right?
Ignore all that.
Second, this will mostly impact movies, online gaming, and sporting events. There are some really cool applications for that, and if you’re in one of those spaces, you need to start thinking.
This isn’t going to affect newspapers, magazines, books, B2B products, etc. If you’re in one of those spaces, my advice is to demote anybody who recommends a new strategy based on the Apple Vision Pro.
Resources
========
This is a good video on the Apple Vision Pro.
• Apple Vision Pro ...
My book on the future of enhanced reality.
https://www.amazon.com/Intruder-Greg-Krehbiel-ebook/dp/B00A7BRZZU
13
views
An introduction to cognitive biases
Our minds are very complex, and our choices are often influenced by strange factors. Today I want to touch very briefly on the psychology of persuasion and focus particularly on cognitive biases. I can’t do justice to the topic in three minutes, but I’d like to pique your interest by mentioning a few of the factors that publishers should consider in their marketing and sales efforts.
Cognitive biases are mental shortcuts that we take to help us process the world more quickly. Evolutionary psychology might suggest that cognitive biases are generally useful – otherwise, why would we have them – but they can also lead to errors in judgment.
One of the more common is confirmation bias, which is the tendency to latch on to information that confirms our existing beliefs and to ignore or downplay contradictory evidence.
I’m not suggesting that you use confirmation bias, but that you be aware of it. For example, my friend Del sells local honey. People believe that local honey helps with allergies. Del has looked into this and says it’s not true – no matter how much he’d like it to be true. The funny thing is that it’s hard for him to convince people out of that belief, even though it would help him to promote it.
Anchoring bias is our tendency to rely too heavily on the first piece of information we receive. This is particularly relevant to headline writing. Think of all the times you read a headline, then read the story, and realized the story was deceptive. It’s also relevant in pricing. By presenting a higher price first, you establish a perception of value, and then when you present the lower price, people feel like they’re getting a deal.
Fear of loss is a very powerful tool in the marketers toolkit. People are more afraid of losing something than they’re motivated to gain something. This applies to messages like “Don’t miss out, only two days left.”
Scarcity bias is similar to fear of loss. It’s the tendency to place a higher value on something that is perceived as scarce or limited in quantity. Marketers can use scarcity tactics by highlighting limited-time offers, exclusive deals, or limited stock availability to create a sense of urgency and drive consumer action.
The bandwagon effect is the tendency to believe what everybody else believes. Marketers take advantage of the bandwagon effect with social proof, like customer testimonials, user-generated content, or even just claiming that it’s the popular choice.
The framing effect refers to how you present data. Is the meat 80% lean or 20% fat? Is $100 the full retail price of the jacket, or is it 50% off the full retail price of $200?
Choice overload – or choice paralysis – happens when there are too many options. We usually think it’s good to have options, but not always, because it’s work to figure out which is the right one?
I remember once when I was on my way to a night class, and I was dead tired. I wanted a cup of coffee and a candy bar. I stopped at a little roadside market that had 25 varieties of coffee and 600,000 candy bars. It was the only time in my life I wished I was in the Soviet Union, and my only choice was “coffee” and “candy bar.”
2 more.
The halo effect is when you extend one positive feature to the whole product, or brand. The attractive salesman must be a good person and wouldn’t lie to you, and the taller candidate must be the better leader.
The priming effect is when one stimulus influences another decision. You play Italian music and more people at your restaurant buy Italian wine, or you show happy people using your product, and people associate the product with being happy.
Obviously there are ethical implications to using these biases to manipulate consumer behavior. In my opinion, as long as you’re selling a good product, and you’re not lying, you can use these techniques without being a rotten bastard. But you’re going to have to figure that out on your own.
18
views
How to use ChatGPT for content creation
There was some conversation at the Renewd Summit earlier this week about how to use artificial intelligence in content generation. Today I’m going to give you six ideas to consider to incorporate AI-powered writing into your workflow.
1. Get new ideas. Let’s say you write a newsletter about electric power transmission, and you’re trying to fill out your editorial calendar. You can ask ChatGPT, “How are advancements in renewable energy impacting electric power transmission systems?”
If you’re not sure what to ask, ask ChatGPT what to ask. You can say, “I write a newsletter on electric power generation and I need new ideas for articles. What prompts should I use?”
It really is a chat bot. It’s very different from Midjourney, where you have to construct the perfect prompt, and it doesn’t remember from one request to the next what you asked.
2. Get keyword suggestions. Ask ChatGPT something like this: “I want to write an article about how advancements in renewable energy are impacting the electric power transmission system. What keywords should I target in the article?”
3. Create a first draft. Once you have an idea for an article, ask ChatGPT to write it for you. You can give it a target word count, and you can even ask it to write in your style, provided you first upload some content so it knows what your style is. You can also ask it to write for a particular audience, or age group.
But don’t use ChatGPT’s text, and don’t believe any of its alleged data. Be sure to fact check it, because it does hallucinate sometimes. It’s funny that we talk about computers hallucinating, but … that’s the crazy world we’re living in now.
4. Check your word choices. Once you have your article the way you want it, ask ChatGPT to make suggestions about how to make it more valuable for your target audience. For example, upload the article to ChatGPT and ask “how can I make this article more appropriate and useful for people in the electric power transmission industry?”
5. Summarize or expand. Once you have your final article, use ChatGPT to write a brief summary, or ask it to expand the article into a longer, more details version. You can also ask it to write a Tweet or other social media post to promote the article.
If you’re really ambitious, you can (6) use Open AI’s application interface to integrate ChatGPT into your content management system.
13
views
Is the word "webinar" holding you back?
Yesterday I attended the Renewd Summit in Arlington. Renewd.net is the free open community for professionals dedicated to sharing best-in-class practices for increasing subscriptions, building communities, and enhancing live and virtual events. In a way it’s the modern manifestation of the newsletter publishers association.
In one of the first talks, Diane Schwartz mentioned that how you name something can have a big effect on the perception of an event.
For example, is it a webinar or a seminar?
Is it a summit or a meeting?
Is it a trade show or a conference?
How you label your event can be very powerful.
Let’s start with formality. Some words – like conference, summit, or symposium – might convey a sense formality, professionalism, and importance, while other words – like workshop, training, or meeting – suggest a more interactive, hands-on approach.
Then we have audience expectations. A seminar or workshop might attract people who are seeking practical skills or knowledge, while a symposium or conference might seem more academic.
You also need to consider the emotional appeal of the words. “Summit” or “forum” may evoke a sense of collaboration or leadership, while colloquium or roundtable implies more intimate and intellectually stimulating discussion. I also wonder if words like “workshop” might alienate introverts, who might be afraid they’re going to be called on to participate. Some people don’t like that.
The words you choose can also affect perceived value. “Summit” or “symposium” might imply higher value or prestige while “meeting” or “training” sound more routine.
The name you choose can affect your position in the market. It’s probably a good idea to do some testing to see which words work best for your event for your audience. Don’t just stick with something boring like “webinar.” I’ve read a couple articles suggesting that the word “webinar” is a turnoff and depresses response.
Here are some other words to think about.
Panel discussion
Think tank
Forum
Lecture series
Retreat
Meet-up
Boot camp
Seminar series
There are a lot of words out there. Before planning your next event, you might consider doing some research. A thesaurus. ChatGPT. A keyword tool. “Answer the public,” etc. And when you roll out the idea, do some testing.
7
views
Composable vs. packages customer data platforms -- what's the difference?
As if the CDP landscape wasn’t complicated enough, now we have the additional complication of composable vs. integrated, or packaged CDPs.
To unpack this, let’s start with composability. From Wikipedia …
Composability is a system design principle that deals with the inter-relationships of components. A highly composable system provides components that can be selected and assembled in various combinations to satisfy specific user requirements. In information systems, the essential features that make a component composable are that it be:
self-contained (modular): it can be deployed independently – it may cooperate with other components, but dependent components are replaceable
Stateless – which means it treats each request as an independent transaction
Although not everybody agrees with the stateless requirement.
Proponents of the “composable” CDP [I’m citing an article that I’ll link below] see the CDP itself as a software platform that could be broken up into components. Their argument is that companies should not buy an all-in-one CDP to handle everything from data collection, to integration, to profile unification, but rather purchase separate components for each function.
It’s not clear to me that a “composable” CDP is actually a CDP. One of the main distinguishing marks of a CDP is that it creates a comprehensive view of each customer. But in a composable environment, that may be happening in other places, like in the data warehouse.
The question, of course, is not whether a composable system meets somebody’s definition of a CDP. The question is whether it will work for your use cases and your staff. I can help you figure that out.
Here are five things to consider about composable CDPs.
1. The Swiss army knife question. None of the tools on a Swiss army knife are best of breed. They’re basically adequate. But the point is that if you have a Swiss army knife, you have all those tools at your disposal. When you evaluate CDPs, keep that analogy in mind.
2. Tied to that is build vs buy. A composable CDP means you have to select all the components. Do you want to do that?
3. Closely related to build vs. buy is how many integrations do you want to make? Integrations can be tough to build and maintain. They break. Do you have a team to keep them going?
4. Do you have a data lake, or a data warehouse? If so, you might think it’s duplicative to also dump everything into the CDP.
5. Are you a data-driven company, and do you have data engineers on staff? If so, the composable option might be good. If you think of “CDP” as a function rather than as a platform, why does the data have to live in any given bit of software?
That’s a very short introduction to the subject. If you want to know more, read the first link below, or give me a call and we can chat about it.
Sources
Composable CDP Vs. Integrated CDP: What’s The Difference?
https://cdp.com/articles/composable-cdp-vs-integrated-cdp/
What is a CDP?
https://www.cdpinstitute.org/learning-center/what-is-a-cdp/
Composability
https://en.wikipedia.org/wiki/Composability
116
views
The June edition of The Krehbiel Letter
In today's podcast I give a brief overview of my latest letter, which discusses ...
Is Google in danger of being replaced? It sounds silly, but it's a real possibility.
What comes after the webpage? The web page is not as significant as it used to be. Why?
Marketing to different generations.
Rethinking Micropayments.
And more.
You can download the letter today at this address.
https://krehbielgroup.com/the-krehbiel-letter/
2
views
The challenge and dangers of redesigns
Bo Sacks distributed an article the other day about the need to refresh your design if it hasn't been updated in a while. The article is mostly about print, but the same can apply to a website.
Designers amaze me. They can tweak one little thing — a color here, a font there — and make something look so much better.
I don’t have that talent. If men still wore ties to work, I’d want Jeeves to help me choose mine.
Looking modern is often a good idea. I remember evaluating some vendor software that looked like 1990 – it was 2010 at the time – and fighting the impression that it was old technology simply because of the interface.
But as I mentioned before, when someone makes a claim, I instinctively think about things that might disconfirm that claim. It’s just the way my brain works.
Following that thought, are there any negatives, or dangers in updating your design? Yes, actually.
The first is simply resistance to change. Change can be uncomfortable, and a sudden change can be disorienting. That might argue for slow, incremental changes rather than wholesale changes.
The second is you might be in a niche where old-fashioned is part of your brand. Looking too modern might make people think you’ve abandoned your traditional, long-time values.
The third is that a prettier, more modern design doesn’t necessarily mean a better user experience.
When I started out in publishing, I saw a box of direct mail pieces sitting outside the marketing director’s office. I picked one up to take a look. I knew almost nothing about publishing back then, and even though I was in editorial, I was curious about everything, including marketing.
I’d never seen an uglier marketing piece. The colors and the design were awful (not that I’m an expert on either of those things, by the way), and the pictures and the type face looked like they were from 1950.
When I asked the marketing director why we sent such an ugly thing, she said she agreed that it was ugly, and it was personally embarrassing, but it out-performed every attempt to replace it. They’d do split tests of the ugly mailer against nicer designs, and the ugly one always won.
I mentioned this to my friend Leslie Laredo, and she asked if there could be some confusion here over correlation and cause. For example, it would be wrong to conclude “ugly = better.” In fact, it would be wrong to conclude that we knew why the ugly brochure worked better. But A/B tests are fairly reliable, and in this case, the uglier design always prevailed. The why was unclear, but the fact was pretty well established.
Back to the article. Peter Sands, who is making the case for occasional redesigns recognizes what I’m talking about. It’s not enough for the designer to like the new design. The readers have to like it.
So whenever you do a redesign, you have to have a method in place to test it and make sure it’s helping you. The bare fact that it looks nicer isn’t enough. It has to sell more, keep people on the page longer, get visitors to read more articles, reduce customer service calls, or just make people feel better about it. But you need some metrics.
Sources
=======
Does your newspaper look tired and out-of-date?
https://www.inpublishing.co.uk/articles/does-your-newspaper-look-tired-and-out-of-date-21830
11
views
Is the "curse of knoweldge" holding you back?
My friend Ralph had a job at a fast-food restaurant in high school and he complained about customers who would ask for a complimentary cup, despite the big sign on the wall that said “no complimentary cups.”
The thing is, the sign is very obvious if you’ve worked there 15 hours a week for three months. It’s not as obvious if it’s your first visit. You can’t expect people to read every sign everywhere they go.
That’s an illustration of something called “curse of knowledge,” which is a cognitive bias that occurs when you understand something and assume everyone else understands it too.
It’s something to watch out for in
* customer service,
* website design,
* project management,
* sales …
it creeps up all over the place.
Remind your customer-facing staff, like my friend Ralph, that they have special knowledge and familiarity that your customers don’t share.
When designing a website, you and the programmers have been poring over it for months, and you know what everything does and why. But your visitors don’t. So here are some ways to combat the curse of knowledge.
* Where possible, use the standard features everyone is used to. Or in other words, when in doubt, copy Amazon.
* Before you roll out a design change or a new feature, have several people who know nothing about the project review it.
* Have Apple users test your Android instructions and vice versa.
* Create reporting mechanisms to track problems. If you expect people to follow a certain path, check to see if they’re behaving the way you expected.
* Make it easy for customers to report when they’re confused or frustrated.
* Reinforce the attitude that if a customer misunderstands, you have not communicated well enough.
The curse can get in project management as well, and I’m afraid I can be guilty of this one.
“The updates are always in that spreadsheet I shared with you during the kickoff meeting. Didn’t you bookmark it?”
Just because you said something doesn’t mean people heard you, understood you or took action based on what you said.
In this context, the curse of knowledge can also go the other way, where the technical people are aware of a problem, but the project manager is out of the loop. They assume he knows.
Here are a few general things you can do to combat the curse of knowledge.
* Overcommunicate. Yes, it’s annoying, but it’s better than miscommunication.
* Avoid buzzwords and industry lingo. People might not know what you mean, and they don’t want to look stupid by asking for clarification.
Hint – always ask for the clarification. As Lincoln almost said, it’s better to be thought a fool now than to remain ignorant and prove yourself a fool later.
Here's a longer treatment of the subject.
https://martech.org/how-the-curse-of-knowledge-may-be-hurting-your-business-and-what-to-do-about-it/
8
views
Are micropayments the right solution for publishers?
Micropayments seem to be the idea that won’t catch fire but also the idea that won’t die.
It seems that publishers are starting to take another look at micropayments because ad revenues are declining, and subscription revenues aren’t always making up the difference. Are micropayments a desperation move, or do they belong as part of the paid content ecosystem?
Advocates of micropayments will say that readers will only subscribe to a couple brands, but are willing to pay for content from many more. To put it another way, lovers like subscriptions. Daters want micropayments.
Right now, publishers tend to have some mix of these three options. Free access, access after registration, or access after a subscription. Giving readers the ability to purchase single articles creates a new revenue stream and a new way for readers to connect with a brand.
Generally speaking, I’m a big skeptic of this “connect with the brand” talk. More often than not that’s a marketing delusion. I have zero brand loyalty to the shoes I wear, for example. But in terms of publishing, think of it this way.
First, think in terms of source. There are sources that the reader trusts, and sources the reader doesn’t trust. We can break down the sources the reader trusts into two categories. The smaller category is sources the reader is willing to subscribe to. The larger category is sources the reader trusts, and might consider paying for, but won’t subscribe to.
Second, think in terms of types of content. There’s content the reader will pay for, and content the reader will never pay for, because it’s easy to find for free elsewhere. Once again, we can break down the content the reader will pay for into content the reader will subscribe to, and content the reader might consider paying for, but won’t subscribe to.
It’s obvious where micropayments fit in here, specifically, sources and content the reader trusts and is willing to pay for, but won’t subscribe to. At least not now.
Here’s the problem. If this is such an obvious and logical division – as it seems to be – why haven’t micropayments caught on?
I think there are a few reasons.
From the reader’s perspective …
Any sort of regwall or paywall is an irritation. Unfortunately, we’ve trained a generation of readers to expect to get content for free. The micropayment solution would have to be as easy or easier than a regwall or paywall.
Nobody wants to set up yet another account with somebody – who’s going to spam you forever and ever, amen – just to pay $2 to get a lousy article. So the solution would have to be able to use a payment option the reader has already created, like Apple Pay, Google Pay, Paypal, etc. Also, it might be a good idea to have a method to give away an article or two when someone first signs up, just to sweeten the deal.
On the publisher side …
It seems like a lot of work. Somebody has to pick which articles are for sale, set prices, assign SKUs, and then report on all that. Is it really worth the effort? An effective micropayment strategy would have to make that process very simple and fit into the publisher’s existing data scheme.
The publisher might believe micropayments will detract from subscription sales, but we’ve already said that micropayments address the audience of people who are willing to pay, and trust the content, but aren’t yet willing to subscribe. So if it’s done right, it opens doors rather than closing them.
The publisher needs to get the reader’s information. Getting a buck for an article is okay. The real money is in the relationship.
There are a lot of other things to consider here. For example, might different age groups have different attitudes to micropayments?
But I think I’m overtime on today’s podcast, so I’ll leave it there. But if this is an issue that interests you, I’ve done a fair amount of thinking about it, and I’d be happy to talk it through with you.
11
views
The death of 3rd party cookies and tracking
Have you heard of “intent-based marketing”? According to an article by Sara Fischer, It’s Dotdash Meredith’s response to the alleged death of the 3rd party cookie.
To understand how this works, let’s start with how things used to work.
You visit a website, say 123.com, and 123.com participates in an ad network, called ads4everybody.com. If you’ve ever been on any site that participates in the ads4everybody network, you would already have their cookie in your web browser. That’s a 3rd party cookie because you’re not on ads4everybody.com, you’re on 123.com.
Every time you do something on any site that participates in the ads4everybody.com network, it sends more data into your profile. After a while, ads4everybody.com knows – or guesses – a lot about you, so it can target ads to your specific interests.
I said “your profile,” but it’s supposed to be anonymized. Ads4everybody pretends they don’t know it’s you – they just have an ID for you. An anonymous id. Wink wink.
When web browsers block 3rd party cookies, they block the ads4everybody.com cookie, because you’re not on that site. But the web browser can read the 123.com cookie, because you are on that site.
Got it?
So 123.com can put a 1st party cookie in your browser and track what you do on the 123.com site.
Now let’s say 123.com is one of the 40 sites owned by Dotdash Meredith. All 40 of those sites collect information with 1st-party cookies and feed that information into a central database. That central database – with all the information from those 40 sites – can put people in audience segments and make predictions about their interests and likely behavior.
So it’s basically the same thing as an advertising platform using 3rd party cookies, it’s just that it only works on sites owned by Dotdah Meredith.
They claim they’re not using cookies at all. But they’re using the word “cookies” to mean 3rd-party cookies – and you have to watch out for that.
If they didn’t want to use cookies, they could use browser fingerprinting.
Here’s how that works. When you visit a website, your browser makes a request to the server. That request includes a lot of data, like your IP address, the browser you’re using, your operating system, screen resolution, active plugins, language, time zone, etc. All those settings create a fingerprint that’s close enough to being unique.
If all 40 Dotdash Meredith sites tie online behavior to that fingerprint, and then send that fingerprint to their central database, it accomplishes the same thing. They can follow your profile across all Dotdash Meredith sites.
I asked someone at D/Cipher – which is this ad system from DDM – about how this all works, and here’s what she said.
We target content, not people. The output of D/Cipher is recommended site taxonomies which we use for contextual targeting (opposed to 1P or 3P based audience segments).
That makes it sound like a content recommendation engine to me, but I’ll follow up and try to parse that out.
The bottom line, however, is the death of the 3rd party cookie does not mean the death of tracking.
Sources
https://www.axios.com/2023/05/16/dotdash-meredith-debuts-intent-based-ad-targeting-tool
20
views
What comes after the webpage?
The “People vs. Algorithms” podcast has an episode called “The Deterioration of Webpages.” It raises some very interesting questions about the future of media companies. If you’re in the media space, you should listen to it.
Here are my comments and opinions on their analysis. Some of it follows what those guys say, and some of it’s my own.
Why are web pages ending?
First, there’s so much more competition.
* Podcasts
* Youtube
* Twitter
* Tik Tok
* Email
* Apps
Why do you need to go to a web page?
Second, as I’ve mentioned before, the web experience on mobile is terrible because of the ads. As more and more content consumption occurs on mobile, that will take away from the web page.
Those are things that are happening right now, but AI will do at least two things to hurt web pages even more.
1. By simply providing an answer to a question, there will be no need to visit the page.
2. AI will start flooding the internet with content, so most pages will be created by AI.
As Alex said on the podcast, the web will be a bloodbath.
So Alex raises the question: What is the least disruptible by AI?
Unfortunately, whatever that is today, AI will probably disrupt it tomorrow.
Troy said a publisher should ask “How am I optimizing my content for the next generation of search?”
I don’t like that. Google has played far too large of a role in dictating terms to publishers. You must do things this way if you want to be discovered.
Obviously you have to be discovered if you want to have a media business, but to whatever extent they possibly can, publishers should keep that under their own control, and not dance to Google’s tunes.
Troy also said “The center of your media brand has to be the quality of your ideas, not the distribution channel.”
I like that. And my suggestion is to think of the reader first. How can you help the reader?
Think of technology as a way to serve your audience.
You have to make money, so you need to worry about audience and monetization and all that. Fine. But if you put the monetization first, you make a horrible product.
Here are my recommendations.
* Put the content consumer first.
* Start with the problem you’re solving for people, and allow the distribution channel to grow out of that.
* Maybe you’ll need to be on many different channels.
But Brian is right, don’t assume you’ll start with a web page.
References
People vs Algorithms
https://podcasts.apple.com/us/podcast/people-vs-algorithms/id1642958293
The Krehbiel Group
https://krehbielgroup.com/
2
views
5 practical tips for email deliverability
I was getting ready to send the latest issue of The Krehbiel Report, which is my e-newsletter – by the way, if you haven’t signed up, you can sign up for free at krehbielgroup.com – and when I’m getting ready to send a message I like to send it to different accounts to make sure it displays properly. I sent a test message to my protonmail address, and I got this message.
“This email has failed its domain’s authentication requirements. It may be spoofed or improperly forwarded.”
Yikes.
That surprised me, because I had worked with the DNS experts at my hosting company to make sure everything was set up correctly to send email from my account. It turned out there was an additional step I had to take with Mailchimp, and I wouldn’t have known about it unless I had sent the email to protonmail.
Here are five practical steps you can do to check on the status of your email sends.
1. Have a DNS expert review the settings on your domain. I hope you did that a long time ago, but there’s no harm in checking.
2. Look at the reports you get from your ESP. They can give you insights into things that might be wrong.
3. Seed your campaigns with test emails on a bunch of different platforms and make sure they’re all getting into the inbox.
4. Use mail-tester.com. It will generate a test email address for you. Send to that email address, and it will give you a report on what might be wrong with your messages.
5. Check your domain’s reputation at senderscore.org.
Sometimes these services ding you for dumb stuff. It’s the same with Google’s page-speed checker, and similar services.
Don’t get too panicked about every little thing. But where you can, take their suggestions.
For example, mail-tester.com recommended I add alt tags to my images. That’s an easy fix, and it bumps up my rating a touch. So why not?
Remember to sign up for The Krehbiel Letter at
https://krehbielgroup.com/the-krehbiel-letter/
3
views
Can publishers control email delivery?
Yesterday I had the privilege of watching Brian Morrissey and Sean Griffey record the latest rebooting podcast at a MACMA event in Washington. Relevant links below.
Brian mentioned that publishers don’t want to depend on other tech platforms. If you live by the algorithm, you die by the algorithm. And Brian is exactly right about that. Publishers have invested way too much of their time, talent, and energies on platforms that then change the rules and undermine everything they’ve done.
Handing over the distribution of your content to a platform makes you very unstable, Brian said. And he’s right.
Sean said that’s exactly why he likes email, because that’s the only distribution platform you can control. I wrote in my notes, “somewhat.”
Sean is right that you can control email better than many other options, but even there you’re subject to factors beyond your control.
The email service provider the publisher uses can have a significant impact on delivery depending on how well they manage the technical side of the process. A smart publisher will stay on top of this, but it’s not entirely in the publisher’s hands.
Popular email services such as Gmail, Yahoo, or Outlook, have their own spam filters and algorithms that sort emails to decide which ones go into spam, which into promotions, which into the inbox, and which don’t get delivered at all.
There are additional layers of filtering behind that, depending on where you’re sending the email. For example, if you’re sending to an employee at a hospital, the hospital will have filters in place to keep people from being pestered by unwanted emails.
If you’re not careful about these things, you can end up on a blacklist and very little of your email will get through.
Even if you do everything right, and your email gets through to your reader, you also face the prospect of user error. In some email interfaces, the “mark as spam” button is right next to the delete button. Once a user marks you as spam, it can be a challenge to get delivery restarted.
Aside from all that, there are problems with email delivery metrics, which are essential to the business of e-newsletters.
Any email sent to an Apple address might be marked as opened whether or not the recipient ever opens it.
Another interesting problem with email is that when you send to a large organization, there are bots that read the email and click on all the links to make sure they’re safe before delivering the email to the recipient. This seriously interferes with standard email metrics.
And, by the way, that sounds like a great business opportunity for someone. Use AI to filter out clicks from bots.
Anyway, Brian and Sean are both correct that publishers should try to keep their own fate in their own hands as much as possible, and email is the best option for that. But it’s not without its challenges.
Sources
The Rebooting Show
https://www.therebooting.com/podcast
MACMA
https://www.the-macma.org/
3
views
Use artificial intelligence to optimize your paywall
When advertising revenue gets disappointing, publishers often wonder if they can charge for access to their content, and the most common way they do that is with a paywall.
Paywalls raise a lot of interesting questions, and as a consequence, there are a lot of different strategies.
One strategy is to show a little of every article, but require people to register or subscribe to see the full article.
How much of an article do you show? Sometimes that decision is driven by Google. You have to show enough to get the article indexed for the right keywords so people can find it.
Another strategy is to allow some number of free views before presenting the registration wall or the paywall.
How many articles should that be? And how do you decide?
It’s very likely that the optimal number of free views before presenting the paywall varies based on how people find your content. It might also vary by how frequently they come to your site. For example, a person who comes once a week on Sunday to get the crossword puzzle is a different sort of prospect than the visitor who wants to read all your articles about a particular topic.
Yet another strategy is to make some articles free to all, and put premium content behind the paywall.
And there are other variations on these themes.
The problem publishers face is that creating one rule for all these different types of users won’t optimize your paywall as well as creating different rules for different types of visitors.
Maybe the people who come from LinkedIn are most likely to convert after three views while the people who come from a Google search are hardly likely to subscribe at all.
But who has the time for all that analysis, and how would you test it?
That’s where your friendly neighborhood AI comes in.
Rather than sitting around a conference table, debating whether to show 2 or 3 pages before you present the offer, let AI figure it out.
There’s still some work to do. For example, you’re going to have to tell the AI what kinds of things to look for, and you might want to bone up on different types of paywall strategies to decide what options the AI can test.
But this method seems so much better to me than trying to find the one rule to catch everybody. (I wanted to do a play on Tolkien there, but it wasn’t working.)
11
views