The December 2024 edition of The Krehbiel Letter
Greg Krehbiel
With a focus on Customer Data Platforms, I help publishers and other businesses optimize their marketing, technology, operations and fulfillment functions. If you have a technology problem, contact me.
Dear Publishing Professional,
In this issue: Review the checklist below before starting your next tech project. Also, we’ve entered the “AI is magic” phase. Beware! Plus revenue diversification, troubleshooting tips, and more.
How to save your tech project from disaster
Here’s a checklist to review before you start a big technology project.
Establish clear goals and don’t stray from them without a very serious reason.
Align your goals with your overall strategy. It’s okay to have skunk works projects that sit outside the fold, to some degree, but new projects need to fit in with your company’s overall strategic direction.
Beware new technology as a sneaky way to introduce cultural change.
Get buy-in from key stakeholders. Anyone who is affected by the new initiative has to believe in the goals and methods, and not grudgingly or with lip service. “He who’s convinced against his will is of the same opinion still.” If you don’t get genuine buy-in, you’ll have resistance later.
Make sure your departments are aligned on the purpose or use of the new technology.
Put someone in charge and give him the authority to do the job. Avoid rule by committee.
Budget for appropriate staff and resources. Total cost of ownership includes internal and external staff and other resources to get the job done.
Pick the right technology. Skimping on the RFP process will come back to bite you.
Don’t underestimate integration challenges. Just because a vendor tells you they have an integration with another service doesn’t mean it’s a good one or easy to implement.
Don’t do too many things at once. There can only be so many top priorities. Get one job done before you start the next.
Now apply this to Customer Data Platforms
As you know, I focus on customer data platforms. Here are some additional rules that apply to CDPs, and to other technologies as well.
Understand what a CDP is and does. Like other technologies, CDPs aren’t magic and they won’t solve your problems unless you know how to use them. For an easy, short introduction, read my new book, “What is a Customer Data Platform? And why should I care?”
Pick the right type of CDP. Make sure they have experience in your industry and with your kind of business, and make sure they can explain clearly how they’ll address your use cases.
Take a hard look at your data. If it’s a mess, you’ll either have to pick a CDP that includes data transformation or pre-processing services, or you’ll have to address that problem with another service before you ingest your data into your CDP.
Define your use cases. List the specific ways you intend to deploy your CDP to reach your strategic goals, and make sure you review this before you sign with a vendor. Your use cases should show a clear path to value that justifies your investment in the technology.
Don’t trust out-of-the-box capabilities. They’ll tell you they can do everything and serve you fries with it. Be skeptical. Make sure they show you exactly how they can meet your needs, and even when they have, expect it to be harder than you think.
Beware integrations. Just because Vendor A has an integration with Vendor B doesn’t mean it’s up to date, works well, or does what you need with your data. (Yes, I’m repeating this.)
Get expert help. You should expect to train your internal staff to do the work … eventually. For the first couple of years, you’re going to need help – whether that’s from the vendor, from their integration partner, or from somebody else.
Remember privacy and compliance. Make sure your CDP project aligns with what you need to do for GDPR, CCPA, etc. Some CDPs can help with this.
Factor in long-term maintenance and growth. CDPs require ongoing management. If the CDP is the central hub for all your customer data, all those connections are going to change as technology evolves on each side. Your business will change over time, and your CDP implementation will have to change with it. It’s not a “set-it-and-forget-it” tool.
Expect costs to increase. However your CDP plans to bill you, those stats will increase over time, and so will your bill. Make reasonable projections and add that to your total cost of ownership.
I know a little about these things, so if you need help, please give me a call.
ChatGPT is amazing, but it has limits
Beware ill-informed ChatGPT solutions.
ChatGPT and other large language models (LLMs) can do some amazing things, which can easily mislead you into believing they can do other things that seem related, but actually aren’t. To avoid being fooled, you need to understand the basics of how an LLM works.
A thread on LinkedIn suggested that you could load a series of email subject lines into ChatGPT and it could tell you which ones were most likely to get an open.
The question betrays a misunderstanding of how an LLM works. LLMs generate text based on patterns in the data they’ve been trained on. An LLM can tell you that one word is similar to some other word, or that one phrase is closely associated with some other phrase. Using that method it can mimic human response pretty well, but it can’t predict human behavior.
ChatGPT has not been trained on email engagement metrics. It can recognize patterns in words, but that’s a very different thing from predicting click behavior, which would require it to have access to real-world email results. It doesn’t.
ChatGPT can suggest or refine subject lines based on theories about persuasive language – according to the data it’s been trained on. Don’t expect it to work magic.
Revenue diversification vs. the hedgehog
Some people say that in order to be stable amidst the ups and downs of fortune you need to have seven independent revenue streams, such as earned income, profit income, interest income, dividend income, rental income, capital gains, and royalty income.
Similar advice is often given to publishers. You need to make money from subscriptions and ads, print and digital, content and data, then add events, licensing, partnerships, and so on. A presenter at OX7 highlighted their success with real estate.
At the same time, books like Good to Great praise “the hedgehog principle,” which is to excel in one core area. You can find that core area by asking …
Some would add, “what are you passionate about?” That reminds me of some common (bad) career advice, “Do what you love and you’ll never work a day in your life.”
No. Do what other people will pay for and learn to be passionate about it.
Diversifying beyond what you’re good at just makes you “a bit less good at lots of things,” according to Louise White.
Publishers can reconcile diversification and the hedgehog principle by ensuring that all revenue streams …
It’s also good to constantly test and reprioritize. Sometimes (probably rarely) it makes sense to go off in a new direction. Nintendo started off making playing cards.
Executive, mentor thyself
How many times have you asked for advice only to hear things that (1) yes, you did need to hear, but (2) you knew already? Next time you’re facing a challenge, step away from your routine and pretend it’s a friend who’s having the trouble, not you. What would you tell him?
One of Jordan Peterson’s 12 rules for life is to treat yourself the way you’d treat someone you’re responsible for. It’s an easy lesson to learn but hard to apply. You’re a pretty smart person. Learn to take your own advice.
A troubleshooting tip from Dear Old Dad
My NASA engineer father always told me that when you’re confronted with a range of possible solutions, do the easy thing first, because if you do the hard thing first, and it turns out the easy thing would have fixed it, you’ll feel very stupid.
I’ve learned to add a few things to Dad’s advice.
The most important is to challenge your assumptions. Part of the challenge in solving a problem is often that you’re assuming the wrong thing. You think the garbage disposal is broken when it’s really the electrical switch.
It can take some mental effort to sort that out. Start by explaining the problem to someone else – or a pet, or even a rubber duck. The act of putting it into words can clarify your thinking, although having a second mind thinking about the problem is good.
Other ideas ….
You get the idea. You have to get your brain out of “I’m too close to it” tunnel vision.
It can also be very effective to take a break. Work the problem, then go do a Sudoku, or play the guitar. Do something that will distract your mind. Your subconscious mind will work the problem in the background. (Is that why Google has ping-pong tables?)
Is LinkedIn cringey?
Elon Musk recently said he immediately loses respect for people who post on LinkedIn. He said it’s cringey, and I wondered why. It turns out some people see LinkedIn as a huge collection of performative BS. Everybody is “super-excited” about this or bragging about some other thing. The excessive exaggeration is common fodder for memes.
What do you think about LinkedIn? I’d love to hear from you. Drop me a line, please.
Sincerely,
Greg Krehbiel
President @ Creative Licensing International | Content Licensing, Copyright Protection
1moGreg Krehbiel, you ask at the end of this issue "Is LinkedIn cringey?" When LinkedIn is about business and all the issues that impact a career, or even a personal story here and there, I find it helpful. Like your newsletters, informative. It is when it veers into extreme politics, as it did in this last election cycle, then it gets cringy for me. I am not speaking for other LI participants, just for me. I would appreciate if all those with extreme political opinions, share them on Bluesky, X (formerly Twitter), TickTock, etc.. In my opinion, those are much better forums for political debate. Thank you for asking.