You may have noticed my dear friend as the cover image. That’s because this is also the Halloween Edition. I didn’t want to scare you off, after all, these are my forecasts (okay the ones you get to see) for 2025. I always keep several back, for clients – Jack is one of them, Nosferatu is another. I didn’t want to frighten you either with LMS, LXP, Learning Platform, Employee Development and whatever another vendor wants to call themselves to distinguish from uh, the competition. Which only works if someone types into a search engine with that wordage.
I decided to go with two delightful images to represent two different types of forecasts. One for the good, the other for bad – as in this is a bad idea but it’s coming, and yeah, who thinks this is a great idea? I mean the person who thought Coke Birthday Cake was a great idea – probably oversaw Coke Spice and read a book about Pepsi Clear. All on the trash heap of yuck and waste of money.
Good Forecast Image
For those folks who are wondering who is their future son-in-law – tada!
Bad Forecast Image
For those folks who are complaining about their missing ketchup packet. Darn drive through.
Forecast Track Record
I can’t recall, it’s been more than a decade is over 92% accurate. The only one that caught me off guard was AI last year – but I think a lot of people slide into that category. Get it? GORY. See, fear the Son-in-law to be.
Seriously though, a miss, and yep – that is what happened.
Forecast 1
Your future son-in-law.
AI is coming. I know people are going to be like DUH. Allow me to explain what is happening here and my reference. There are going to be three types of vendors when it comes to AI
a. Those that embrace – All hands on the ship – Move forward, even if we don’t know what we are doing – or the LLM pros and cons (the latter is familiar already)
b. Those that are sort of embracing because a competitor has it, pushes it, and so, they are like those folks who still believe ILT is making a comeback. I am seeing this early stage – and I can’t believe I am saying this – but seriously? When did change because a seven-letter word? Actually two – no.
c. Won’t do it because the client or prospect needs to ask for it. Part of this will slide into the lousy forecast, not that they aren’t, okay, unwilling to jump in, but the whole client/prospect thing – that is about as relevant and useful as polling.
The first group – “embrace”
Breaks into two categories
Knows what they are doing – has a plan, strategy and understands the ins and outs of LLMs, feedback loops, hallucinations and the text aspect, either has a team in place who know AI and Learning/Training – which is what you want OR the head person knows the above.
I’ve talked about the lack of adding a SLM for a mobile or laptop – to run on it, but that’s up to the vendor, even though I believe it would be beneficial in many ways – it’s not like you need to use all the parameters nor max text in a prompt window.
Knows they have or have multiple LLMs – doesn’t know the pros and cons of them, lacks the understanding of the whole text in system aspect around hallucinations, isn’t staying up to date (very common), and uses the whole LLM agnostic – which means nothing, if you do not know what it means to begin with.
This isn’t like plug and play here.
Lastly, there are some very awesome LLMs out there, that I have yet to find one vendor use. Strange.
I digress. The embrace but lack of knowledge is of great concern to me.
Because they are selling you on what it can do, we they themselves do not know the necessities they need in the system (let’s ignore the feedback loop here).
Forecast 2
Be honest, you often wonder if your next-door neighbor turns into this, when they close the door. I’m thinking 75 percent chance, yes.
Common AI Solutions coming to system near you in 2025
- Content Creator Tool – this is that who authoring tool item, which pushes out text, and the solid ones – and trust me, these AI content creators that I have seen and forecasted are not using multimodal, which, uh, numerous LLMs can do. Anyway, this is pure text generation – with an edit option. Ideal for fast push. Quality of content? It is underwhelming unless you are using multimodal (which leads to another fun forecast)
- Assessment/Survey – This will be like butterflies – you will see them everywhere. Will create questions, based on simple text entry, and do various other things. Some offer editing, others do not. You can add them within a course/content or at the end. Will you have a quiz bank with the first version of these rolling out? No. This would require a massive change within the assessment side of the platform.
- Generative text prompt window on the learner side. I already have seen a few out there; again, it pushes out text. And again, if the vendor has a multimodal LLM, they can do far more than text—but that costs the vendor money, and then they have to charge you. The key to pushing out text via prompts and saving token fees is going specific right out of the gate on whatever you are asking or seeking.
- Help Guide—pop up or on the side of the system. Simply type in what you need to know—show me how to find reports—and then it tells you. The plus ones are those that have the text highlighted, so you can go right there. I have already seen one that does a show me, tell me, let me do it, which is the ideal way to train people. Thus, expect more to be coming.
I’m not sold on the whole personal agent’s aspect. However, one vendor has them – I will see them next week- and another says they have a “copilot” aspect, which they do not because copilot isn’t a full throttle of autonomous personal agents that can do far more than copilot.
I surmise the copilot twist/spin you will see more in 2025 with vendors that incorporate it.
Still, I am not seeing a large amount of them.
The skills tied to mapping or skills tied to content, recommendations/playlists is growing, but there is something most people are unaware of.
A lot of it is machine learning – which, while yes, is a form of AI, is not what people think of AI today – the generative aspect.
I expect more skills tied to content/job role angle to grow in 2025. Two vendors say they are working on Skills being mapped to the content without you having to do it, let alone the need for machine learning.
One of the vendors has been playing the shell game – so getting them to show it is more complicated – which is always a red flag. The other is developing it. Oh, the red flag one?
Okay, they admitted it is being developed, which is about as valuable as the infamous tech spin “We have it on our roadmap.” If someone says that it means zip. My roadmap could be for 2045.
AI for reports and metrics. I’m not seeing it. The reports are somewhat doable, but depending on what you want, they take investment and time.
Now, the multimodal, where I can put a PDF book in the window, and it summarizes it, is already doable, yet it hasn’t grabbed hold in the space.
I don’t expect it to. Some folks will do it—even the whole text-to-video angle—but the best ones that can do it are third-party tech, and even then, the output is limited.
Plus, it will take time to generate.
Speaking of generating, today, the speed output for AI with content creation is slow. If you are expecting three seconds, that doesn’t exist.
As AI improves, other items will improve. AI is flea in the stage.
Every LLM on the planet has pros and cons, which is why you need at least two as a vendor.
And even if a vendor says they have guardrails and a R.A.G and it’s accurate, it isn’t.
There is no solution out there because there is no LLM that is 100% accurate. So, if a vendor says it, you can say it’s not true.
Forecast 3
What your boss thinks of you behind closed doors
Financial Struggles (It’s not in a good state)
As noted on my LinkedIn thread, sales in the industry as a whole, are flat. Way too many prospects are in the pipe (pipeline), which means they are not moving forward.
The industry continues to do RIF (Reduction in Force) within their company. There are a lot of companies doing this. I had a meeting with this senior exec at a vendor (not to be named). They noted they would get back to me the next day. Well, the next day, they were let go. And this person was near the top of the food chain.
If sales are flat, then what follows is that the re-organization is always followed with some wordage on how this is going to make the company better. Well, laying off people doesn’t really say that. It says your sales are not meeting expectations for that quarter or year, and thus to RIF, this will offset it.
The industry – way too many vendors – are not generating high numbers – growth wise. Sure, they will pitch 85% or I’ve seen 140% growth compared Y/Y (year to year). Sounds awesome in a press release, but seeing or knowing the actual numbers isn’t published. I mean that growth could be from eight clients to 12.
Which is fine if they are brand new and started in the beginning of the year – it should be higher, but look it takes time to sign and close off on a deal.
Then there are vendors who have a good or even a great year but do to various economic factors beyond their control – in some cases – the numbers drop. Then there are other vendors whose numbers spiked then dropped because of mismanagement (happens a lot), system failure repeatedly – which you would be surprised how many folks stay with a system despite others fleeing – and acquisition that has gone south.
The mismanagement is the biggest problem. There is no doubt in my mind, I could take 100 vendors and find at least three that are being mismanaged. The numbers should be better, the system too, and why isn’t growth continuous on the upswing?
Yes, there are vendors who are doing well in 2024, lots of sales, growth – so this isn’t impacting them as compared to some of others.
Here are key factors in what I have found as a problem in 2024
- Capital infusion, and the PE doesn’t know what they are doing, thus the vendor that gave money too, is having problems themselves, unable to generate enough to offset. If you do not have the cash flow, then those upgrades while impressive may be stuck in the woods.
- Overspending ignoring the signs to pull back. This is a constant in the industry, which is why there are so many vendors who haven’t broken the 10 million USD mark in a given year or has, but when you look at the financial numbers are struggling to turn a profit. I know over vendors who are sales wise near 100M USD, and yet, profit wise, still hasn’t hit it. Sure, they could have incurred a lot of debt from acquisitions, and other areas, but if you aren’t getting the profit here, then RIF is a guarantee, and so is R&D
The overspending is a big problem, and I see nothing to change that. Not even prospects in a pipe that is longer than six to 12 months, over hiring of staff, while at the same time doing RIF – think of the logic there. The marketing aspect is huge. If you are investing a lot of money, say in trade booths/shows, going to a lot of them, making sure you are having prime location with a massive size booth, the outlay isn’t cheap. The growth – from the expectation of closing a lot of deals in a short time is not realistic.
The avg. close time pre-pandemic was 18 months. After pandemic the numbers dropped on avg to 12 months. Now? Back to 18 months or freeze. It is hard to buy a learning system, if your company is scaling back on costs (which a system is), and RIFs in a period of time.
Worse, many of these companies (your targets) are gutting or shuttering Training/L&D (your main buyers), HR, Marketing and some sales.
As a vendor you lack a strategy to handle the gutting or laying off folks in Training or L&D or both at some companies, and thus, the person you are close to signing as a client, just got laid off. If you think they are calling you, forge it.
This is happening not just in small or mid-market, but even F500 – the golden goose every vendor I have ever met wants, even if their focus in mid-market.
What I always find interesting as it relates to marketing in our industry is that you need it – it is big, far bigger than you realize, then you turn around and cut the folks overseeing it, ignoring the impact.
It’s like a company I know who shoved their entire HR team under finance. The logic? Zero.
Anyway, costs, profit issues, and realistic growth – okay, growth the vendor should be at, based on audience type, vertical segment(s), size, should be at X and if the vendor isn’t there – I would have serious concern.
Regardless of capital raised. And number of users – who cares. For all I know that could come from three clients. Sounds great in marketing.
Oops, you just laid them off.
Forecast Four
This is your child when you tell them they only get X number of candies, and then you secretly go and eat a bunch of it.
Upswing in pricing
Remember that issue about financial growth, the number of clients, and RIF?
And something called profit?
Well, a vendor needs the growth and profit piece offsetting the costs.
If the pipe is frozen and sales are flat, a vendor needs to do something—besides lay off people or scale back services or R&D.
How do they do it?
Increase the cost of the system or learning tech or content – which is already overpriced.
The cost of a system—I meant the pricing of a system—is misleading in terms of cost to the vendor.
There are many factors involved here, but from a pricing standpoint, the quote—nobody has a secret formula here.
The “salesperson” who goes rogue and charges what they want isn’t rogue—the head of sales accepts it.
Is it an unacceptable behavior? Absolutely – but you have no idea. Does it happen?
Far more than you realize.
Some vendors quote high, knowing or assuming that you have a procurement team OR that you will negotiate down the price.
I hate this strategy simply because if I see X amount and my budget is Y amount, you will never get into the negotiation round, assuming the system is wanted.
It’s a risky move yet common.
Some vendors are crushing it in terms of profit and growth. How? Dev team costs and blue ocean strategy.
The latter works, the dev team—the team could be from Central or Eastern Europe—yet I talked to a couple of vendors selling low and being profitable and growing strongly (as you would expect if your system is very low compared to the market), and their dev team is in their state or country—say, the UK, the US, or Canada.
Token fees, though, are going to get into this whole quandary, and the industry still hasn’t come up with a plan for charging the customer.
If it stays on the admin side, with content and assessments, the token fees are not going to impact it, and those vendors who have it usually do not charge. Some do.
Token fees for AI, depending on the LLM, but overall are, super tiny, as I noted in the last post.
In 2025, vendors will still not be locked into one pricing strategy.
The ones entering are going—assuming they stay on the admin side, without anything really hitting them as an extensive AI process per se, which will likely eat it.
Once you go to the learner side, and with multimodal, those token fees will go up.
If you have 5,000 people hitting the window, asking questions, adding a PDF for a summary, seeking image to video or text to video, or the system takes your site and produces the code you need (it can already do this), or you want to type in a bunch of text, tie it around an LLM, and then expand it with AI (you can do this with some), that isn’t free.
Think of 10,000, 50,000, or 100,000 people on the learner side. Or how about those customers/clients or your association members. The numbers will change.
As a vendor, you won’t be eating that cost.
Thus, what is your approach?
I see only two clearly, and one I believe will be the more common one—the bundle approach.
You buy a bundle—maybe the vendor calls it a bucket—but you pay basically a bag of credits (which are the tokens), and then when you run out, you just buy another bundle—you as the client.
If the vendor is smart, they will add a virtual wallet (expect to see more vendors do the VW and have their end-users use it to buy courses/content, even external content, on their own).
That part about VW with the end-user angle differs from what I am referencing regarding VW and bundles.
The bundle is the easiest approach and doesn’t require the vendor to face many challenges.
The other option is an add-on cost when buying whatever has the learner side piece.
This option is a poor approach to me. Explaining token fees to someone, unless they understand them and how they are calculated, is similar to someone trying to explain calculus to me when I was 16.
Yeah, exactly.
Forecast Five
Something to remember every time you ignore what is in front of your face
50% of learning system vendors will not have AI by the end of 2025.
Frankly, the number of vendors without AI should be lower, ideally zero, or at most 10%. However, based on current trends, it is projected to be much higher than this.
One of the key factors contributing to this trend is the common theme of prospects or clients not actively seeking AI in their learning systems. This lack of demand is shaping the direction of the market.
The client’s piece bothers me because I know vendors whose clients’ decisions around the vendor’s roadmap exceed 45% (percent). There is a vendor whose roadmap is driven 75% (percent) by the clients.
For a moment, I thought the vendor was the expert.
Guess not.
Anyway, the mantra of this approach, which seems to be ‘if the client wants it, we’ll do it ‘, is bad for the industry. It’s bad for innovation, bad for progress, and bad for the potential of AI in learning systems.
If the clients told you that you need to have emojis on the screen, I guarantee no vendor would do that, even if the vendors they talk to are F50 or Global 100.
I often hear that people seeking a system (related to the vendor themselves) never ask about AI.
Virtual wallets are not a new concept. I know a vendor who had one in 2011. Nobody says, “Hey, what about a virtual wallet!” Yet, they make sense in so many ways.
Technology advances learning, regardless of whether it ends up good, bad, or what they were thinking. I’m not sold on the metaverse in 2025 or the wide usage of VR or MR (aka XR) for learning or training, but I am sold on AI.
Some vendors will be amazing, and some will be awful and just push it in because they have to.
Some clients will see zero benefit from it, but who is to say that the person who made that decision will still be at the company in three years? Three years is something you, as a client, have to consider.
Not now.
Therein lies the problem.
Bottom Line
I never cared for Shakespeare. We (in hs) were forced to watch Romeo and Juliet.
Hamlet bores me. Have you ever seen it live? My forecast says you will fall asleep.
I once dozed at a play in London featuring Sir Patrick Stewart and Sir Ian McKellan.
I awoke at the break after someone tapped me on the knee.
I blamed jet lag.
I predict that at the next seminar, whether on-site or via the web, at least 25% of people will sit in the back, and a quarter of them will not pay attention.
They won’t leave out of guilt.
The person at the seminar thinks they hit a home run.
Everything is great.
The show person thinks the person is great, ignoring the signs.
That says otherwise.
Let’s bring them back!
And the show does.
Because if you can’t trust your audience to decide for you,
Who can you?
This guy?
All images generated by
AI. Excluding your future son-in-law.
E-Learning 24/7
