I’m late with this year’s predictions post because I’ve discovered that writing while recovering from knee surgery, zonked on painkillers, is a surprisingly difficult endeavor. Onward, through the fog. And apologies for the delay.
I’m always humbled by the act of making predictions. A few months ago, I was in London, sipping champagne at the BAFTA, improbably discussing the mast on Mike Lynch’s superyacht. As I talked and sipped, with my mind already in 2025 predictions mode, I couldn’t help but think: I’ve made a few predictions about Lynch in the past, but how could anyone have predicted this?
Life is indeed stranger than fiction.
On that note, let’s begin our eleventh annual Kellblog predictions post. As always, I’ll review my 2024 predictions (with my generous self-scoring) and then make ten predictions for 2025. This is neither business nor investment advice and this content is provided for information and entertainment purposes only. See my FAQ and T&Cs for disclaimers. See note [1] for my policy on political content.
2024 Predictions Review
1. Election dejection. Hit. The election certainly was distracting. The media generally did emphasize “odds, not stakes” in their coverage. My comment about “testing the once-veiled political neutrality of Silicon Valley,” was the understatement of the year with Elon Musk, A16Z, Sam Altman, the All-In Bros, and several others coming out in direct, vocal and fiscal support of Trump. Andreessen’s take was the most interesting, effectively saying they made me do it, accompanied by explanations (some might say rationalizations) to justify their position as self-declared, single-issue voters. In effect, to modernize an old argument, “what’s good for VC is good for America.” In the end, I suppose it shouldn’t be surprising that when the president puts a For Sale sign on access, that some come forward as interested buyers of power.
2. A slow bounce back in startup land. Hit. I correctly called 2024 as a transition year where the Silicon Valley system would purge itself of recent excesses. ARR growth rates continued to get hammered. ARR multiples hovered around historical means, around half of what they were during ZIRP. See these slides from Aventis.
While I was correct that 2024 would be a tough year, I was over optimistic in thinking we’d turn the corner. I now think the bloodletting will continue in 2025. AI will be a huge driver of the rebound, because of both the large VC investment it attracts and its ability to convert headcount budget into software budget (e.g., AI SDRs).
3. The year of efficient growth. Hit. Efficient growth was the watchword in 2024, with companies delivering increases in both profitability and Rule of 40 scores. Again from Aventis:
Rule of 40 scores increased more moderately than profitability, a reflection of companies’ struggles with cracking the code on efficient growth.
Investors did increasingly look at ARR/head as an overall efficiency measure. Bessemer’s new Rule of X gathered momentum as a key SaaS metric because it better accounts for the ~2.2x greater importance of growth over profit in explaining valuation.
4. AI climbs the hype cycle. Hit. While I’m not sure this requires explanation, I’ll share two observations. First, per Pitchbook, AI grabbed 36% of VC deal value in 2024 on its relentless upward march. Frankly, I’m surprised that figure wasn’t more than 50%.
Second, in the enterprise at least, I think Salesforce made the launch of the year by doing what they do best – neatly packaging industry evolution into a simple three-part message and broadcasting it to the world. I’m not talking about technology innovation; I’m talking about the service they perform for the market by widely broadcasting key positioning messages including (a) we are in the agentic era of AI (the previous two being predictive and generative) and (b) it’s safe for enterprises to get into the AI water.
Great marketers remove fear from the equation in new technology adoption. While profit-motivated in the macro, enterprises are risk-averse in the micro because executives literally bet their hard-earned careers on the success or failure of new technology projects. Credible announcements from enterprise leaders do far more to grease the skids of enterprise adoption than the endless, ever-inflating prognostications from Sam Altman, whose views were summarized by one critic as, “we are now confident that we can spin bullshit at unprecedented levels, and get away with it.”
5. AI-driven GTM efficiency. Hit. There has certainly been an explosion in AI-powered, go-to-market tools from startups. Mega-vendors, keenly aware of the deadly potential of disruptive technology, have not been caught flatfooted, either. (That’s the often-ignored, second-order effect of everyone now having read The Innovator’s Dilemma.) While I don’t think we have yet captured Battery’s 30% increased efficiency target, I believe we will in 2025, particularly for more mature SaaS businesses. New and AI-driven SaaS businesses will likely be investing so much in growth that it will be hard to see those same-store sales productivity increases when they are mixed with the large investments in new capacity (which is a fundamental limitation of the CAC ratio as a growth efficiency metric). But, overall, I think we are well on the way to achieving the GTM productivity improvements promised by AI GTM tools. Keep investing in them and experimenting with them. What’s a competitive advantage one day is often table stakes the next.
6. Beyond search. Hit. To paraphrase REM, it’s the end of the (Internet search) world as we know it. This MIT Technology Review article does a great job of explaining the evolution of search and how conversational interfaces are replacing the search box and generated answers replacing lists of links. This will have a profound impact on businesses that rely on Internet search for traffic, leads, and customers, from publishers to e-commerce providers. And it will impact any business that relies on digital marketing, such as paid search, SEO, or content marketing. So, basically everyone.
Marketers should understand these impacts and get ready for a future of zero-click marketing. While HubSpot’s SEO crash recently made headlines, I agree with Kyle Poyar that they’ve adequately hedged themselves against this. The question is, of course, have you? Lest all this sound too scary, I offer this excerpt from Rand Fishkin.
I believe that this is not an apocalypse for digital marketers.
These are important things that we need to consider, and we need to, as a result, invest in zero click kinds of marketing and change our entire thought process around what we’re doing online with digital marketing. But influence has always been better than traffic. Traffic was always a vanity metric. I love my friend Wil Reynolds who posted this video about showing how their traffic, Seer’s traffic, his company’s traffic had dropped 40%, and it seemed like the end of the world, but sales were up 20% because traffic is not the same as conversions. Traffic is not the same as customers. Traffic isn’t even the same as fans.
So, like REM, I feel fine.
7. From RAGs to riches. Hit. I like RAG because it’s a practical approach that solves or mitigates key problems with LLMs (e.g., hallucinations, explainability, sourcing), all while leveraging their tremendous power. In 2024, I think RAG established itself as a cornerstone technology for enterprise AI. These two posts provide a detailed review of RAG’s progress in 2024. Menlo’s The State of Generative AI in the Enterprise report shows RAG as the dominant and fastest-growing design pattern in enterprise AI.
8. Outbound finds its proper place. Hit. I think that companies, assisted by the rapid adoption of AI SDRs (e.g., Piper), are increasingly figuring out some key truths about SDRs.
- Inbound SDRs are an extension of marketing and, due to their fairly rote work, are increasingly being replaced by AI SDRs.
- Outbound SDRs are an extension of sales and, due to their relatively complex work, are not easily replaceable by AI SDRs.
- Unfocused outbound is generally an unproductive activity. You are better off investing in inbound and partners if you don’t have defined, high-value targets.
- Outbound SDRs are best used as part of targeted account programs, such as ABM, aimed at high-value customers. Think: is the juice worth the squeeze?
9. The reprise of repricing. Partial hit. The best data I’ve found here is in a report from Carta, which suggests that I was a year late: option repricing appears to have peaked in 2023. That said, this chart contains only one quarter of 2024 data. The 2H24 version of this report should be out soon, so we’ll know more in a few weeks. Either way, if your company is still digging out from valuation overhang, it’s never too late to consider repricing. Look at last year’s predictions post for more.
10. Peak podcasting. Miss. Podcasts continued their upward march in 2024. While I’d argued that podcasts would peak in 2024, both market forecasts and industry trends suggest that podcasts will continue to grow in the years to come. The demise of Internet search and the associated need for companies to build their own first-party audiences will drive podcasts to grow in importance. While I’d written that 2024 might be the last good year to start a business podcast, I think 2025 will be a good one as well. So, if you don’t have a business podcast yet, think about starting one. Just make sure you produce good content.
Kellblog Predictions for 2025
1. America gets what we deserve. We voted for it, both via the electoral college and the popular vote, so we’re going to deserve what we get. That will include:
- A more brazen, more conflicted, and less constrained Trump. In short, we’ll see “Trump, Unbridled.” The unlikely bedfellows that elected him will discover exactly what they ordered and exactly who the administration is going to serve. Trump will face less resistance on both the internal front (i.e., intra-party, intra-staff) and external front (i.e. Democratic). Decreased internal resistance will result from fealty-based screening and fear-based leadership, making quick examples of those who step out of line. Decreased external resistance will come from a mix of advance obeying, a sense of futility, and continuous (if incorrect) mandate rhetoric. If the Democrats brought knives to a gunfight last term, this time they’re bringing cupcakes.
- A more divided country. I’d initially thought the more brazen approach would result in buyer’s remorse, but I now think it’s more likely to result in increased division, with supporters doubling down in response to each fresh outrage. Aided by a more fearful and less hostile media, Trump’s apologists may need to contort to new degrees, but they will invariably support virtually anything he says or does. Thus, the country’s divide will widen, with one side believing that we’re making the tough decisions needed to restore America’s greatness and the other thinking we’re destroying many of the things that made America great in the first place.
- A more distracted country. I think of the government like plumbers. I have little interest in what they do and how they do it. I don’t view plumbing as a spectator sport. I just want things to work. But we have now signed up for four more years of stunts, boasts, bluffs, brags, parade jumping (e.g., Stargate), hyperbole, constitutional crises, and trial balloons. Trump is a master at centering attention on himself, has turned shamelessness into a superpower, and paralyzed the traditional media in the process. I’ve always been surprised that we haven’t seen clear opportunity costs associated with all this distraction. In 2025, I think we will.
Get used to hearing “unprecedented” a lot. It, as was once said, will be wild.
2. The broligarchs enjoy their 15 minutes of fame. For some, the agenda was preemptive defense. For others, a desire to deregulate AI, crypto, or big tech M&A. For a few, a chance to grab power and live in the spotlight. For many, the ideological pursuit of sci-fi-inspired visions.
We know who the broligarchs are and why they’re here. A surprising number hail from the PayPal mafia. We know that they’ll all get their 15 minutes of fame. The big question is how long will they last?
Given Trump’s mercurial personality, the revolving door of “best people” in Trump’s inner circle, the sizes of the various egos, and the fact that the broligarchs are all much smarter than Trump, I think the general answer will be: not long.
For every person who hangs on, I think we’ll generate several Rex Tillersons who don’t. As a reminder, while his “fucking moron” quote was never publicly confirmed, here’s what Tillerson did say about his experience:
“It was challenging for me,” he said, “coming from the disciplined, highly process-oriented Exxon Mobil corporation, to go to work for a man who is pretty undisciplined, doesn’t like to read, doesn’t read briefing reports, doesn’t like to get into the details of a lot of things, but rather just kind of says, ‘This is what I believe.’ ”
It won’t be easy for the data-driven tech bros to handle such arbitrary decision-making. But to steal a line from Airplane: “they bought their tickets, they knew what they were getting into. I say let ‘em crash.”
I think the odds of any given individual hanging on will be an inverse function of their desire for power. The more they’re conducting business as usual and simply looking out for their firm’s or industry’s interests (e.g., Cook, Bezos), the longer they should be around. The more they’re trying to work in the inner circle (e.g., Musk, Sacks, Ramaswamy), the shorter.
Heck, Ramaswamy couldn’t even last one Scaramucci before getting blown out. While Musk’s $270M may have bought himself a longer tenure, featuring multiple lives, we’ll see how many times he gets to embarrass Trump before being blown out himself.
3. The startup ecosystem purge continues. As mentioned above, I think the cleanse that started in 2024 will continue into 2025. In many ways, startup investing is like playing craps. You play for a long time, accumulate bets on the table, and either win slowly as different bets pay off at different times – or lose a lot all at once when the shooter rolls a seven. Personally, I’ve never had more angel investments sell, cease operations, or return money than I have had in the past 12 months. There are two opposing forces in play: cash reserves and exit multiples. I think that many startups have strategically decided to sell, but don’t want to start a process in what’s clearly a buyer’s market. Look at these feeble M&A-specific exit multiples from Aventis:
Thus, many startups are tightly managing their cash reserves to buy time and hopefully sell into a better market. I believe that as multiples start to bounce back many of those in waiting will be able to achieve their exits.
4. Attention is the new oil. In 2006, Clive Humby, coined the phrase “data is the new oil,” to suggest that data would power the information economy in much the same way as petroleum powered the industrial economy. Today, I think we can replace “data” with “attention.” In his upcoming book, The Sirens’ Call, Chris Hayes argues that every single aspect of human life is being reoriented around the pursuit of attention. Attention is a kind of resource, he argues, it has value, and if you can seize it, you seize that value.
This harkens back to Jeff Hammerbacher’s 2011 quote, “the best minds of my generation are thinking about how to make people click ads.” Today, you might update that with “click anything,” as best demonstrated by the bizarre game Stimulation Clicker, which ends up part game, part real-life reenactment, and part parable.
We are moving, Hayes argues, from the information age to the attention age. The masters of attention, such as Trump and Musk, already understand this and are leveraging it to their advantage. The rest of us need to learn how to play the game, both on offense and defense. I think that will accelerate in 2025.
5. The world wide web, as we knew it, is dead. Born: 1989. Died: 2024. The original web vision was for an open, world wide network of hyperlinked content, freely accessible to all.
That worked until the information wants to be free crowd got (rightfully) squashed by paywalls to protect creators. Then Web 2.0 came along, creating a read/write web, with user-generated content, so that individuals could not just read, but publish and share content without requiring any technical skills. The mobile explosion extended connectivity but undermined the vision as applications and app stores (with their heavy platform fees) replaced web browsing and websites – resulting in oddities such as the inability to buy an e-book in the iPhone Kindle application.
In recent years, platform providers (e.g., Twitter, LinkedIn) declared war on the hyperlink, unapologetically downranking content that included links beyond their walled gardens. Google’s ever more ambitious front-running (e.g., featured snippets, AI-generated answers) provided the final nail in the coffin, decimating search traffic, and replacing it with the zero-click search.
When was the last time you saw an in-line hyperlink, particularly on a corporate website? Why are newsletters and Substack replacing blogs and WordPress? Why do people bury links in comments and replies? Why can’t WordPress auto-post to Twitter? Why did Seatguru stop updating its content years ago? Why are sites like SlideShare so ad-laden as to become unusable? Why, when I have 20K+ followers, do I have posts that get only 500 views? In a world with algorithm-driven feeds, what does “follow” even mean anymore?
These are the death throes of the world wide web. Why must platforms invariably undergo enshittification? It’s the tragedy of the commons all over again.
While web 3.0 and a Read/Write/Own paradigm is theoretically coming to save us, I’m not holding out much hope. As interesting as some of those ideas are, Web 3.0 strikes me as too much of a hodgepodge of agendas and ideas. I think the current web 3.0 (which is actually web 3.0 v2) has roughly the same odds of success as its predecessor, web 3.0 v1, aka, the semantic web.
So, for now, I think we’ll remain stuck in the Hotel California era of the web: you can check out any time you like, but you can never leave.
6. Working for the algo. You hear a lot of concerns about AI replacing jobs. But I’m also concerned about something else: about us working for algorithms as opposed to algorithms working for us.
What do I mean? Have you ever:
- Taken actions to improve your credit score?
- Put “link in bio” in an Instagram post?
- Followed LinkedIn hygiene tips in an attempt to go viral?
- Avoided certain topics to avoid getting de-boosted?
- Jiggled your mouse a bit while working from home?
- Executed a two-factor authentication (2FA)?
Then congratulations. You’re not working for the man. You’re working for the algo.
CAPTCHA is my favorite perverse example because you have a human trying to prove to a computer that they’re human. Or 2FA, where you have a human trying to prove to a computer that they are who they say you are. Ponder that for a second.
In 2025, I think we’ll increasingly be working for the algo. When I’m performing transactional tasks, I already feel like I’m spending as much time on CAPTCHA and 2FA as on the tasks themselves. And I definitely dislike the menial work I do to tune my content for maximum reach.
Working for the algo isn’t necessarily bad. But it does pose a lot of questions about who is making or tuning it. Bluesky’s custom feeds are one approach to solving one of the many problems here. Passkeys help with security. I’m sure we’ll see other solutions arise as well.
7. The death of SaaS is greatly exaggerated. Satya Nadella made headlines with a three-minute commentary during a recent BG2 podcast appearance which many translated to: SaaS is Dead!
While this is a somewhat fashionable thing to say these days, let’s first look at what Satya actually said:
“Yeah, I mean, it’s a very, very, very important question, the SaaS applications, or biz apps. So let me just speak of our own Dynamics. The approach at least we’re taking is, I think, the notion that business applications exist, that’s probably where they’ll all collapse, right in the agent era, because if you think about it, right, they are essentially CRUD databases with a bunch of business logic. The business logic is all going to these agents, and these agents are going to be multi-repo CRUD, right? So they’re not going to discriminate between what the back end is. They’re going to update multiple databases, and all the logic will be in the AI tier, so to speak. And once the AI tier becomes the place where all the logic is, then people will start replacing the backends, right?”
Translating the surprising amount of technobabble, he’s making an old-age argument that a business application is “just” a UI tied to a database with some business logic, the implication being: how hard can that be? Workday’s $70B, Salesforce’s $330B, and SAP’s $330B market caps all say “pretty hard” to me. Or, if not technically “hard” per se, that there’s nevertheless a lot of value in tying those things together.
Satya builds upon this to say that the business logic can now be handled by agents, again a repackaged argument that once was made about rules engines, business process automation, and low-code development tools and one that trivializes the domain expertise built into business applications. I think the quote says more about Microsoft and their worldview than it does about the future of business applications.
To bring some data to bear here, I found this interesting chart in this excellent deck from Aventis, which asks companies what inning SaaS adoption is in at their firms.
I think the best short answer I’ve seen to this question comes from Jason Lemkin:
To paraphrase Mark Twain, reports of the death of SaaS have been greatly exaggerated.
8. An unlikely revival of branding. In an era of efficient growth and highly scrutinized marketing budgets, it’s surprising to predict a revival of branding. But I think one’s coming because I’m increasingly hearing statements like:
- We’ve spent the past two years optimizing pipeline generation efficiency.
- Now we need to focus on winning more deals.
- SaaS products are increasingly lost in a “sea of sameness,” and we are thus unable to differentiate at a product level.
- Branding is therefore the last bastion of differentiation
- So, we need to win deals based not on product superiority, but on brand value and experience.
In short, since we can’t differentiate our product, we need to differentiate our company.
I have two problems with this logic:
- As someone raised in product marketing: you can always differentiate your product. If you can’t, it’s time to turn in your marketing badge and gun.
- As someone with a child who works in CPG: if my daughter can differentiate fermented milk (i.e., yogurt), then we should darn well be able to differentiate a complicated piece of enterprise software.
But I do understand how a demandgen-oriented CMO – as most are these days – could get caught up in this logic. So, before you embark on a branding program, ask yourself three questions:
- Are you sure you can’t increase your win rate the old-fashioned way — through product marketing and market research (e.g., win/loss analysis, sales enablement, sales training)?
- Are you guilty of Law of the Hammer bias? Is branding the right solution, or are you simply more comfortable working on branding than product marketing?
- Do you have the time and money required to complete a successful branding program? Will your tenure as CMO be long enough to see the fruits of your labor, or will your successor send you a posthumous medal of honor for your contribution?
Whether done for the right or the wrong reasons, I think we’ll see a revival of branding campaigns in 2025. If you’re doing one, make sure you’re in the first group, by ensuring that you’ve exhausted product marketing solutions to the problem.
9. PR is the new SEO. It turns out that one of the best ways to optimize inclusion in ChatGPT results is, per Rand Fishkin, “getting your brand mentioned alongside the right words and phrases in authoritative media.
In other words: PR.
Here’s a link to Rand’s five-minute explainer video.
I guess that if you live long enough, everything comes full circle. This is good news for marketing teams that kept an active PR function and agency during the dark times. It’s bad news for those who turned off PR and will now need to restart from scratch.
There are numerous techniques that marketers must learn to build their LLM optimization skills while still running traditional SEO programs in 2025. Here are a few of the better articles I’ve read on the topic.
- Emily Kramer’s post summarizing an interview with Flow Agency
- FirstPageSage’s guide to ChatGPT optimization
- PenFriend’s in-depth post on LLM optimization
10. LinkedIn enters the social media death cycle. LinkedIn is at a fork in the road. With users fleeing other social networks (e.g., Twitter, Facebook) and trolls, bot-nets, and the like wanting to increase their reach, there is an increasing amount of non-work content on LinkedIn. For example, jokes, memes, snark, and political content. And that’s not to mention the gray zone content where business leaders are making political commentary.
This trend has not gone unnoticed by users, and I think they generally don’t like it.
LinkedIn can go down the usual path to enshittification, relying on engagement as their North Star metric. Because this content is highly engaging, the engagement scores are through the roof: look at the numbers in this screen clip.
The problem is, of course, by allowing and amplifying this highly engaging content, you get more engagement, right up until the point your site becomes a hellscape and nobody wants to use it anymore (e.g., X). Then the hapless platform provider finds that network effects also work in reverse: the more your friends stop using a site, the less incentive you have to go there.
Or they can make the tough decision and focus on their original vision, purpose, and positioning: a social network for work. While they have taken modest steps, such as a feed preference to turn off political content, the features simply don’t work. If they want to preserve their status as the social network for work, they’ll have to do much, much more. And that’s not to mention getting core work social network functionality, such as job seeking, to work properly.
While I think they’re a smart organization, the sirens’ calls of engagement are strong. I’m predicting that in 2025 they only take half-hearted measures to preserve their positioning and thus enter the social media death cycle. Some would argue they’re already in it.
Thank you for reading all the way through. I hope you’ve enjoyed this post, and I wish you a happy and healthy 2025.
# # #
Notes
[1] As described in my FAQ, I generally avoid political content on Kellblog unless I’m using politics to illustrate strategy and messaging. My annual predictions post is an exception because I like to start at the macro level and, particularly in recent times, that will likely involve some intersection of business and politics. Instead of attempting the impossible (i.e., pretending to be neutral), I will allow my views to leak out in the process but, rest assured, I’m not trying to change your mind about anything. Note that I’ll delete comments that try to engage in political conversation as opposed to comments about the predictions themselves. If you are interested in my broader views, you can follow me on Bluesky where I post on a broad range of topics (largely just sharing and commenting on what I read) including SaaS, VC, Silicon Valley, strategy, politics, current affairs, France, the Grateful Dead, databases, humor, and others.
