2025 is coming fast, and AI is no longer just an exciting idea but is slowly becoming the foundation for businesses’ operations and growth. These AI predictions for 2025 are based on my experience working across digital marketing, product management, and customer experience for over a decade.
What gets me most excited, though, is how AI is becoming less about the shiny buzzwords (like LLMs this, GenAI that) and more about actually solving real problems. Whether redefining how search intent connects to landing pages or figuring out how to make personalization feel less creepy and more helpful, AI is proving itself as the ultimate toolbox for businesses willing to invest in doing it right.
Together with my team at Monks (huge shouts to Krasimir Bambalov), we’ve helped global brands like Starbucks, finance apps, and other large organizations use AI to understand customers, optimize experiences, and drive data-led growth. Whether analyzing search behavior, integrating real-time personalization, or experimenting with smaller language models (SLMs), I’ve seen what works and what doesn’t. And now, I want to share my thoughts with our amazing community.
So, let’s dive into the five things I think will shape how we all use AI in 2025—and how you can start thinking about them now.
Table of Contents
1. AI-Driven Customer Insights Will Become Table Stakes
Here’s the thing: if you’re still debating whether to use AI, you’re already behind.
By 2025, AI will be the bare minimum for making sense of the overwhelming amount of customer data businesses collect. The focus won’t be on if you use AI but how well you use it to drive decisions and improve experiences.
Why This Matters:
- Unified Data Models, anyone? Silos are out. Businesses need to unify their data so every team—marketing, product, CX—can make decisions based on the same information.
- You really need to be there in real-time. Adjusting strategy every quarter is no longer enough. AI will let businesses respond to real-time customer feedback, behavior, and market shifts.
- AI tools are becoming more user-friendly, empowering marketers, CX professionals, and product managers to make data-informed decisions faster. However, this doesn’t mean the role of data science is diminishing—if anything, data scientists are becoming even more critical as they validate, refine, and guide how these tools are applied to ensure accuracy and strategic alignment.
My Perspective:
I’ve seen first-hand how well AI models perform sentiment analysis and query intent modeling, which helps completely transform decision-making. For example, when working on Starbucks with Krasimir Bambalov, we used AI to analyze customer feedback across regions (using Natural Language Processing), which helped shape a product optimization roadmap.
What You Can Do:
- Start Small: Use AI to tackle one specific pain point—like churn prediction or customer feedback analysis. Focus on tools that analyze your data securely and meaningfully.
Let’s pause for a moment (LOL): when I say “AI,” I’m not talking about plugging your sensitive data sources into public Large Language Models like ChatGPT. That’s a shortcut, not a strategy. I’m talking about leveraging AI solutions purpose-built for your business, designed to work securely with your proprietary data.
This means investing in tools like sentiment analysis platforms, predictive analytics engines, or customer feedback processors that integrate directly with your systems. AI isn’t just generative—it’s analytical, predictive, and transformative when applied with the proper infrastructure.
Reminds me of what my friend Eddie Aguillar would say:
I’m also not talking about “using Zapier” to automate two processes and calling it “AI”.
- Incorporate customer voice data into your workflows to create actionable insights and iterative improvements.
- Work with data, CX, and product teams to ensure AI applications align with real business goals, not just buzzwords. Don’t know how? Start here.
Have you already implemented AI in your CX practice? read more about how you can show the ROI it brings to the wider business.
2. Smaller Language Models (SLMs) Will Gain Traction
We’ve all heard the hype around big AI models like ChatGPT or Gemini, and it’s for ALL the right (and wrong) reasons. But not every business needs the biggest, flashiest model. By 2025, SMLs will make a significant impact because they’re efficient, affordable, and great for specific tasks.
Why This Matters:
- SLMs are perfect for handling regional languages, cultural nuances, or niche industries where one-size-fits-all doesn’t work.
- These models are cheaper and often more secure, making them perfect for industries like banking or healthcare.
- Specialized AF. Need AI for summarizing documents or analyzing customer feedback? SLMs can do that without the extra bloat of larger models.
My Perspective:
When working with financial clients, I’ve seen how smaller, more targeted AI models make a huge difference. They deliver faster results for tasks like query analysis or sentiment detection without needing massive infrastructure.
In recent years, LLMs have been the stars of natural language processing (NLP). With their massive scale, they’ve been praised for tackling diverse and complex language tasks. However, as Yejin Choi, a leading AI researcher, points out, “the AI industry’s obsession with large language models has overshadowed the potential of smaller language models .”
SLMs are a breath of fresh air—they’re more sustainable, cost-effective, and require fewer computational resources, making them accessible to more businesses. As demand for real-time text analysis grows, this matters a lot. Why burn through energy and budgets when smaller models, fine-tuned with high-quality, specific data, can deliver the same (or better) results for targeted use cases?
It’s time to shift the conversation from model size to what drives success: your data’s quality and relevance. SLMs aren’t just “smaller”—they’re smarter in how they fit into business needs without overcomplicating or overconsuming.
What You Can Do:
- Evaluate Your Needs. If you don’t need a massive LLM, explore cost-efficient SLMs for specific tasks.
- Optimize Resources. Use SLMs for privacy-sensitive or niche projects where data security and performance are critical.
3. Personalization Will Expand Beyond the Web to Real-Time Experiences
Personalization isn’t new, but it’s often shallow—think of those generic “Hello [First Name]” emails. By 2025, personalization will evolve (finally?) into something much more profound: a real-time, omnichannel experience that feels effortless for customers. Whether online, on an app, or in-store, customers don’t see different “channels.” They expect brands to know them, remember them, and deliver what they want seamlessly.
Why This Matters:
- Personalization is more than tracking what customers browse or buy. It’s about connecting behaviors across channels. Relevancy is key. And delivering experiences that are validating, resonating, and helpful will win over anything else you can do from a marketing perspective.
- Imagine your business knowing what a customer might need before they do. Tools like Netflix’s recommendation algorithm or Amazon’s predictive ordering are great examples of this shift. This started a while back, and AI will take it to the next level.
- Real-Time Personalization in Action! Brands are starting to use AI and data to engage customers on the fly. For example, a retailer could send a personalized discount to a shopper’s phone while they are window-shopping in the store.
My Perspective:
When working with retailers, we explored how app data could enrich in-store interactions. For example, if a customer frequently orders cold brews via the app, how can we use that data to personalize their next in-store experience—maybe by recommending a new cold brew flavor or offering an exclusive reward?
This blend of online and offline personalization isn’t a nice to have; it’s becoming a standard. It’s about creating intuitive experiences where customers feel seen without feeling overwhelmed or “watched.”
Examples in the Industry:
- Sephora’s Omnichannel Personalization: Sephora’s Beauty Insider program tracks online and in-store purchases to offer tailored recommendations and promotions. (Source)
- Verizon’s Proactive Personalization: Verizon uses AI to predict customer needs during calls, connecting them with the right agents and offering targeted retention strategies. (Source)
- Nike’s Data-Driven Engagement: Nike leverages app activity and in-store purchases to deliver hyper-personalized product suggestions and exclusive member rewards. (Source)
What You Can Do:
- Data Consolidation! Use the appropriate tools for your business to integrate customer data across your website, app, and offline channels.
- Experiment with Real-Time Offers. Test personalization in a real-time context.
- Leverage AI for Predictions. AI can help analyze behaviors to predict needs. For instance, a subscription service could proactively suggest refills or renewals based on past behavior. (PS: not something new necessarily, just not explored enough)
- Start investing in testing transitions between channels. Make sure customers have a consistent experience no matter where they interact with you—whether that’s browsing online or engaging in-store.
4. Search experience optimization is going to blow TF up! Watch this space.
You can rank at the top of Google all day, but it’s a wasted effort if your landing page doesn’t deliver what people are searching for. By 2025, Search Experience Optimisation will be where all is at. This is my biggest bet for 2025.
Why This Matters:
- Search Intent vs. Page Intent: Krasimir Bambalov and I worked with financial brands, and we’ve seen how often landing pages miss the mark. Users search with a clear intention but bounce if the content or UX on the page doesn’t align. We’ve seen this do massive damage to searchability and discoverability.
- Reducing Dissonance with AI: For a financial client, we used NLP to identify gaps between what users searched for and what landing pages offered. AI revealed where better navigation, clearer CTAs, or more motivating content were needed.
- Calculate to what extent you meet your users’ intent. Krasimir and I created this metric called “Similarity” to measure the relevance between a user’s search query and the content of the landing page that they land on.
What You Can Do:
- Analyze whether your top-performing pages truly meet search intent.
- NLP tools can help map search queries to user-friendly page designs.
- Think Holistically! SEO, CRO, and UX teams must work together for seamless experiences. (this reminds me of this tweet from Jono Alderson, whom I met at SMX London this year, and all I could utter to him was some fangirl stuff lol)
ALSO: It’s time for my DBZ reference!
5. AI Will make CX explode in Highly Regulated Industries
By 2025, AI will reshape customer experience strategies in traditionally regulated industries like banking, telecommunications, and insurance. These sectors, often constrained by compliance and risk, are finding innovative ways to leverage AI to meet customer expectations, streamline operations, and scale personalized services.
PS: In my last 2 years at Monks I’ve been working heavily with banks and financial institutions and GOD, I love the thrill of finding creative ways to use data to experiment.
Why This Matters:
- Customers in banking or telecom expect hyper-relevant interactions, but compliance often slows personalization efforts. Natural Language Processing is helping these sectors analyze customer intent in real-time to deliver tailored yet compliant experiences.
- Efficiency Gains Without Sacrificing Quality. As I have also seen in the State of AI in CX Report for 2024, AI has been critical in reducing manual workloads. For example, Telia reduced contact volumes by 25% while increasing Net Promoter Scores.
- Taking action from data? Finally? At scale? AI enables companies to analyze open-text customer feedback, a historically untapped resource. This leads to insights that directly inform product decisions, service improvements, and operational strategies. MY FAVE use case.
Example in Action:
- Telia: This telecom giant implemented AI-driven solutions to analyze customer feedback, optimize processes, and deliver more efficient service. They achieved a 30% increase in CSAT and higher NPS while embedding AI into their business operations. (Source)
My Perspective:
My work with banks and other highly regulated industries showed me how AI tools like sentiment analysis and predictive modeling transform CX strategies. For example, predictive analytics can help identify customers at risk of churn, while sentiment analysis pinpoints recurring pain points in support interactions. These insights allow businesses to act faster and more effectively, improving retention and customer satisfaction.
What You Can Do:
- ROI of using AI? I wrote about this extensively here.
- Use tools to process and analyze open-ended customer feedback at scale. This can uncover actionable insights you might otherwise miss. I like the one Iqbal Ali is building – check it out here.
Also, I recommend reviewing this resource from Craig Sullivan, Iqbal Ali, Johann van Tonder, Marcella Sullivan – AI Playbook for Research, CRO and Experimentation.
- Use AI to ensure regulatory alignment, such as identifying patterns in customer complaints that align with compliance mandates.
Conclusion: All Roads Lead to Data—The Backbone of AI Success
As we look ahead to 2025, one thing is clear: AI is only as good as the data it’s built on. Whether you’re leveraging predictive analytics to personalize experiences or smaller language models to solve niche problems, it all starts with clean, actionable, and relevant data. The best AI solutions in the world won’t fix bad inputs. Simply put: crap in, crap out.
Here are the key takeaways from these predictions:
- By 2025, AI will no longer be a competitive differentiator—it’ll be a baseline expectation. Companies that fail to integrate AI into decision-making and customer experience strategies risk falling behind.
- Smaller, Smarter Models Will Thrive. It’s not always about size. Smaller language models prove that targeted, efficient AI can deliver incredible results while consuming fewer resources.
- Customers demand relevance, whether they’re online or offline. Unified customer profiles and predictive AI tools will bridge the gap across touchpoints.
- Search Experience Optimization is where it’s at! Ranking is just the first step. AI-powered search experience ensures you align search intent with user action, creating seamless post-click experiences.
- Non-tech industries Will Be the Next AI Powerhouses. Highly regulated industries like banking, telecom, and healthcare are proving that AI isn’t just for tech companies. These sectors harness data and AI to improve efficiency, meet compliance, and scale CX.
To succeed with AI, start with your data. Good AI isn’t magic. So, what’s your next move?
PS: I wouldn’t be able to talk about AI with so much conviction and confidence without the help, support and collaboration of Krasimir Bambalov. One of the best data scientists in the industry that is highly skilled in both LLMs and SMLs, stats, probablistics, ML, NLP and you name it.
I have been fortunate enough to work side by side with him for the last 2 years and we have delivered some insanely innovative and exhaustive work in textual analysis and we are prepping together a science paper to be released this year or early 2025 that explores a comparison of performance between LLMs and SMLs. Thank you Krasimir :).
Connect with Krasimir here: https://www.linkedin.com/in/krasimir-bambalov/