A memorable brand is one that has a leg up on the competition. Not only have they found a way to differentiate their business model from their peers, they’ve found a way to leave a lasting impression on consumers’ minds.

In this post, we’ll explore three research-backed strategies you can use to create a brand that’s more memorable.

3 Ways to Make Your Brand More Memorable

Turning cold and warm leads into customers takes time. According to Tooltester, it can be anywhere from five to 50 touches/interactions to turn a lead into a customer. Even inactive customers may need multiple touchpoints as well (between one and three).

So you need to make every interaction count.

By making your brand one that’s easier to remember, you won’t have to constantly reestablish who you are with each touchpoint. If you’re looking to try this out, here are some research-based techniques that can help improve memorability:

1. Tell Stories vs. Statistics

There was a study published in 2022 on “Stories, Statistics and Memory.” The aim of the research was to determine what kind of information campaigns tend to stick more effectively in the minds of consumers. Specifically, the researchers looked at storytelling versus the sharing of statistics and facts.

To evaluate the staying power of each type of content, they measured how rapidly the content faded from users’ minds over the course of a day. On average, stories faded by 33%; statistics faded by 73%.

But why does this happen? You might think that learning a single, powerful statistic would be easier to remember than a whole story.

In reality, it’s not the brevity of the information shared that makes it so memorable.

For starters, statistics can be difficult to remember unless you’re actively studying them to be recalled at a later time. We have so many numbers thrown at us in our lives—addresses, phone numbers, dates, metrics, measurements and so on. A statistic has to have a ton of potency and relevancy to be memorable.

Now, not all stories are memorable either. However, stories with unique, distinctive and/or surprising details can improve recall. While users might not remember the entire story, they likely will remember standout details.

As for which details improve memorability, you want to focus on reality versus concepts. When we share statistics, we’re asking someone to envision the concept of a number or percentage.

“82% of children who used our flash cards performed better in math classes.”

It’s hard for our minds to form a clear idea around this. Yes, we get that there’s improvement and it’s impressive that more than four-fifths saw improvements in their studies. But there’s not much else there to work with.

When we share stories with others, we’re painting a mental picture. We describe the place, the people, the problem and resolution, etc. For example:

“My son was failing his third grade math class. He failed test after test, no matter how long he studied at his desk every night. That’s when me and my wife came up with these flashcards. They were specially designed with colors and graphics that were shown to improve memorability. We practiced daily with him for weeks. On that next test, he got a B-minus! He was so proud of himself! Since then, MemoMath Cards have helped 12,000+ parents and their kids make big strides in learning!”

Storytelling is a powerful tool in marketing. With it, we can:

• Show empathy for our users’ pains or obstacles
• Paint a mental picture that depicts how the pain was resolved
• Provide more relatable and realistic details about our solution
• Connect with users on deeper emotional levels

It can be difficult to watch your child struggle in school, especially when you see them putting in the work. This story encapsulates that situation much more effectively than a statistic. Parents who relate to this situation may actually see their child struggling with their studies and coming home crying when they hear the story told.

This isn’t to say that statistics aren’t useful in marketing. If you’re looking to quickly convince someone that your products or services are more effective than the competition’s, a statistic may be able to persuade the prospect’s decision. But if you want a prospect to really remember what it is that differentiates you from the pack, stories are the way to go.

2. Use Frequency to Become a Familiar Face

There’s a phenomenon called the Mere-Exposure Effect. It describes our preference for something because it feels familiar.

How does this play into memory? Well, studies have proven that repeated exposure improves recognition.

Here’s an example of how the Mere-Exposure Effect works:

You’re watching something on YouTube and an ad comes on for AutoZone.

You watch it, but barely. You’ve seen AutoZones as you drive around town. You just never give them much thought.

It seems like every few days, you see or hear another ad for AutoZone—on YouTube, the TV, Instagram and even the radio. Again, it doesn’t really trigger a need or much interest in you. But you’re now joining in and singing the jingle at the end.

“Get in the zone. AutoZone.”

You get into your car a couple months later and the oil light is illuminated on your dashboard. What do you think of? For some people, it means a trip to the autobody shop to get the oil changed. For more DIY types and gearheads, you might prefer to do it yourself. And, realistically speaking, if this ad targeted you, you’d probably fit into that segment. So, you might be thinking about a quick trip to AutoZone for oil.

This is how brands can use the Mere-Exposure Effect to increase the frequency with which target users encounter their brand.

This is why regular marketing in the form of blogging, newsletters and social media is so valuable for brands. Even if prospects aren’t ready to convert today, you’re becoming a more familiar (brand) face with each interaction.

Multi-channel marketing and advertising will help, too. That way, users don’t become blind to your content or ads if they repeatedly encounter them on the same platform. A new setting and different content will allow you to recapture their attention.

Just be careful about overdoing it. Consider running tests to find the best days/times to put your content in front of people. Also test the frequency with which you publish content or show ads. You’ll want to find the sweet spot where they recognize your content without feeling like it’s become intrusive or annoying.

3. Pique Their Curiosity

In a 2014 study published in Cell, they sought to understand why people learn new topics more effectively when it’s something they’re curious about. Using functional magnetic resonance imaging (MRIs), they studied how the brain responded to different types of information.

The researchers compared how much information people would retain and for long when it was intrinsically motivated (i.e., curiosity) versus externally rewarded. In the end, they found that activity in key parts of the brain was enhanced when curiosity was the driving force.

Let’s use the example of reading.

When you were in school, you were assigned countless books to read. You’d have to complete certain chapters each night or week, and then complete assignments or tests related to them.

How many of the assigned books do you still remember to this day? Some of the stories may have resonated with you to the point where you remember them or at least certain details from them. For instance, there was a book called Night by Elie Wiesel that we had to read in high school. Although I enjoyed it at the time, all I can remember about it now is that it was set during the Holocaust.

Compare that to the books you’ve read on your own over the years.

How many of them do you remember? You might not recall all of them or every perfect detail, but I have a feeling they’re much clearer in your mind than what you were assigned to read. For instance, I read Stephen King’s The Stand around the same time as Night. It has more than 1,100 pages and dozens of characters in it. I can still tell you who they were and describe the plot and high points.

So, knowing this, how do we translate this concept to branding and marketing?

If we’re looking at this on a larger scale, you want there to be curiosity around your brand. If you can surprise people, present them with a novel concept or have them longing for answers, you’ll increase the memorability of your brand.

For some brands, this might not be realistic. If your company offers something that many others do (think of like a plumber or an IT services provider) or you don’t want there to be any surprises or shocks when it comes to your brand, then you might not be able to leverage curiosity in that way. But you certainly can with the marketing content or ads that you run.

I’ll give you an example. I follow a lot of local farms in Florida and Georgia on Instagram. I enjoy learning about what they produce and seeing what kinds of events they host. So, this kind of content is nothing new.

But earlier today, I was scrolling through my feed when this post was recommended by @thefarmatokefenokee:

In the video, co-owner Doug Davis answers the question:

“Do I need to work on the farm?”

I was initially confused. This is the first time I’ve encountered this farm on my feed, so I didn’t understand why visitors would have to work there. It definitely piqued my interest. As I watched the video, I realized he’s talking about an opportunity to buy land and a cabin on the farm.

This is what I mean by smaller-scale curiosity. You can use your website or app, social media posts, blog, and even ads to address questions or matters that get users to stop and say, “Wait a minute. What is this about?”

I might not be in the market for homeownership. However, finding this post allowed me to discover that this farm has other appealing opportunities, like visits with the animals, overnight stays and farm-to-table meals. It’ll definitely be a while before I forget about this farm.

If you can get your content in front of targeted prospects and pique their interest, you’ll make a strong impression on them, too.

Wrapping Up

Is memorability the only thing that matters? Consider it from the perspective of the user:

You’re at a party and talking to your brother-in-law and sister about how your tooth aches. They recommend a dentist who happens to be there. He introduces himself, mentions his business name, says his office is downtown and invites you to visit his website and book an appointment.

You walk around the party and you’re hit with pain when you take a sip of your drink. Someone notices and introduces himself as a dentist. He works on Main Street right next to that coffee shop you love. Then he tells you about a woman who came in with persistent toothaches. She didn’t have cavities. The problem turned out to be gum recession. Since her treatment, she’s been pain-free for six months. He gives you his business card and tells you to call him.

Which of these dentists would you end up contacting?

Both were local and seemed professional. The first one was recommended by your brother-in-law and sister, but you can’t remember much about him or even the name of his office. You could always call your sister to get the info, but you can’t stop thinking about the other dentist and the story about the woman with tooth pain. Plus, you have his business card, so you can look him up right now.

Does a personal brand recommendation trump a brand that gives a stronger and longer-lasting impression?

I think it depends on the context. Someone who’s had bad experiences with dentists may prefer the personal recommendation, even if it means jumping through some hoops to get the info. On the other hand, some people may find the second dentist’s proactive and empathetic approach to be more reassuring. But that’s how it goes in marketing. We try different strategies in the hopes of connecting with as many people as possible.

That said, it’s important not to conflate memorability with trustworthiness. There are plenty of brands with memorable slogans, brand packaging and messaging that have a less-than-stellar reputation. So the memory-building strategies above need to be used in combination with ethical brand-building strategies in order for the memorability effect to be worthwhile.

Vector-only retrieval misses exact terms, relationships and use-case-specific context, so multi-layer indexing and per-experience retrieval configuration decide whether agentic RAG stays trustworthy at scale.

Pure vector search starts to bend under that load. It is good at semantic similarity, and that matters. But enterprise AI does not only ask for ideas that sound alike. It asks for the exact clause, the policy version, the affected customer, the downstream product and the paragraph next to the paragraph retrieval found. When one search strategy has to answer all of those questions, quality degrades as content volume and query variety grow.

There is also a shift in how people ask. Users who grew up on keyword search are moving to intent-based AI search: they phrase questions conversationally and expect the system to infer what they mean. That looks like a straight win for semantic vectors, but it widens the gap. The same knowledge base now has to serve a vague natural-language question and an exact-term lookup with equal confidence, and you still have to recognize when a question needs keyword weighting instead. Single-strategy retrieval cannot hold that range.

Where Vector Search Quietly Misses

Vector search maps text into a semantic space, so “heart attack” and “cardiac arrest” can land near each other even when the words differ. For a sales team asking broad content-library questions, that may be enough. “How should we talk about onboarding for mid-market accounts?” is a meaning problem, and semantic vector search handles meaning well.

The miss shows up when the question needs a different kind of match:

• A legal team asks for the indemnification language in the 2024 master agreement. That is an exact-term problem, so full-text retrieval should carry more weight than semantic similarity.
• A retailer asks which products a flagged supplier touches. That is a relationship problem, so a knowledge graph needs to walk supplier, component and SKU relationships.
• A support agent asks why an answer changed after a policy update. That is a context problem, so retrieval may need the adjacent paragraphs, version metadata and source freshness signal, not only the nearest chunk.

These are ordinary enterprise questions, not edge cases. Vector search can still be part of the answer, but it cannot be the only path if the system has to serve legal, sales, operations and customer-facing workflows from the same knowledge base.

Why Better Embeddings Do Not Fix Coverage

The easy answer is to upgrade the embedding model. A better embedding can sharpen semantic neighborhoods, but it still optimizes the strategy vector search was already using.

That does not solve the legal lookup, because the problem is not that the contract clause is semantically misunderstood. The problem is that a unique identifier, date or phrase should rank as itself, not as one more nearby meaning. It does not solve the supplier-chain question either. The missing object is an explicit relationship, not a better sentence embedding.

A single strategy can be high quality and still leave a coverage gap. If the system retrieves the wrong evidence, the model can only write a polished answer from the wrong evidence.

Multi-Layer Indexing Is The Architecture Move

None of this means “vector bad, keyword good.” No single search strategy is enough for every use case. Multi-layer indexing is the move: semantic vectors, full-text or keyword search, metadata filters and knowledge graphs operating over the same governed knowledge layer.

That matters because the right mix changes by experience. Sales may weight semantic vector search heavily. Legal may weight full text and metadata. A retail operations assistant may combine graph traversal with vector search so a messy natural-language question can still land on the right product relationships. The system does not need a separate pipeline for each one. It needs one knowledge layer with retrieval strategies that can be configured per experience.

In an Agentic Knowledge Layer, ingestion, chunking, indexing, ranking and retrieval configuration sit underneath many AI experiences. A strong implementation lets teams tune chunking strategy, ranking logic, search weighting and metadata filtering without reindexing the corpus every time a new use case arrives.

Context expansion is the small example that makes the larger point concrete. If retrieval finds the right paragraph but the answer needs the paragraphs around it, the fix should not be a rebuild. It should be a tunable retrieval parameter for that experience.

Agents Make Retrieval Failures Compound

Retrieval quality matters even more in agentic workflows because the system does not retrieve once. It plans, retrieves, grades what it found, decides whether to loop, retrieves again and then synthesizes across those results.

One weak retrieval decision can become the starting point for the next reasoning step. A vector-only sales assistant might survive that if the task is broad and semantic. A legal assistant that misses an exact clause, or a retailer assistant that misses a supplier relationship, can drift across the whole plan. The agent looks like it reasoned poorly, but the reasoning was built on a retrieval strategy that never had the right evidence shape.

Retrieval strategy is an architectural decision, then, because it sets the evidence boundary for every step above it.

Architecture First, Tuning Knob Second

In Progress Agentic RAG, retrieval can become a tuning knob, but only on one condition: the layer has to expose retrieval strategy as configuration before teams can tune it per use case.

Configurability is what separates a hard-coded vector pipeline from an Agentic Knowledge Layer. Progress describes 30+ retrieval strategies that can be configured per experience, including search weighting, ranking, chunking and metadata filters. The goal is not to run every strategy on every query; it is to stop forcing legal, sales and retail operations through the same retrieval shape.

Configuration only helps if teams can see which configuration is working, and tuning a retrieval strategy blind is guesswork. The loop has to close on evidence. REMi, the platform’s evaluation layer, scores responses continuously, and its Context Relevance metric is the one that speaks to retrieval: it measures whether the chunks the system pulled were actually relevant to the query. When Context Relevance falls for a class of questions, that is the signal to change the retrieval mix for that experience rather than swap the model.

That tuning happens in the admin, not in a rebuild. Teams adjust and compare retrieval strategies in RAG Lab, then use Prompt Lab to replay the same questions and confirm the change held before it ships, comparing prompts and models where that matters. Measuring quality, changing the retrieval strategy and confirming the change all live in the same knowledge layer, so improving a weak experience becomes an adjustment instead of a re-engineering project.

So the real question for a CTO is not “Is vector search good enough?” It is “Can this architecture change retrieval behavior without rebuilding the knowledge base?” If the answer is no, the first retrieval decision becomes the ceiling. If the answer is yes, each AI experience can use the retrieval mix its work actually needs.

What separates a pilot that shines on one polished demo set from an architecture that survives production variety is one capability: reshaping retrieval per experience without a rebuild. To test whether your current RAG layer can make that jump, start a 14-day free trial or book a demo.

FAQ

Is Hybrid Retrieval Always Better Than Vector Search?

No. Hybrid retrieval is better when the use case needs more than semantic similarity. A sales content assistant may work well with vector-heavy retrieval, while legal or operations workflows often need exact terms, metadata and relationships to carry more weight.

What Does Multi-Layer Indexing Change In Practice?

It lets one knowledge layer support different retrieval needs without creating a separate pipeline for every AI experience. The same corpus can support semantic search, full-text lookup, metadata filtering and graph-based relationship queries, with different weights by use case.

Why Does This Matter More For Agents Than Chatbots?

Article Summary

Continuous RAG evaluation turns accuracy from a sampled opinion into scored, traceable evidence for governance and quality teams. Instead of one reviewer’s read on a handful of answers, every response gets scored for context relevance, answer relevance and groundedness, and those scores accumulate into a trend line a quality lead can actually inspect. Accuracy stops being a launch-day impression and becomes a standing control you can point to.

The production question is sharper: can the team prove why an AI answer was trusted, which source supported it and what changed when accuracy drifted? Manual spot-checking rarely leaves that kind of evidence. Continuous evaluation does.

Why Spot-Checking Is Not a Defensible Control

A manual review usually starts with good intent. Someone reads a few answers, decides they sound relevant and signs off. For a pilot, that can feel practical. For production AI, the method leaves too much unmeasured.

Scale and consistency are the immediate problems. A retrieval-augmented generation (RAG) system can drift when source content changes, retrieval ranking shifts or users ask questions outside the original test set. That gap widens at scale: many enterprise AI programs run through one central data or AI team, and no central team can hold a domain-specific quality bar for every part of the business the system serves. When a finance answer and a legal answer do not share the same definition of “good,” hand-checking gives you no consistent way to judge either. One reviewer may value a concise answer, while another expects citations, caveats or a fuller explanation. A handful of sampled answers will not show where drift started or give governance a repeatable measurement.

The deeper problem is evidence. If a reviewer asks what was checked and whether yesterday’s answer still holds today, a spot-check has an awkward answer: someone looked. That is thin paper for an audit file.

Accuracy needs to become a measured signal on every answer, not a sampling exercise at the end of a project.

The Three Scores That Make Accuracy Measurable

The useful question is not simply, “Was the answer good?” That question hides the cause — and at scale, neither hand-checking nor an automated judge answers it on its own. A RAG answer can fail because the system retrieved the wrong material, because the model ignored good material or because the answer sounds right while leaning on facts that were never retrieved.

REMi, the evaluation layer in Progress Agentic RAG that scores every response for relevance, accuracy and groundedness, separates those cases into distinct signals.

• Context Relevance shows whether retrieval brought back context that fit the user’s query.
• Answer Relevance focuses on the response itself: did it answer the question asked?
• With Groundedness, the test is source support — whether the answer can be traced back to the retrieved context.

Those metrics matter because each answers a different governance question. Groundedness is the hallucination check. If an answer says a policy allows something, the evaluation question is not whether the sentence sounds plausible. It is whether the retrieved source material supports that claim.

The review conversation changes. Instead of asking a subject matter expert to reread random answers, a quality lead can inspect the score that moved, the source that was retrieved and the answer that lost support.

How Score Patterns Point to the Failure

The three scores are most useful when read together. A single low score says something went wrong. The pattern tells you where to look.

That is the difference between quality theater and diagnosis. If retrieval missed, you investigate chunking, metadata, ranking or how the query was rewritten. If generation failed, you look at the prompt, model behavior or response policy. If the system grounded itself in the wrong source, you have a retrieval problem wearing a compliance-friendly outfit. The answer looks traceable. The trace still points to the wrong place.

For an AI quality engineer, the value is speed. For a governance lead, the value is defensibility. A finding that says “bad answer” is a complaint. A finding that says “high answer relevance, low groundedness” is a record of the failure mode.

Continuous Monitoring Makes It a Standing Control

One clean evaluation pass at launch does not prove the system will stay accurate. A knowledge base changes. Source documents get replaced. Users stop asking demo questions and start asking the messy ones that matter. The system that looked healthy at rollout can drift without anyone touching the model.

That is why continuous scoring matters. Progress documents REMi performance views that track quality over rolling seven-day and 30-day windows. Those windows turn isolated judgments into a trend line. If Context Relevance drops after a content sync, the quality review can start with retrieval evidence instead of a generic debate about whether AI is “getting worse.”

The same loop can expose knowledge gaps. Unanswered-question tracking shows where users asked for information the system could not retrieve or support. That gives the content owner a remediation backlog: add missing documentation, improve indexing, review the next score window and verify whether the gap closed without rebuilding the whole pipeline.

The audit record should connect the source update, the sync event, the retrieved context, the generated answer, the REMi scores and the reviewer decision. That chain lets a governance lead explain the input path before defending the AI outcome.

Automated evaluators are not magic, and they should not be treated as final truth. A scoring model can be miscalibrated. A metric can miss a domain-specific nuance. The answer is not to go back to gut feel; the answer is a visible control loop. Define the score threshold, name the owner, open an exception or remediation record when the threshold is crossed, and require review before expanded use continues.

Manual review still has a job. It should investigate the cases the scores surface and validate whether the measurement aligns with business risk. It should not be the only thing standing between a production AI system and users who assume the answer is safe to trust.

Accuracy Has to Leave Evidence Behind

A production RAG system earns trust by proving its work repeatedly. Launch approval is only the first checkpoint; standing evaluation is the control.

Continuous evaluation gives governance and quality teams a shared language for that proof: retrieval fit, response fit and source support. The labels are different because the failures are different.

Those are not abstract quality labels. They are evidence fields for deciding whether an AI experience deserves broader usage and more responsibility.

If you are evaluating RAG for production use, ask for the scores behind the answer, the trend after source changes and the control that runs when a score drops. To see the measurement loop, start with the REMi evaluation model or book a demo.

FAQ

Does Continuous Evaluation Replace Human Review?

No. It changes where human review spends its time. Instead of reading random answers and hoping the sample represents production, reviewers can investigate score patterns, calibrate thresholds and focus on cases where quality risk is visible.

Which REMi Metric Matters Most for Governance?

Groundedness is usually the most audit-relevant metric because it asks whether the answer is supported by retrieved context. Context Relevance and Answer Relevance still matter because they explain whether the failure came from retrieval, generation or a mismatch between the two.

What Should a Team Do When Scores Drop?

Start with the metric that moved. A Context Relevance drop sends the team toward retrieval, indexing, chunking or metadata. When Groundedness falls, look for unsupported answers; when Answer Relevance falls, review the generation step and how the question is being interpreted.

Manufacturing runs on precision. Production schedules are tight, supplier relationships are load-bearing, and when something goes wrong in the data flows connecting your operation to its extended supply chain, the consequences land on the plant floor fast.

That risk is more tangible than it used to be. In late August 2025, a publicly reported cyber incident in the automotive sector disrupted production for weeks, rippling through its supply chain and downstream partners. The affected company later disclosed £196 million in direct extra costs, with the wider UK economic impact estimated at about £1.9 billion, and described it as the most economically damaging cyber event in UK history.

This example is an extreme case. But the dynamic it illustrates—a cyberattack that doesn't just compromise data, it stops lines—is one every manufacturing IT and operations leader now has to plan for.

File transfers sit directly in the path of that threat. Every purchase order sent to a supplier, every engineering drawing shared with a contract manufacturer, every production schedule distributed to a logistics partner—each one is a data exchange that crosses your perimeter. And in many manufacturing organizations, those exchanges still run on infrastructure built for a simpler time: legacy FTP configurations, email attachments, shared drives with permissions nobody has reviewed in years.

Modernizing that infrastructure is what this post is about, and specifically how to do it without disrupting the operations that depend on it.

Why File Transfers Are a Supply Chain Vulnerability

The manufacturing supply chain is a network of trusted connections. Attackers exploit that trust.

When a threat actor wants access to a large manufacturer, the path of least resistance is rarely a direct assault on hardened internal systems. It's more often a smaller supplier, a logistics partner or a contract manufacturer. It’s likely to be an organization with a trusted data relationship and fewer security resources. File transfer channels are frequently that entry point: a server that hasn't been audited, an email workflow with no access controls on the receiving end, a shared folder where permissions have drifted over time.

The challenge isn't that manufacturers are careless. It's that file transfer workflows in most organizations grew organically, using tools that solved a narrow problem when the environment was simpler. Together, they create fragmented visibility. Teams may not always know what was sent, who accessed it, whether it arrived intact or whether a workflow failed silently overnight.

That fragmentation is both a security problem and an operational one. A corrupted file can reach the plant floor before anyone notices. A spec revision can reach three of four suppliers and miss the fourth. A transfer that fails overnight surfaces only when a production scheduler starts chasing answers at the start of the next shift.

What's Actually Moving Through Those Transfers

The data flowing through manufacturing file transfer workflows isn't generic. It's specific, sensitive and valuable.

Engineering drawings and CAD files represent years of R&D investment. Production schedules reveal capacity, demand and supplier dependencies. Compliance documentation underpins regulatory standing and customer contracts. Pricing and procurement data is commercially sensitive. For manufacturers operating in regulated industries, some of this data carries legal obligations around how it's transmitted and who can access it.

Plain FTP typically transmits without encryption, which may expose data in transit. Email attachments generally lack centralized controls once they leave your outbox. Shared drives can accumulate permissions nobody actively manages. And none of these methods produce a reliable audit trail for the question every IT and compliance team eventually faces: "Can you show us exactly what moved, when, and who had access?"

What a Modern Approach Looks Like

Secure file transfer in manufacturing should do more than move data. It should help protect data, support automation and provide visibility into exchanges.

That's the role of managed file transfer. Rather than relying on a patchwork of disconnected tools, MFT provides a governed layer for file movement across plants, systems and partners—with encryption, access controls and audit logging built in rather than bolted on.

Progress Automate MFT is built on this foundation, enabling you to orchestrate secure file workflows from a centralized cloud console while keeping execution where your data lives. Unlike traditional monolithic or other cloud-based MFT solutions, Automate MFT leverages lightweight agents to run file transfer tasks, helping manufacturers maintain control across segmented IT and OT environments as well as distributed sites. What the MFT approach delivers in principle is consistent regardless of environment: files move through a controlled, logged, encrypted channel rather than an ad-hoc one, and when something fails, the system can provide insight into where and why issues occurred.

For manufacturers managing dozens or hundreds of supplier relationships, that consistency matters operationally as much as it matters for security. New partners are onboarded into a standardized, secure workflow rather than a one-off arrangement. Every exchange is auditable and controlled, regardless of which partner is on the other end.

Compliance Requires a Paper Trail

Beyond the security case, file transfer practices have compliance implications that manufacturers can't afford to overlook.

US defense contractors operating under CMMC 2.0 face specific requirements around how controlled unclassified information is transmitted and logged. Manufacturers exchanging personal data involving EU residents may need to demonstrate appropriate data-handling under regulations such as GDPR. "We emailed it" is not typically a defensible answer under audit. More broadly, the ability to produce a complete, accurate record of data exchanges is increasingly a baseline expectation from customers, regulators and auditors across industries.

Ad-hoc transfer methods don't produce that record reliably. A governed MFT approach can help generate this record automatically, as a byproduct of every transfer, without manual record-keeping or after-the-fact reconstruction.

Modernizing Without Stopping the Line

The standard concern about changing file transfer infrastructure is disruption. These workflows touch production systems, supplier integrations and compliance routines, and, in manufacturing, changing everything at once isn't realistic.

The practical path is incremental. Start with the highest-risk or highest-volume exchanges: production data feeds with contract manufacturers, engineering file distribution to OEM partners, compliance workflows for regulated components. Each migration can improve security and visibility without requiring a full-scale overhaul.

The goal isn't new technology for its own sake. It's closing the gaps in encryption, logging and access control that accumulate when file transfer workflows grow organically—and replacing them with a foundation designed to improve security, auditability and scalability with the supply chain.

The Attack Surface You Control

Manufacturers can't eliminate every risk in a supply chain that spans hundreds of organizations. But the file transfer infrastructure connecting that supply chain is within your control. And for many manufacturers, it remains one of the most exposed and least monitored parts of the perimeter.

A managed file transfer approach can help address that gap.

Automate MFT is one way to get there, but the right starting point is a conversation about your specific environment, workflows and risk priorities.

Request a demo and see how Automate MFT can help you secure and manage the data flows your supply chain depends on.

AI is reshaping digital experiences, and not always in the way organizations expected. While generative AI has accelerated content creation, the real challenge has shifted to delivering personalized, trustworthy experiences at scale.

In this Q&A, Pierre Azzam, Founder and CEO of Belong, explores why governance, structured content and operational maturity matter just as much as AI itself, and what organizations can do today to prepare for a future of adaptive, AI-driven digital experiences.

About Belong

Belong is a Dubai-based digital product agency on a mission to “make digital matter,” by building “Living Platforms,” digital ecosystems grounded in empathy, powered by data and scaled through technology, designed to learn, adapt and evolve to deliver continuous value over time.

Established in 2012, Belong partners with leading organizations to craft and manage digital transformation programs across a wide digital ecosystem. As an implementation partner of Progress Sitefinity, Belong specializes in creating and engineering intelligent, personalized and AI-enabled digital platforms that evolve with the needs of the business and its users.

Section 1: How AI Is Changing Digital Experience Expectations

How Are Teams Using AI in Their Digital Experience Stack Today? And Where Is It Still Falling Short?

Organizations are primarily using AI to handle repetitive, high-volume tasks such as content generation, first-pass translations, tagging, workflow acceleration and search optimization. These are areas where the cost of being slightly wrong is low but the gains in speed and throughput are significant.

However, AI still struggles with contextual understanding, maintaining brand voice and operating reliably in customer-facing scenarios without human oversight.

What’s Driving the Current Urgency Around AI Adoption in Content and Experience Teams?

Three forces are driving urgency, none of which are purely about the technology itself.

The first is competitive pressure. As early adopters begin delivering more personalized and responsive experiences, others are forced to respond to avoid appearing outdated.

The second is internal. Marketing leadership is expected to define a clear AI strategy quickly, often before they’ve had time to fully understand its implications.

The third is cost efficiency. AI is emerging as the first credible response to rising content production costs, especially across multilingual environments.

How Is AI Changing the Way People Discover and Interact with Content?

Discovery is shifting from navigation and keyword search toward conversational intent. Users increasingly don’t land on full pages. They land on synthesized answers generated from multiple sources.

This has two implications. First, the unit of optimization is no longer the page but the passage. Second, content that survives summarization is structured clearly—built around claims, evidence, conditions and exclusions. Traditional editorial formats are less likely to be preserved in AI-generated responses.

Key Takeaway: Content needs to be structured for retrieval and summarization, not just page-based consumption. Clear, well-structured information is more likely to be surfaced and reused by AI systems.

Section 2: The Limits of Traditional Content and Experience Models

Where Do Traditional CMS and Digital Experience Approaches Start to Break Down Today?

Traditional content management system (CMS) architectures are organized around pages and the first break-point is the assumption that the page is the unit of experience.

Increasingly, the unit of experience is a dynamic composition assembled at request time, drawing from a content graph rather than a page tree. Platforms that treat pages as atomic struggle to support this model, while platforms designed around structured content, flexible APIs and integrated workflows are better suited to support this model at scale.

The second break-point is the assumption that content is authored once and reused many times. AI-assisted personalization inverts this model, requiring content to exist as multiple variants that can be assembled dynamically based on context.

As a result, CMS workflows shift toward shorter content units, more variants, richer metadata and fewer monolithic pages.

Why Are Static Pages and Predefined User Journeys Becoming Less Effective?

What is becoming less effective with static pages is the assumption that a single experience can serve every audience equally well.

The emerging model is a stable, indexable page with conditional elements that adapt based on context. The page still exists, remains auditable and is still what AI systems retrieve—but the experience within it becomes flexible.

Predefined journeys present a different limitation. They assume the organization understands the user’s goal better than the user does and that assumption has weakened significantly.

Teams seeing the best results are replacing rigid, linear funnels with intent-driven interfaces—designed to recognize where users are in their decision process and respond accordingly, rather than forcing them through a predefined sequence.

What Challenges Do Teams Face When Trying to Keep Content Relevant Across Different Audiences and Channels?

The most overlooked challenge is governance, not content creation.

AI can generate channel-specific variants quickly but the bottleneck has shifted upstream—to deciding which variants are approved, by whom, against which brand and compliance standards and how those decisions are recorded.

Many organizations have not modernized their governance models to match the speed of content generation.

Platforms like Progress Sitefinity CMS that integrate governance directly into the authoring workflow—rather than treating it as a separate system—are better positioned to scale content effectively.

Key Takeaway: As AI accelerates content creation across channels, governance becomes the primary constraint. Organizations need clear ownership, approval structures and compliance controls embedded into workflows to manage content at scale.

Section 3: The Operational Burden Behind Content at Scale

Why Does Managing Content at Scale Still Require So Much Manual Effort in Most Organizations?

Managing content at scale requires significant manual effort because the bottleneck is not in writing but in coordination. Authoring typically represents a small portion of the total effort. The majority is spent on approvals, translations, image rights, legal reviews, scheduling and stakeholder alignment.

While AI is improving content creation, it has limited impact on these coordination-heavy processes.

Workflow automation, structured approvals and content scheduling tend to unlock more capacity than adding additional AI writing tools on top of existing systems.

Where Do Teams Typically Lose Time or Efficiency in Their CMS Content Workflows?

Three patterns consistently reduce efficiency in CMS content workflows.

First, content reviews and approvals often become bottlenecks, particularly when multiple stakeholders—such as legal, brand or executive teams—are involved.

Second, translation rework occurs when source content changes after localization has already been completed, requiring teams to redo work across languages.

Third, asset reformatting remains highly manual. Images are frequently resized, recropped and re-exported for different channels, even though image transformation can be handled at the platform level.

Despite available capabilities, many teams continue to rely on manual tools and the cumulative cost of this inefficiency is significant.

What Kinds of Tasks Are Realistically Ready to Be Automated Today and Which Ones Are Not?

Some tasks are well suited for automation today, particularly those that are structured, repeatable and low-risk.

These include categorization, summarization, tagging, translation support, image and video reformatting, workflow assistance and recommendation logic. These outputs typically remain within a human review loop and integrate directly into existing CMS workflows rather than requiring separate systems.

Other tasks are less suitable for automation—especially those where AI-generated output reaches customers without oversight.

This includes areas such as product descriptions, legal disclosures, compliance-related content or chatbots handling sensitive queries like pricing or eligibility.

Organizations that remove human review may achieve marginal efficiency gains but take on disproportionate regulatory and brand risk. The more effective approach is to automate processes where humans remain responsible for final approval.

Section 4: Why Personalization Is Still Hard to Get Right

If Personalization Has Been Around for Years, Why Do So Many Organizations Still Struggle to Make It Work?

Many organizations invest in digital experience platforms but underutilize them as basic CMS tools, limiting their ability to deliver meaningful personalization.

A primary reason is the lack of dedicated resources and operational focus. Personalization is often treated as a feature rather than a sustained capability that requires ongoing content production and management.

As a result, organizations struggle to produce the volume of content variants needed to support personalization at scale, even when the platform itself is capable.

Another key challenge is ownership. Personalization is typically managed by digital marketing teams, while content creation is distributed across different functions. Without alignment, personalization efforts tend to result in minor variations rather than meaningful differentiation.

What Are the Biggest Limitations of Rules-Based or Segment-Driven Personalization Approaches?

Rules-based personalization becomes difficult to manage at scale because it relies on predefined assumptions about user behavior using “if/then” logic.

This approach struggles to capture real-time context or adapt to sudden changes in user intent. As the number of rules and segments increases, systems become harder to maintain and less responsive.

The shift toward intent-based or behavioral personalization is not about eliminating rules entirely but about replacing the parts of the system that cannot adapt, learn or respond dynamically.

To remain operationally viable, this evolution needs to happen within the same platform, rather than introducing additional tools that increase system complexity.

What Tends to Break as Organizations Try to Scale Personalization Across More Use Cases?

Two things typically break as organizations try to scale personalization, often in this order.

First, content production cannot keep pace with the increasing demand for variants. As personalization expands, the volume of content required grows significantly and many teams lack the capacity to produce and maintain it.

Second, governance structures fail to scale. What begins as a manageable set of variants with clear ownership can quickly become fragmented, with no consistent control over what content exists, who owns it or how it is used.

Without a clear content modeling strategy—defining priorities, audience segments and content variations—personalization efforts become difficult to manage and sustain.

Section 5: Moving Toward More Adaptive, Real-Time Experiences

What Does ‘Real-Time’ or Adaptive Personalization Mean in Practice?

Adaptive personalization refers to systems that continuously interpret contextual and behavioral signals to adjust experiences dynamically based on user intent, delivering relevance in the moment.

More advanced implementations move beyond predefined rules, allowing the system to learn what works over time. This requires a feedback loop where user interactions—such as impressions, clicks and outcomes—are captured and used to refine models and improve future decisions.

What Challenges Do Teams Face When Trying to Respond to User Intent in the Moment?

The most difficult challenge is accurately identifying user intent.

Intent is rarely explicit. A user engaging with a specific page may be researching, comparing options, ready to act, returning for information or simply browsing. The system cannot determine intent without additional context.

Inferring intent from behavioral signals is possible but only when those signals are clean, consistent and connected across sessions and devices.

Without integrated systems that consolidate behavioral data from multiple touchpoints, organizations often spend more time resolving identity and stitching data together than delivering personalized experiences. The personalization engine itself is rarely the limiting factor—the surrounding data infrastructure is.

How Do You Move from Predefined Journeys to Experiences That Can Adapt Dynamically?

The transition to adaptive experiences should be incremental and disciplined.

Attempting to replace an entire predefined journey at once—such as moving from a structured funnel to a fully adaptive experience in a single release—often fails. Supporting content, data and workflows are typically not mature enough and operational complexity increases faster than teams can manage.

A more effective approach is to retain the overall journey structure while introducing adaptive elements within it. Individual steps become responsive to user behavior, while the broader flow remains intact.

Over time, more steps can be made adaptive, gradually shifting from a fixed funnel toward a more flexible content graph that the system navigates based on user signals and intent.

Section 6: Trust, Accuracy and Control in AI-Driven Experiences

One Concern with AI Is Trust—How Can Organizations Deliver Accurate and Reliable Experiences?

Trust is becoming the defining factor in whether AI-driven experiences succeed or fail. Three elements must work together to support accuracy and reliability.

First, AI must be grounded in trusted enterprise content. Retrieval-Augmented Generation (RAG) architectures like Progress Agentic RAG help to base responses on approved sources, rather than relying solely on the model’s training data.

Second, the content itself must be accurate, audited and up to date, with clear governance over what information can be used. AI systems can only be as reliable as the information they retrieve and use to generate responses.

Key Takeaway: Trust in AI-driven experiences depends on grounding outputs in verified content, supported by governance and auditability. Without accurate, controlled and up-to-date information, AI systems cannot reliably deliver correct or compliant responses.

What Risks Do Teams Need to Manage When Introducing AI into Customer-Facing Experiences?

Teams need to manage several risks when introducing AI into customer-facing experiences. These include hallucinations and factual inaccuracies, which can lead to misinformation or incorrect statements about products, pricing or eligibility.

There is also a risk of regulatory missteps, particularly in industries where content must meet strict compliance requirements.

In addition, AI-generated content may not align with brand voice or tone, creating inconsistency across customer interactions.

These risks become significantly more serious when AI is deployed without proper grounding, governance and human oversight.

How Important Are Governance, Auditability and Content Control in This New Model?

Governance, auditability and content control are not optional—they are foundational to whether AI builds or erodes trust at scale.

Governance defines ownership and boundaries: who is responsible for AI-generated outputs, what the system is allowed to do and how escalation is handled when limits are reached.

Auditability provides traceability, enabling organizations to understand how outputs were generated, investigate issues and create feedback loops for continuous improvement. It also builds stakeholder confidence by making it possible to address errors or misstatements.

Content control means that AI outputs are grounded in accurate, current and approved information. Without it, organizations risk generating responses that are inconsistent, outdated or non-compliant.

Section 7: Preparing for AI-Led Discovery

How Is Content Discovery Changing as More Users Turn to AI Tools Instead of Traditional Search?

Content discovery is shifting from keywords to citations.

Brands that have historically optimized for search engines and human readers are now also being interpreted by AI retrieval systems which prioritize structured, factual and well-attributed content, often relying on metadata and content structure to interpret meaning.

As a result, content needs to be machine-legible without losing its voice—structured around clear claims and supported by evidence. This approach improves both human readability and the likelihood of being accurately summarized and reused by AI systems.

Key Takeaway: AI-driven discovery favors content that is structured, factual and easy to attribute. Organizations need to move from keyword optimization toward creating clear, evidence-based content, supported by structured content models and metadata that AI systems can reliably interpret and cite.

What Challenges Do Organizations Face When Trying to Make Their Content Visible and Useful in These Environments?

Organizations face several challenges when trying to make their content visible in AI-driven discovery environments.

First, discoverability is no longer the same as traditional search ranking. Content can be well-optimized for SEO and still not be surfaced by AI systems which rely on different signals and sources, often prioritizing structured and well-attributed content over traditional ranking factors.

Second, citation patterns are inconsistent and difficult to predict. AI systems do not follow transparent ranking rules, which introduces a new form of model risk for organizations relying on these channels for visibility.

Third, content freshness plays a larger role. Large content estates with outdated or rarely updated material become less likely to surface, even if the information remains technically correct.

As a result, the optimization landscape has expanded and many organizations are still adapting to how visibility works in these environments.

What Should Teams Be Thinking About Now to Stay Discoverable in an AI-Driven Landscape?

To remain discoverable in an AI-driven landscape, teams should focus on four core disciplines:

1. Structured content — Organizing information into reusable content models with metadata and flexible delivery, making it easier for AI systems to interpret and retrieve
2. Semantic clarity — Content that is written in clear, explicit terms, with unambiguous meaning
3. Authoritative attribution — Linking claims to verifiable sources, including visible references, timestamps and supporting evidence
4. Content freshness — Regularly updating content so that it remains relevant and more likely to be surfaced by AI retrieval systems

While these principles are not new, they now serve a dual purpose: supporting both human audiences and AI systems that interpret, summarize and cite content.

Key Takeaway: AI discoverability depends on how well content can be interpreted, trusted and kept up to date. Teams should prioritize structured content, metadata, clear meaning, verifiable sources and ongoing maintenance so that their content is retrieval-ready and remains visible in AI-driven environments.

Section 8: Making AI Practical Without Adding Complexity

Why Do Some AI Initiatives Add More Complexity Instead of Reducing It?

AI initiatives often add complexity when they are introduced on top of disconnected systems and fragmented workflows, without first simplifying underlying processes or establishing scalable foundations. In these cases, AI amplifies existing inefficiencies rather than resolving them.

A common issue is adding new AI capabilities that overlap with tools or functionality already present in the stack. This creates duplication, increases operational overhead and makes systems harder to manage.

A useful test before adopting any AI capability is whether it replaces an existing part of the workflow or simply adds another layer. If it sits alongside existing systems without consolidation, it is more likely to increase complexity than reduce it.

What Should Organizations Prioritize if They Want to See Real Value from AI Quickly?

Organizations should prioritize narrow, high-frequency tasks with clear inputs and outputs, embedded directly into existing workflows rather than introduced as separate processes.

The most effective approach is to augment skilled teams rather than attempt to fully automate entire workflows.

Measuring a focused set of metrics from the outset is critical to understanding impact and guiding iteration.

Adoption is more likely to scale through internal champions than through top-down mandates.

Organizations that see early success tend to start small, iterate quickly and prioritize delivering practical outcomes over over-engineering solutions.

Where Is the Most Practical Place for Teams to Start if They Want to Move Toward More Intelligent, Adaptive Experiences?

The most practical place to start is operational enablement.

This means structuring content, consolidating data, simplifying workflows, strengthening governance and identifying repetitive tasks where AI can deliver measurable efficiency gains.

Most organizations already have the necessary inputs—analytics, CRM data, search queries and form submissions—but these behavioral signals are often underutilized.

The real shift is not purely technological but organizational. It requires clear decision-making processes around who acts on which signals and how quickly those decisions are executed.

A practical starting point is to audit existing intent signals, define a small number of high-value audience segments and select one visible experience to make adaptive first.

Key Takeaway: The best starting point for adaptive experiences is not new technology but improving operational foundations. Organizations should focus on structuring content within their CMS, activating existing data and introducing AI in a small, targeted use case before scaling further.

Learn more about how Progress Sitefinity CMS is positioned to help your team get started with the power of AI in creating intelligent content and personalized experiences.