31 min read
31 min read
What happens when a company thinks it knows its customers, but in reality only knows an outdated version of them?
That risk is more common than it seems.
Organizations invest time in building segments, buyer personas, behavioral profiles, and needs maps. The problem isn’t in doing this. The problem arises when that information is assumed to be valid for too long and begins to be used as if the customer were a stable, predictable, and virtually unchanging entity.
But people don’t work that way.
The same customer may make very different decisions depending on the moment, the urgency, the channel, the economic context, their recent experience with the brand, or even what they’ve just heard from someone else. They may seek autonomy in one interaction and need guidance in the next. They may be price-sensitive today and prioritize convenience tomorrow.
Even so, many customer experience decisions are still based on profiles that were created months ago and are rarely reviewed with the same rigor as other business metrics.
That’s where the problem begins.
When the customer becomes a snapshot, the organization stops observing them in motion. It begins to design for a representation of reality rather than for reality itself.
And that difference may seem small in a presentation, but it becomes enormous when it affects decisions about service, channels, communication, personalization, customer journeys, and value propositions.
This article does not question the value of buyer personas, segmentation, or customer profiles as working tools. It discusses what happens when those tools stop evolving at the same pace as the people they’re meant to represent—and a question that few organizations ask themselves often enough: How many decisions about experience, service, communication, or value propositions are being made today based on a version of the customer that no longer fully corresponds to reality?
-The problem isn’t taking the snapshot
-The customer changes even if the PowerPoint doesn’t
-A persona is for understanding, not for pigeonholing
-What the customer does today may say more than what they said six months ago
-Signals matter when they change a decision
-CRM doesn’t need more data; it needs context
-Customer Experience should adapt as the customer changes
At a strategy meeting, someone opens a presentation and a slide appears that everyone recognizes. It shows a photo, a fictional name, an age, an occupation, some motivations, frustrations, preferred channels, and a phrase that attempts to summarize how that customer thinks.
The team nods in agreement. Marketing knows this person. Sales does, too. Customer Service has used this profile in a workshop. It’s even possible that this profile has been used to redesign a customer journey or make important decisions.
There’s nothing wrong with that scene.
In fact, building a customer profile is one of the most useful ways to force an organization to stop talking about “the market” as an anonymous mass. Good research allows us to recognize patterns, understand needs, and put a face to behaviors that, when viewed solely in a database, would be difficult to interpret.
The problem begins afterward.
It begins when that slide still appears exactly the same a year later.
When no one asks whether that customer is still buying the same way, whether they switched channels, whether they now shop around more before deciding, whether their relationship with the brand has strengthened or weakened, or whether what was identified as an important need still carries the same weight.
That’s when the snapshot ceases to be a tool for understanding the customer and begins, silently, to replace that understanding.
The ICX methodology is based precisely on an important distinction: researching, building segments, developing Persona Playbooks®, and mapping Customer Journeys generates very valuable initial insights, but those insights begin to lose their relevance if the process ends with the delivery of the document.
We could illustrate this in a very simple way:
We research the customer
↓
We build segments and personas
↓
We design their customer journey
↓
We define recommendations
↓
We deliver the project
↓
The organization continues to operate
↓
The client continues to evolve
The last step is the one that’s usually left out of the presentation.
>> How to Map the Customer Journey in 2026 to Reduce Churn and Complaints <<
And that matters because a Persona was never intended to be an exact copy of a person. It is a representation. It summarizes common patterns to facilitate decision-making. The problem arises when we forget that distinction and start treating a methodological construct as if it were a permanent description of every individual belonging to that segment.
Consider, for example, a customer whom a company identified as highly digital. She shops online, looks up information on her own, and almost never uses assisted channels. Everything points to her preferring autonomy.
Until she runs into a problem.
Her order doesn’t arrive. There’s a charge she doesn’t recognize. She needs to modify a contract term or resolve an issue that directly affects her money.
Suddenly, that “digital and independent” customer wants to talk to someone. And she wants it now.
Have You Changed Your Contact Person?
Probably not.
The context has changed.
That nuance may seem minor, but it has significant consequences. If the company designed its customer experience assuming that “this type of customer prefers self-service,” it might continue to steer them toward FAQs, chatbots, and forms at the very moment when they most need human assistance.
The picture wasn’t necessarily wrong. It just couldn’t tell the whole story.
That’s why customer insight requires two things that are sometimes confused.
On the one hand, it needs structure: segments, personas, customer journeys, needs, and moments of truth. Without them, it’s very difficult to turn thousands or millions of interactions into something understandable.
On the other hand, it requires movement: recent behavior, transactions, feedback, changes in the relationship, new preferences, and signals that let us know whether what we thought we knew is still true.
In the ICX methodology, the early stages build that foundation using data, behavior, the Voice of the Customer, and research, and transform it into segments, Persona Playbooks®, and Customer Journeys. But the model doesn’t end there: it incorporates an ongoing phase of intelligence and activation designed to keep that knowledge alive, detect changes, and enable the experience to respond to them.
That difference completely changes the conversation.
It’s no longer just about asking:
“Who is our customer?”
We also have to ask:
Photographs still have value. They help us recognize the person in front of us.
The danger lies in stopping to look at them because we think we already know them.
And that’s where the next problem arises: while our presentations remain unchanged, the client doesn’t stand still.
It’s also helpful for you to know:
>> What is Customer Experience and what is it for? <<Six months after that meeting, the presentation remains exactly the same.
The same person. The same motivations. The same pain points. The same journey. Even the same phrase that summed up what that client expected from the brand.
There’s just one detail: the client who inspired all of that has now been in a relationship with the company for six more months.
They either made a purchase or didn’t. They tried another channel. They compared alternatives. They had a good experience or accumulated several small frustrations. Perhaps they started spending less, switched categories, filed a complaint, stopped responding to communications, or began dealing with a different branch.
None of that automatically changes the chart.
And that’s where one of the most common contradictions in Customer Experience arises: we devote a great deal of effort to building knowledge about the customer, but far less to asking ourselves how long that knowledge will remain valid.
Time, by itself, does not render research incorrect. A deep-seated need can persist for a long time. Some expectations change slowly, and certain patterns can remain perfectly useful years later.
What does change much more rapidly, however, is the specific relationship a person has with a company.
Let’s consider a customer who has been shopping regularly for three years.
In January, they make four purchases.
In February, four times.
In March, three times.
In April, twice.
In May, one.
In June, none.
From the Person’s perspective, nothing has likely happened. He is still roughly the same age, has the same occupation, similar needs, and many of the characteristics that allowed him to be classified in the first place.
From the perspective of their behavior, something is happening.
January → February → March → April → May → June
4 purchases → 4 → 3 → 2 → 1 → 0
That sequence tells a story that no static description alone could convey.
Maybe they found an alternative. Maybe a recent experience strained the relationship. Perhaps their financial situation changed, they no longer need the product as often, or they’re simply shopping differently.
The data doesn’t yet explain why. But it’s already telling us something very important: the relationship has changed.
That’s where a more mature approach to customer experience begins to ask different questions.
Not only that:
Who is this customer?
Also:
What's going on with this customer?
The difference between the two questions may seem semantic, but it completely changes the way the experience is managed.
The methodology we’re using as the basis for this article specifically incorporates metrics capable of tracking these trends: recency and frequency of purchase, spending trends, retention by cohort, time to churn, reactivation rate, category affinity, reliance on promotions, and value by segment, among others. On the experience side, it includes metrics such as NPS by segment, Voice of the Customer feedback, complaints, moments that most impact the experience, and validation of persona attributes.
None of these metrics can replace a Persona.
What they do is something different: they allow us to verify whether the story we tell about the customer still aligns with the story the customer is writing through their behavior.
And this isn’t unique to retail.
At a bank, a customer may still belong to the same segment even as they begin to reduce their balances, stop using certain products, or increase their service inquiries.
At an insurance company, a customer may renew their policy for several years without any issues until a bad experience during a claim completely changes their perception of the company.
In a B2B company, an account may retain exactly the same firmographic profile—industry, size, revenue, location—while decreasing its use of the service, changing its internal contact, or beginning to explore alternatives.
At a university, a student does not cease to belong to a certain profile just because they begin to show signs of dropping out. But those signs are precisely what should change the way the institution interacts with them.
The profile remains the same.
The relationship does not.
This distinction is very helpful in understanding the problem.
Some data helps describe the customer:
And there are other signs that indicate something is changing:
The first set helps establish context. The second set helps interpret the moment.
A company needs both.
That’s why the CX Intelligence methodology is based on a different approach than a one-time study that’s then filed away. Initial insights are supplemented with transactional data, digital behavior, purchase history, surveys, feedback, and analytical models that allow us to continue observing what happens next.
This also forces us to accept an uncomfortable truth: a good decision made based on outdated information can turn into a bad decision without anyone having made an obvious mistake.
A campaign might have been perfectly designed for a particular segment when it was created.
A customer journey may have accurately reflected the most significant pain points when it was researched.
A persona may have accurately represented a group of customers when it was developed.
The problem arises when the organization continues to use those responses without reformulating the questions.
Because while the company sticks to its definitions, the customer gains new experiences.
And each experience can alter—even if only slightly—what the customer expects from the next one.
That’s where we need to make an even more important distinction: if a customer can behave differently without necessarily ceasing to belong to the same profile, then a Persona shouldn’t function like a box into which we place people.
It should function as a tool that helps us understand them.
And that’s not the same thing.
Learn more about this key topic for reducing churn:
>> CX as an Early Warning System for Critical Processes <<
And that’s not the same thing.
A Persona can help us better understand a group of customers. It can organize scattered information, reveal patterns, explain motivations, and provide different teams with a common language to discuss the experience.
What it shouldn’t do is become a label that determines, in advance, how we expect a person to behave.
Let’s imagine that a company has identified one of its customer profiles as “practical and independent.” This is a customer who values speed, prefers to solve problems on their own, and uses digital channels whenever possible.
The description works.
That is, until that same customer needs to make a decision they consider important.
They no longer want to figure it out on their own. They read comparisons, ask other people, review terms and conditions, call an advisor, and maybe even visit a branch.
Has he stopped being practical and independent?
Not necessarily.
You may still be practical and independent in most of your interactions. What changed was the importance of the decision you were facing.
That detail is crucial because People tend to summarize behavioral trends, not write the script for every future interaction.
When we forget that difference, we start to pigeonhole people.
The organization stops thinking:
“Customers with these characteristics tend to behave this way.”
And it begins to assume:
“This customer behaves this way.”
It seems like a small change in the sentence. In practice, it’s huge.
Perhaps a more useful way to visualize it is this:
The Persona emerges at the end of that exercise in understanding. Not at the beginning of the client’s journey.
However, once it has been developed, many organizations end up reversing the flow:
Persona → “this client is like this” → let’s design everything based on that conclusion.
That’s when a tool created to simplify complexity begins to obscure it.
The ICX methodology treats segments, Persona Playbooks®, and Customer Journeys precisely as models built on evidence—not as conclusions that are never revisited. The logic starts with combining data, behavior, the Voice of the Customer, and research; converting that evidence into experience models; and subsequently maintaining them through a continuous process of intelligence and activation.
There is one word that is particularly important in this logic: evidence.
A Persona should not be relied upon simply because it is well-designed, because the presentation was approved, or because the organization has been using it for three years.
It should be justified because customer behavior continues to support the hypotheses it contains.
That is why CX Intelligence even includes the validation of persona attributes as part of the analysis.
That changes the relationship we have with the tool.
We no longer just ask:
“Which Persona does this customer belong to?”
We can start asking more useful questions:
The “Person” then ceases to be a rigid classification and becomes a hypothesis that reality can confirm, refine, or challenge.
Let’s return for a moment to the practical, self-reliant client.
On Monday, they check their balance using an app and take care of it in less than a minute.
On Tuesday, they purchase a recurring product without asking for help.
OnWednesday, they receive a charge they don’t recognize.
OnThursday, they need to change an important term of their service.
In just two days, you can go from wanting zero human interaction to considering it essential to speak with someone.
If we looked only at the initial interactions, we would conclude that they need self-service.
If we looked only at the most recent interactions, we might think exactly the opposite.
Both interpretations would be incomplete.
The interesting thing isn’t deciding which of the two represents the “real customer.” Both do.
What changes is the moment they are experiencing.
And this explains why personalization based solely on profiles can fall short. Knowing that someone belongs to a certain segment helps establish a foundation. But to decide what experience to offer them right now, we also need to understand what’s happening in their relationship with the company.
The methodology itself brings this idea to CRM through variables such as Persona Affinity, Relationship Health, Current Signal, and CX Opportunity. In other words, the profile remains relevant, but it coexists with signals that allow us to interpret the current state of the relationship.
That nuance is important.
It’s not about eliminating Personas in favor of thousands of completely individualized experiences. That would be unrealistic for most organizations.
It’s about avoiding the opposite extreme: designing as if all customers who share a profile should always receive the same response.
Perhaps that is the most practical distinction.
Patterns help with design.
Context helps with decision-making.
A Persona can help us define the type of language that usually works, the most relevant channels, the predominant needs, or the friction points we need to address.
But when it comes time to interact with a specific person, we need something more.
We need to know if they’ve just made a purchase.
If it’s been weeks since they last did.
If they had a problem.
If they responded negatively to a survey.
Whetherthey’re strengthening their relationship with the company.
If is scaling it back.
If is repeating a usual behavior or if something has just changed.
Because a company may know the Person perfectly well and still get the experience wrong.
Not because their research was poorly conducted.
But because it was responding to who it thought the customer was, when the situation required understanding what the customer is actually experiencing.
And that’s exactly where the next step in this conversation begins.
If we want to prevent a Persona from becoming a mere box, we have to look beyond its description.
We have to look for the clues left by the real person.
Because, in many cases, what the customer is doing today can tell us more about the experience they need than what they told us when we researched them six months ago.
Is your experience truly aligned with what your customers expect?
Discover the ideal time for your UX strategy:
>> When to Implement UX Design in Your Business Strategy <<
Because, in many cases, what the customer is doing today can tell us more about the experience they need than what they told us when we researched them six months ago.
Let’s consider a simple situation.
In a survey, a customer explains that she values convenience, that she tends to be loyal to brands that work for her, and that she prefers to always shop at the same place. The conversation is productive, the information is consistent, and several people with similar characteristics respond in a similar way.
A profile is created.
Six months later, that same customer continues to say she values convenience and likely still considers herself loyal. But something begins to happen in her transactions.
First, she buys a little less.
Then she starts switching to another supplier.
After that, she stops buying a product category she used to purchase regularly.
, she responds to a promotion, comes back once, and then disappears again.
If the company looks only at what the customer said, the relationship appears stable.
If it looks at her behavior, the story is different.
And neither source is necessarily wrong.
That’s one of the most interesting aspects of trying to understand customers: people can’t always explain exactly how their own decisions are changing.
A customer may still think that a brand is their first choice and, at the same time, have started buying more frequently from a competitor. They may say that price isn’t the most important factor and yet consistently respond to certain promotions. They may claim to be satisfied and quietly begin to reduce their engagement with the company.
There isn’t necessarily a contradiction.
What a person says usually explains how they interpret their relationship.
What they do reveals how that relationship is actually unfolding.
We need both.
In Customer Experience, we tend to associate the Voice of the Customer with surveys, interviews, feedback, or qualitative studies. These are indispensable sources because they give us access to something that a transaction alone can hardly explain: emotions, expectations, frustrations, motivations, and reasons.
But the customer also “speaks” when they don’t respond to a survey.
They speak when:
These are not answers to a research question.
They are behavioral clues.
And they often appear before the customer has an explicit conversation with the company about what is happening.
The ICX methodology incorporates precisely this dual perspective. On the one hand, CX Intelligence seeks to understand the customer experience, its challenges, and the necessary interventions; on the other, Commercial Intelligence analyzes behaviors, opportunities, and what might be changing in the business relationship with the customer. These are not two separate conversations: one helps us understand how the customer experiences the relationship, while the other allows us to observe how that relationship manifests itself in specific decisions.
This combination avoids a fairly common mistake: treating experience and behavior as if they belonged to different worlds.
Because a poor experience can also manifest itself in a transaction.
It’s just that it’s rarely labeled as a “bad experience.”
Let’s imagine another scenario now.
A customer of a financial institution has been using their card regularly for two years. They have never filed a major complaint, and their previous surveys show high levels of satisfaction.
One month, a charge appears that they don’t recognize.
He calls the customer service center.
The call takes longer than expected, and resolving the issue requires several steps.
Finally, the problem is resolved.
He doesn’t file an additional complaint. He doesn’t post anything on social media. He doesn’t request to cancel the card.
From a traditional perspective, the incident is closed.
But over the following weeks, something different happens:
Regular use → incident → decreased use → shift in shopping habits → almost no transactions.
The customer never said:
“This experience undermined my trust, and I’m going to start using another card.”
They didn’t have to.
Their behavior began to speak for them.
That kind of signal is precisely what a static view can overlook. The profile remains the same. The demographic data hasn’t changed. Perhaps the segment hasn’t either. Even the latest satisfaction survey may still show a positive result.
What changed was the trajectory of the relationship.
And to see it, you have to look at the timeline—not just the data.
That’s why variables such as recency, frequency, spending trends, retention, reactivation, price sensitivity, affinity between categories, or time to churn are relevant within a Customer Experience Intelligence system. Not because they all need to be analyzed for every customer, but because they allow us to detect patterns that a single snapshot doesn’t reveal.
That difference deserves attention.
A $50 purchase says very little.
A $50 purchase every week for a year tells a different story.
Five weeks without a purchase after that usual pattern says even more.
And a poor service rating right before those five weeks adds a piece of information that could change our interpretation of everything that came before.
We could represent it this way:
Transaction
What did they do?
+ history
Is this normal?
+ Context
What was happening?
+ perception
How did he experience it?
= meaningful signal
That’s where intelligence begins to emerge.
Not when we gather more information about the customer, but when we can connect the dots to understand what might be happening right now.
The methodology approaches this combination by using different sources: transactions, digital behavior, purchase history, surveys, open-ended comments, and analytical models. The goal is not to indiscriminately transfer all that information to the CRM, but rather to first convert it into knowledge that can be used in the customer relationship.
This matters because organizations typically already have a considerable amount of data.
The challenge isn’t always about getting more.
It’s learning to recognize when a change warrants attention.
We need to be careful here, too.
If a customer buys less, it doesn't automatically mean they're dissatisfied.
They may have finished a project.
They may be traveling.
His financial situation may have changed.
They may need the product less frequently.
In B2B, a decline in usage may be due to seasonality or a change in the customer’s operations. In insurance, less interaction may be perfectly normal. In healthcare, it could even be a good sign.
That’s why observing behavior doesn’t mean interpreting every variation as a cause for alarm.
It means establishing enough context to distinguish between noise and signal.
One fewer purchase is a data point.
A sustained decrease over several periods is a pattern.
A sustained decline following several service incidents begins to form a hypothesis.
And if that same behavior appears among hundreds of customers with similar experiences, we are no longer observing just individual cases.
We are learning something about the experience.
That is precisely the leap from collecting data to generating insights.
This shift in the question forces us to reevaluate how we understand the customer.
The Persona still makes sense.
The customer journey still makes sense.
Research still makes sense.
But now they coexist with an additional layer that allows them to be constantly confronted with reality.
Do the customers we identified as highly service-sensitive continue to behave that way?
Do the moments we define as critical still have the same impact?
Is a new source of friction emerging?
Has a segment started to behave differently?
Does a trait we considered central to a Persona truly still explain their decisions?
The answer no longer depends solely on conducting a comprehensive study every few years. Part of it can be found in the signals customers leave behind every day.
And this takes us into even more challenging territory.
Because detecting that something has changed can be interesting for an analyst.
It can generate a dashboard.
It could even make for a great presentation to a committee.
But from the customer’s perspective, none of that matters if the company continues to do exactly the same thing.
A signal begins to have real value when it changes a decision.
A signal begins to have real value when it changes a decision.
Let's go back to the credit card customer.
The financial institution has already noticed that, following a service incident, the customer began using the product less frequently. The pattern does not appear to be random. There is a sequence clear enough to warrant attention.
So far, the organization has learned something.
But the customer hasn’t noticed any difference yet.
There may be a dashboard showing the drop in usage. An analyst may have identified the pattern. It might even come up in a monthly meeting as an interesting case.
As long as no one acts on that information, from the customer’s perspective, everything remains the same.
That’s one of the areas where many customer experience initiatives fall short: they’re good at explaining what happened, but much less effective at deciding what should happen next.
And it’s not always because of a lack of data.
Sometimes the exact opposite is true. There’s simply too much.
We know when they made a purchase, how much they spent, which channel they used, whether they called the contact center, how long the interaction lasted, what they answered in the last survey, and which products they have subscribed to.
The difficult question comes next:
What are we doing differently now that we know all that?
Let’s assume the signal arrives in time.
The institution identifies that this is a valuable customer who, until the incident, had a stable usage pattern and whose usage began to decline immediately after a bad experience.
There are several possibilities:
You could offer them a promotion to encourage them to use the service again.
It could call the customer to try to rebuild the relationship.
You could provide them with access to a more direct customer service channel.
You could acknowledge the problem they had and verify that it has truly been resolved.
Or I could decide not to do anything yet and continue to monitor the situation.
All of these are possible decisions.
The difference is that now the action doesn’t stem simply from the segment to which the customer belongs. It stems from a combination of who they are, what happened, and what signal they’re showing right now.
That’s where personalization begins to take on a whole new meaning.
It’s not just about changing the name in an email.
Nor is it sending a different offer just because someone belongs to segment A instead of segment B.
It’s about deciding which experience makes the most sense right now.
The ICX methodology embodies precisely this logic under the concept of the Next Best Experience. Unlike the traditional Next Best Offer, the response doesn’t always have to be commercial. It can involve facilitating, anticipating, helping, recognizing, recovering, rewarding, selling, or even deciding that the best course of action is to do nothing.
That last point deserves a moment’s consideration.
In an environment where companies can automate virtually any communication, there is a natural temptation: to turn every signal into a message.
They bought something: message.
They stopped buying: message.
Visited a page: message.
Opened the app: message.
Didn't open it: another message.
The risk is ending up using customer intelligence to increase the volume of contact, rather than to improve the experience.
That’s why the idea proposed by the methodology is so important:
Personalization isn’t about communicating more; it’s about intervening more effectively.
A customer who has just solved a complex problem may not immediately need a cross-selling campaign.
Perhaps they simply need the company to verify that the solution worked.
A customer who has been buying the same product for years might appreciate it if the next interaction were simpler, rather than someone trying to sell them something extra.
A user who is browsing information may not need any intervention just yet.
And a customer who has shown clear signs of frustration probably shouldn’t receive the same promotional message as someone whose relationship with the brand is at its peak.
The signal, therefore, should not automatically trigger a campaign.
It should trigger a decision.
Here, there’s an interesting shift in the way we think about the customer experience.
An approach based solely on profiles usually starts with something like this:
The customer belongs to segment X
↓
Segment X receives experience Y
It's organized. Easy to automate. And, in many cases, useful.
But when we incorporate behavior and context, the sequence becomes richer:
Customer belongs to segment X
+
has a certain relationship with the company
+
an event has just occurred
+
is exhibiting behavior that differs from its usual pattern
↓
What is the most appropriate course of action right now?
That’s when we stop using the segment as a response and start using it as part of the response.
The difference may seem subtle, but it significantly changes the quality of decisions.
Let’s consider three customers who belong to the same Persona.
All three have similar characteristics. They buy similar products. They use digital channels and have comparable economic value.
But today they are in different situations:
| Customer | What’s Happening | A Possible Response |
|---|---|---|
| To | It continues to behave as usual | Do not intervene |
| B | Has just had a bad experience | Repair the relationship |
| C | Their use has increased significantly | Recognize and facilitate |
The person hasn't changed.
The recommended experience did.
And that is precisely why the methodology proposes translating the analysis into concrete actions through CX Activation Plays®, such as protecting a valuable relationship at risk, responding to a bad experience, facilitating a repeat purchase, or supporting a customer’s reactivation.
This is also a good test for any customer analytics initiative.
We might ask ourselves:
If we can’t answer those questions, we probably have some interesting insights, but we still don’t have a fully functional end-to-end Customer Experience Intelligence system.
Because the goal isn’t just to accumulate signals.
It’s to turn them into insights.
And that’s where another important difference comes in: intelligence doesn’t have to reside in the same place as all the data that produces it.
A company may have millions of transactions, comments, clicks, tickets, surveys, and historical records spread across different systems. Trying to bring every single one of them into the CRM doesn’t necessarily help the sales executive, the consultant, or the service agent who has to decide what to do with a customer.
What that person needs isn’t always to see twenty years of history.
They need context.
They need to know, for example:
this customer is high-value, the relationship had been stable, an incident has just occurred, their activity has decreased, and there is a risk of deterioration.
That is no longer raw data.
It is an interpretation that enables action.
And that is precisely the bridge to the next part of this conversation.
Because if we want the signals to actually change decisions, then we have to answer a very practical question:
What should the CRM know about the customer to help the organization deliver a better experience, without turning it into a data warehouse?
What should the CRM know about the customer to help the organization deliver a better experience, without turning it into a data warehouse?
The answer is not: everything.
That’s often one of the first mistakes a company makes when it starts integrating customer experience with technology. Since more data is available, it seems logical to feed it all into the CRM: transactions, purchases, support tickets, surveys, digital browsing history, contact history, products, complaints, campaigns, visits, email opens, and clicks.
Before long, the customer profile ends up overflowing with information.
And, paradoxically, it can become harder to understand what’s going on with the customer.
Let’s imagine a customer service representative who receives a call.
In front of her is a customer profile with dozens of fields: sign-up date, products purchased, last purchase, city, segment, campaigns received, preferred category, historical value, NPS, previous tickets, and a long list of activities.
It’s all right there.
But the customer just said:
“I’m calling because I’ve been trying to figure this out for three times now.”
At that point, the agent doesn’t need any more information.
She needs to know what the information she already has means.
Is this the third call about the same problem?
Has the customer had a negative experience in the past?
Is this a key relationship for the company?
Is there a risk that they will leave?
Has anyone already promised a solution?
Should we resolve the issue, escalate it, offer compensation, or simply listen?
A CRM may contain all the data needed to answer those questions, yet still fail to provide any answers to the person handling the inquiry.
That’s the difference between having customer information and having context about the customer.
During a CRM implementation, it’s quite common to hear requests such as:
Each request has its own logic.
The problem arises when no one asks the question that should come first:
What decision do we want to help people make with this information?
If the data doesn’t influence a decision, perhaps it doesn’t need to be presented to the user.
It can still exist. It may be essential for analysis. It can feed into models, segmentations, or metrics.
It simply doesn’t have to take up the same space where someone needs to interpret the customer in a matter of seconds.
The ICX methodology proposes precisely this separation. The data and analytics layer retains transactions, digital behavior, purchase history, surveys, comments, and models. Not all of that detail necessarily makes its way to the CRM; what does travel over is the intelligence that can be used in the relationship: segment, affinity with a Persona, customer value, relationship health, purchase cycle, current signal, CX opportunity, recommended experience, preferred channel, and preferred time.
The phrase used in the methodology sums it up quite well:
“Data lives where it can be analyzed. Intelligence travels to where it can be activated.”
And that idea changes the role of CRM.
It should no longer be limited to answering “What do we know about this customer?”
It should also help answer , “What do I need to understand about this customer right now?”
Let's go back to our credit card customer.
Across different systems, the organization might have data like this:
Individually, these are records.
When viewed together, they begin to tell a different story:
A high-value customer with a stable relationship. After a service issue that was not resolved during the first contact, there is a significant reduction in their usual level of usage. This is a sign that the relationship is deteriorating.
That’s context.
And for the person handling the case, that single sentence can be worth much more than having five screens open with all the records that led to it.
The difference is similar to that between handing over the ingredients and serving the dish.
Data remains indispensable.
But someone—a rule, an analysis, a model, or a team—had to organize them, connect them, and turn them into something meaningful.
This connects directly to the problem with the photograph.
If the CRM only retains relatively stable attributes, the customer profile may end up being just another version of that diagram we created at the beginning of the article:
Premium customer. Digital. Self-employed. High affinity for a specific product. Low use of assisted channels.
All of this may still be accurate.
And yet, they still don’t tell us that they had a terrible experience yesterday.
That’s why it’s helpful to distinguish between two types of information.
What helps us understand who he is:
What helps us understand what is happening:
The first variables provide structure.
The latter provide movement.
And a truly contextual experience needs both.
Here’s another paradox.
Companies can build extremely comprehensive profiles and still continue to offer generic experiences.
A company may know that a customer:
And then send them exactly the same promotion as the rest of the customer base.
There was no shortage of data.
What was missing was linking that data to the decision.
Something similar can happen in different contexts:
In all cases, the underlying idea is the same:
the historical data explains the origins of the relationship; the signal helps determine what to do with it now.
This also has a very practical implication for those who interact directly with customers.
When everything is important, nothing stands out.
If a screen contains a hundred fields, twenty recent activities, and five indicators without a clear hierarchy, the user ends up doing something perfectly understandable: ignoring most of them.
Then something strange happens.
The organization has invested in integrating data to provide a smarter experience, but employees continue to make decisions based on the information they can remember, find quickly, or interpret on their own.
The technology exists.
Intelligence still depends on intuition.
That’s why the discussion about customer experience shouldn’t be limited to which systems are connected. It should also ask what context we’re providing to those who have to take action.
Perhaps they don’t need to see the statistical model that identified the risk.
They need to know that there is a risk.
You may not need to review six months of behavior.
You need to know that the usual pattern has just changed.
You may not need to know all the survey responses.
You need to understand that there is a recent experience that deserves attention before you try to sell something.
That’s what sets CRM apart from a simple electronic customer record.
It turns it into a trigger point.
And that’s exactly where the circle we’ve been building since the beginning of this article begins to close.
First, we take a snapshot to understand the customer.
Then we accept that the photograph ages.
We learned to observe the dynamics that unfold between one photograph and the next.
We turned those changes into signals.
And now we ensure those signals reach the place where someone can make a decision.
But there’s still one piece missing.
Because doing this once brings us right back to the problem we started with.
If we want Customer Experience to truly stop being a snapshot, the system has to keep learning as the customer continues to change.
If we want Customer Experience to truly stop being a snapshot, the system must continue to learn as the customer continues to change. That is, at its core, the point we’ve been arriving at from the very beginning. Not because personas are flawed, not because customer journeys are no longer useful, and not because research loses its value the moment it’s completed. The problem arises when any of these tools becomes the final word on a relationship that continues to evolve every day.
Let’s think back to that first meeting. The team did the right work: they researched their customers, identified patterns, built segments, developed personas, understood the most important moments of the journey, found friction points, and defined recommendations. The project ended well.
But for the client, nothing was over.
The next day, they continued shopping, calling, browsing, comparing, complaining, using products, leaving reviews, abandoning processes, or recommending the brand. Even as the project was wrapping up, the relationship continued to generate new information. That’s where Customer Experience stops being just a design discipline and starts requiring a continuous capacity for learning.
The ICX methodology is based precisely on that logic. The initial stages allow us to understand the customer, model the experience, and identify the gaps that currently prevent us from delivering the expected experience. But there is a fourth stage, Intelligence & Activation, whose function is to keep that knowledge alive, detect changes, and enable the organization to act on them.
The difference between the two approaches can be summarized as follows:
This shift is more profound than it seems. In the first model, the client is the subject of a project. In the second, the client becomes a continuous source of learning.
And that forces us to accept something that isn’t always comfortable: some of our own models will have to change. An attribute that seemed to strongly define a Persona may lose importance. A new friction point may arise at a stage of the journey that previously worked well. A secondary channel may become central. A segment that seemed homogeneous may begin to split into very distinct behaviors.
That doesn’t mean the research failed. It means that reality continued to evolve.
Many organizations track NPS quarterly, conduct annual surveys, or update their Personas from time to time. All of that is still necessary. The problem is that hundreds or thousands of experiences occur between two measurements, and some contain signals that shouldn’t have to wait until the next survey.
A single incident can quickly damage a relationship. A segment may start abandoning a process at a new point. A new policy can create friction. A change in channel can alter expectations. Even a sales behavior can foreshadow churn long before anyone responds to a survey saying they plan to leave.
That’s why continuous learning doesn’t mean researching everything all over again every week. It means having the ability to observe what happens between the major research moments.
Some insights require depth, and others require speed. An interview can explain very well why an experience is frustrating. A sustained change in behavior can quickly alert us that something is happening. A mature organization needs to know when to use each approach.
Perhaps this is the simplest way to summarize the entire article.
Good research produces photographs of immense value. It allows us to freeze time, observe in detail, and understand aspects of the client that would be difficult to see while everything is happening. But a photograph has one inevitable limitation: it doesn’t show movement.
Photography can help us understand who the customer is, what they need, what they value, and how they typically make decisions. Video, on the other hand, begins to show us what is changing, what has just happened, which patterns are strengthening or weakening, and where new friction is emerging.
We don’t have to choose between one or the other.
Photography provides depth. Video provides continuity.
And when the two are combined, the organization stops working with a static representation of the client and begins to understand the evolution of the relationship.
There’s one final part of the cycle that often receives less attention: what happens after we intervene.
Suppose we identify a relationship at risk and take steps to restore it. The work shouldn’t end there. We also need to know whether that intervention produced the expected result.
We might ask ourselves, for example:
Without that feedback, the system remains incomplete. Because it’s not enough to learn from the customer; we also need to learn from our own decisions.
The ICX methodology specifically links insights to concrete initiatives and the subsequent measurement of results. Even when it proposes starting with a pilot, it recommends defining an objective, a priority customer journey, segments, an experience initiative, responsible parties, and metrics that allow us to verify whether the intervention actually generated value.
That small change in verb form sums up a large part of the problem.
Many companies can rightly claim that they know their customers. They’ve done research, they have data, they’ve built customer personas, they’ve mapped customer journeys, and they listen to the Voice of the Customer.
The most challenging question is another:
What mechanism do they have in place to continue getting to know them?
What changed this month? What new behavior began to emerge? Which hypothesis was no longer valid? Which intervention produced a better result? What did we learn that should change the next client’s experience?
When an organization can answer these questions systematically, Customer Experience ceases to be a collection of projects and begins to become a management capability.
The ICX methodology summarizes this logic very clearly: CX Compass builds knowledge, CX Intelligence keeps it alive, and CX Activation turns it into experience. Behind these three layers lies an even simpler sequence: Data → Intelligence → Action → Experience → Growth.
That is probably what we should expect from a Customer Experience strategy today—not that it create a persona so detailed that it never needs to be modified, nor that it produce a definitive customer journey. What matters is that it has the ability to learn when reality contradicts what we thought we knew.
Because the customer will continue to change even after the research is done, will continue to make decisions even when no one is looking at the dashboard, and will continue to accumulate experiences even after the project has been closed.
And perhaps that’s the real mistake of turning the customer into a snapshot: not taking the picture, but believing that once it’s taken, there’s no need to look again.
At ICX CONSULTING, we support organizations that want to stop treating customer insight as a finished deliverable and start managing it as a capability that evolves alongside the relationship.
This work doesn’t start by discarding buyer personas, segments, or customer journeys. It begins by recognizing their limitations: good research allows us to build a valuable snapshot of the customer, but that snapshot needs to be continually cross-checked against what people do, say, and experience afterward. That’s why our methodology connects initial insights with an ongoing layer of intelligence that combines behavior, transactions, Voice of Customer, and recent signals to detect when a relationship begins to change.
From there, the challenge shifts from accumulating more information to turning it into decisions: which customer needs attention, which signal warrants intervention, which experience makes sense at that moment, and what information should be fed into the CRM so that the person interacting with the customer can act with context. Following this logic, CX Compass builds the knowledge, CX Intelligence keeps it alive, and CX Activation turns it into an experience.
The result we seek is not a more detailed customer profile or a more comprehensive customer journey. It is an organization capable of recognizing when what it knew about its customers begins to change, learning from those signals, and adjusting the experience before the gap between the model and reality leads to a loss of relevance, loyalty, or business.
If this conversation sounded familiar to you, it’s probably worth reviewing how much of what your organization thinks it knows about its customers is still true today. That’s what we’re here for at ICX Consulting.
Every organization faces different challenges when it comes to the customer experience.
If your organization continues to manage the customer experience based on segments, personas, or journeys that were created months or years ago—without continuously comparing them to what customers are actually doing today—schedule a diagnostic session with our senior team.
We’ll explore how up-to-date your organization’s understanding of its customers is, what recent indicators might be signaling changes in their behavior or in the relationship, and how to turn that information into more timely, relevant, and actionable experience decisions.
- Explore our success stories at www.icx.co
- Book a www.icx.co with our consultants
- Download exclusive design and CX tools from our CX Toolkit section
Yashin Fonseca : Jul 16, 2026
There is one metric that almost no executive committee examines as seriously as it should: the response rate to its own customer satisfaction surveys.
Yashin Fonseca : Jul 17, 2026
Your customers have changed. Has your experience changed with them?
/ICX_Insights%20content/Folder%20icon%20for%20insights.svg)
/ICX_Insights%20content/Folder%20celeste.svg)
/ICX_Insights%20content/Print%20gris.svg)
/ICX_Insights%20content/Print%20celeste.svg)
/ICX_Insights%20content/Share%20icon.svg)
/ICX_Insights%20content/Share%20celeste.svg)
/ICX_Insights%20content/Subscribe%20icon.svg)
/ICX_Insights%20content/Subscribe%20celeste.svg)
/ICX_Insights%20content/Start%20Here%20icon.svg)
/ICX_Insights%20content/Start%20here%20celeste.svg)