Tiago J. C. Sousa
PTEN

18 November 2026 · 9 min read

Everyone personalises. Almost no one is remembered.

Wall of repeated text "Hi [Name]," with the line "Almost no one is remembered." highlighted

In 2019, a B2B prospecting email had an average reply rate of 8.5%. In 2026 it has 3.43% (Instantly, via TechRound). Over the same period, personalisation went from a first-name field to models that write a different email for every recipient.

We have never personalised so much. We have never been ignored so much.

Source: Instantly, cold email benchmark 2026, via TechRound

This is not a paradox. It is a simple economic rule: whatever becomes free stops being a differentiator. AI made personalisation free. So it stopped being a competitive advantage and became noise with our first name on it.

Relevance is now table stakes

71% of consumers expect personalised interactions and 76% get frustrated when they don't get them (McKinsey). Those numbers justified a decade of investment in CRM and automation. They still hold, but they say less than it seems.

They say that the absence of personalisation is penalised. They don't say that its presence sets you apart. When any competitor can generate ten versions of an email in seconds, "Hi Anna, I saw you opened an office in Manchester" no longer proves attention. It only proves that someone switched on a tool.

B2B buyers changed at the same pace:

Buyers already have filters, human and algorithmic, trained to spot the pattern. The result is a market where every message looks personalised and none of them feels personal (Revnew).

The cost that never shows up on the dashboard

The campaign report shows the good side. After personalised interactions, customers are 1.8 times more likely to pay a premium price (Gartner, 1,464 B2B buyers and consumers, 2025).

The same study shows the side no marketing dashboard captures (Gartner, via Digital Commerce 360):

  • 53% of customers felt personalisation did more harm than good in their last purchase.
  • Those customers are 3.2 times more likely to regret the purchase and 44% less likely to buy from the same brand again.
  • People exposed to personalisation felt twice as overwhelmed by information and 2.8 times more pressured to decide.

Source: Gartner, survey of 1,464 B2B buyers and consumers (2025)

In other words, the gain shows up now and the loss shows up later. The customer pays more today, and that goes into the campaign report. But they buy with less confidence, regret it and don't come back. That loss only becomes visible months later, in a renewal that doesn't happen or a customer who leaves, and almost nobody links it to the campaign that caused it.

Nor can this cost be fixed afterwards. In a study cited by VKTR, five out of six remedies tested, including discounts and explanations of where the data came from, failed to reverse the negative reaction (VKTR). Trust is lost in one message and is not won back with the next.

We are asking the wrong question

In almost every CRM project, the conversation about personalisation starts in the same place. What data are we missing? Which fields do we need to create? Which integration will enrich the profile?

The premise is that more data produces more relevance. The numbers say otherwise:

Beyond a certain point, each extra field adds more risk than relevance. And repeating the same well-segmented message costs as much as getting the price wrong.

Before opening the CRM to create one more field, it is worth writing down on a sheet of paper the three or four things the company wants the customer to understand in every interaction, whatever the segment. Only then does it make sense to decide what data is needed for that. Often it is less than what is already there.

Passive versus active personalisation

Gartner distinguishes between two models. Passive personalisation infers what the customer wants from their behaviour: they looked at product X, they get product X. Active personalisation involves the customer in the decision: it asks, lets them choose, helps them move forward.

The difference is measurable. Customers engaged through active personalisation are 2.3 times more likely to complete a critical purchase decision with confidence (Gartner, via Demand Gen Report). Active personalisation also reduces the feeling of intrusion, because the customer knows exactly what they shared and why (Gartner, via Document Media).

Source: Gartner, survey of 1,464 B2B buyers and consumers (2025)

The Dove x Pinterest case is active personalisation at its best. The "Real Beauty DNA" campaign did not infer preferences. It asked people to define their own beauty standards by choosing images and qualities, and used those choices to shape the recommendations. It reached 27 million people and delivered a 2.9-point lift in association with the Real Beauty message (WPP, cited in MarTech).

More interesting than the 27 million is how the campaign was built. Dove already had its own idea of real beauty, and the personalisation simply gave each person a way to apply it to themselves. The message stayed the same for everyone. Only the path to it changed.

In B2B, the translation is direct. A consultancy that believes technology without adoption is waste does not personalise its email with the recipient's job title. It personalises it with the question nobody has asked them about adoption of the tool they are already paying for. And instead of inferring interest from a download, it asks them what decision is on their desk.

Four decisions before asking the CRM for one more field

1. What never changes. Before defining what varies, write down what stays fixed: the point of view, the tone, the promises, the topics the brand stays out of. These rules come before letting AI generate variations. If a variation can contradict the brand, the problem is not the model. It is the missing rule.

2. Which signals deserve a response. A download is not purchase intent. Three visits to the pricing page in the same week might be. Define explicitly the few signals that trigger action. Everything else stays in the CRM as context, not as a trigger. Treating an isolated behaviour as a confirmed need is guessing, not understanding.

3. Where to ask instead of infer. Map the moments when the customer moves from one task to the next, for example from evaluating options to choosing. Gartner recommends focusing active personalisation precisely on those transitions (Demand Gen Report). At those points, a question, a diagnostic or a configurator is worth more than a dozen inferred data points.

4. When to stay quiet. The most mature automations have silence rules: a cap on contacts per person per week, sequences that pause automatically after a meeting is booked, campaigns suppressed while a support request is open. Respect is also measured by what you don't send.

Then measure against the goal. If the goal is to remove friction, measure whether the person got where they wanted to go. If it is decision confidence, measure regret and repeat purchase. If it is loyalty, measure retention and renewal. Opens and clicks tell you whether the email was delivered and read. They don't tell you whether the customer trusts the company more than before.

"But personalisation works"

It does, and there is no point pretending otherwise. Faster-growing companies generate 40% more revenue from personalisation than their slower-growing peers (McKinsey). Gartner itself measures the 1.8x premium mentioned above.

But those figures measure who personalises well, not who personalises more. And customers are saying clearly that most don't personalise well:

  • Only just over a quarter of consumers say brands understand their interests "very well". More than two in five have stopped buying from a brand because its personalisation felt repetitive, too personal or off the mark (Site Impact, 800 consumers, via CX Dive).
  • 64% prefer personalised experiences, but only 39% think the benefit outweighs the cost of sharing their data (Qualtrics 2026, via CMSWire).

The gap between 64% and 39% is exactly where trust is won or lost. Personalisation itself is not in question. What is in question is the idea that more volume and more data automatically produce better personalisation.

Human first, AI forward

AI has solved "how to personalise". It solved it so well that it made it ordinary.

"Why personalise", and above all "what never to personalise", is still a human decision. It is a strategic decision, not a configuration one. And it is the only one a competitor cannot copy by switching on the same tool.

When everyone seems to know the customer, the winner is whoever has something to say to them.

Sources

  • McKinsey & Company, The value of getting personalization right, or wrong, is multiplying
  • Gartner, survey of 1,464 B2B buyers and consumers (Nov-Dec 2024, published June 2025), via Demand Gen Report, Digital Commerce 360 and Document Media
  • Gartner, surveys of 632 B2B buyers (Aug-Sep 2024) and 646 B2B buyers (Aug-Sep 2025)
  • Customer Experience Dive, ThoughtSpot and Site Impact studies, Gartner reference on data dimensions
  • eMarketer, Amperity 2026 Consumer Priorities Report
  • CMSWire, Qualtrics 2026 Report
  • VKTR, The Line Into 'Creepy' Personalization Isn't Where Marketers Think It Is (30 Oct 2026)
  • TechRound, Instantly cold email benchmark 2026
  • Revnew, AI-generated leads not converting
  • MarTech, Jonelle Chandler, How to stand out in a sea of personalization (10 Nov 2026), Dove x Pinterest case (WPP)