Personalization Limits: Where Marketing’s Ceiling Is

9 min read · 1,938 words

Personalization has a ceiling. Most companies serious about it passed it two or three years ago without noticing. The returns concentrate in three decisions, the cost of going further rises faster than the benefit, and in B2B the whole exercise often personalizes the wrong unit: an individual, not the committee that decides.

I ran a personalization programme at one of the industrial businesses I worked at that promised a lift the business case never delivered. We had segmented email content, web modules keyed to account industry, and a scoring model that fed sales roughly a third more “warm” leads than blanket campaigns did. Pipeline moved. Revenue did not move nearly as much as the model predicted, because we had personalized the parts of the funnel that were cheap to personalize and left the part that actually decided the sale untouched.

What does McKinsey’s personalization research actually measure?

The number everyone quotes is McKinsey’s. Its Next in Personalization research puts typical revenue lift at 10 to 15 percent, ranging 5 to 25 percent by sector. The more interesting figure sits one line below: companies growing faster than peers draw 40 percent more of their revenue from personalization than slower-growing companies do. That is a correlation between two things fast-growing companies are already good at, not proof personalization alone produced the growth. McKinsey’s trillion-dollar value estimate is an aggregate opportunity across US industries, not a guaranteed return per company, closer in spirit to the caveats I raised about data-driven marketing getting some things right and others wrong.

Did Gartner’s prediction that marketers would abandon personalization come true?

In 2019, Gartner predicted 80 percent of marketers who had invested in personalization would abandon it by 2025, citing weak ROI and customer data management as the two drivers. That is now testable, and a softer version held up. Gartner’s own follow-on research found 63 percent of digital marketing leaders still struggling with personalization technology, with only 17 percent using AI and machine learning broadly for it despite 84 percent believing the technology helps. Full abandonment did not happen, mostly because personalization is bundled into martech nobody unplugs. The ROI problem the prediction was built on never went away. It just stopped being measured.

Forrester’s consumer research makes the same point from the buyer’s side. Its 2025 Media and Marketing Survey found only 53 percent of US online adults said they like it when companies personalize interactions, the second year running that sentiment has landed on “meh.” A tactic barely half the audience says it likes is not one with headroom to expand into every touchpoint.

Where does personalization actually pay off, and where does it stop?

The Boston Consulting Group’s Personalization Index, built from 200 brands across seven sectors, found leaders post compound annual growth 10 points higher than laggards, with $2 trillion of value at stake industry-wide. Even BCG’s own sector notes carry a caveat: financial services consumers are “privacy conscious and often skeptical of personalized nudges,” and digital-native categories carry a live risk of “consumer fatigue.” The value is concentrated, not universal, in the same report that produces the headline number.

Across the research, a line is consistent enough to draw. It sits at three decisions: what to show (product or content recommendation), when to show it (timing and channel), and how to phrase it (tone for the segment). Push past those three into identity-level specificity and the incremental lift falls off while the downside risk keeps climbing.

Why does the creepiness cost rise faster than the relevance benefit?

Relevance and comfort do not move on the same curve. Relevance rises in a straight line as you add more data about a person. Comfort rises too, for a while, then bends downward once personalization implies knowledge the customer did not consciously hand over. YouGov’s 2025 report found 54 percent of Americans describe the ads they see as creepy or invasive, across the US and sixteen other markets. Salesforce shows the ceiling on comfort even under the best conditions: in its State of the Connected Customer research covering over 13,000 consumers and nearly 4,000 business buyers across 29 countries, 61 percent said they were comfortable with companies using relevant personal data transparently, up from 52 percent two years earlier. That still leaves close to two in five people uncomfortable with the best-case version of personalized data use. The academic literature has a name for this: the personalization-privacy paradox, reframed in a 2024 study on transformative privacy calculus in Psychology & Marketing, which argues the trade-off consumers accept resets every time a brand’s data behaviour changes.

Most marketing teams treat the creepiness line as a legal or brand-safety question. It is a return-on-investment question. Every rupee spent stitching identity across devices to push specificity past what a customer consciously disclosed is a rupee spent past the point where the YouGov number works against you, not for you.

Why is B2B personalization solving for the wrong unit?

This is where the case against over-investing gets strongest. Deloitte Digital’s B2B personalization research found 73 percent of B2B buyers expect a consumer-like personalized experience, and personalization leaders were roughly three times as likely to have exceeded 2023 revenue goals by 10 percent or more. Taken alone, that argues for going further. Taken alongside how buying committees actually behave, it argues for going further in a different direction entirely.

Gartner’s 2025 survey of 632 B2B buyers found buying groups now run five to sixteen people across up to four functions, and 74 percent of those groups show unhealthy conflict during the decision. The number that should stop every B2B marketer chasing 1:1 personalization is this one: individual-level content tailoring produced a 59 percent negative effect on group consensus, while group-level tailoring, content built for what the whole committee needs to agree on, produced a 20 percent positive effect. Personalizing to the individual does not just fail to help the committee decide. It makes the group worse at agreeing.

That is the structural reason the ceiling exists. A B2B purchase is decided by a group with internal politics and competing incentives, close to the buying-committee argument I made in journey maps built to survive the buying committee. Individual-level personalization treats that group as a single persona with a name on a CRM record. It optimises for the wrong entity, and the more precisely you personalize to the wrong entity, the further you get from what the group actually needs to move.

What does India’s DPDP Act change for personalization specifically?

India’s Digital Personal Data Protection Act, passed in 2023 with its Rules following in 2025 and the Data Protection Board operational from 13 November 2025, sets consent as the primary basis for processing personal data, without the “legitimate interest” latitude GDPR gives European marketers. It bars tracking, behavioural monitoring or targeted advertising directed at children, and penalties for serious breaches start at Rs 50 crore. For an Indian consumer business, that converts identity-stitched, cross-device personalization from a data-engineering project into a legal-risk decision made by counsel, not the growth team.

Indian consumer behaviour already shows the strain regulation is catching up with. A 2026 LocalCircles study of 22,119 respondents across 328 districts, focused on insurance, found 90 percent had received repeated unsolicited calls or prompts, 85 percent had been asked for unnecessary personal information, and 77 percent said platforms used their data to spam them or reject genuine claims. That is not personalization done well. It is personalization infrastructure run without the restraint the DPDP Act now makes mandatory. Zomato and Nykaa both run useful recommendation and reorder personalization in Indian ecommerce, and it works because it stays inside the narrow band: what to show and when, built on data the customer visibly gave the platform through its own browsing and ordering.

What this means for your personalization budget

The practical rule is short. Personalize the decision, not the person. If a piece of data lets you choose which product, which message, or which moment, use it. If it only lets you sound like you know more about the individual than they consciously told you, that is where the spend should stop, because the research above shows the marginal return past that point is flat at best and negative in B2B specifically.

1. Audit against the three-decision list. Recommendation, timing, segment-level tone. If an initiative isn’t improving one of those three, question its business case before its execution.
2. Put a comfort ceiling in the business case. Salesforce’s own data shows even transparent, beneficial use tops out near 61% comfort. Model returns against that ceiling, not against full adoption.
3. In B2B, build for the committee first. Use Gartner’s own finding as the test: does this content help five to sixteen people agree, or does it just flatter one inbox?
4. Treat DPDP consent as a design constraint. A tactic that only works because consent was implied, not given, has no future in the Indian market.
5. Re-test the abandoned 80 percent. Gartner’s 2025 prediction did not fully happen, but the ROI problem it named did not go away. If your programme has never been re-measured against its original business case, that is the fix, not more data.

Frequently asked questions

Does personalization actually increase marketing revenue? Yes, within limits. McKinsey’s Next in Personalization research finds a typical revenue lift of 10 to 15 percent, concentrated in product recommendation, timing and message relevance. The lift does not scale linearly with more personal data, and past a certain specificity the marginal return flattens or turns negative.

Did Gartner’s prediction that 80% of marketers would abandon personalization come true? Not literally. Gartner’s 2019 prediction targeted 2025, and full abandonment did not happen because personalization got embedded into standard martech. But the underlying problem, weak ROI measurement, persisted: Gartner’s own later research found 63% of marketing leaders still struggling with the technology.

Why does personalization feel creepy past a certain point? Relevance and comfort do not rise together indefinitely. YouGov’s 2025 research found 54% of Americans call personalized ads creepy or invasive, and academic work on the personalization-privacy paradox shows comfort depends on whether the data used was consciously disclosed, not just whether the message is accurate.

Does 1:1 personalization work in B2B marketing? Rarely, and Gartner’s 2025 buyer survey shows why: individual-level content tailoring produced a 59% negative effect on buying-group consensus, while group-level content improved consensus by 20%. B2B purchases are decided by committees of five to sixteen people, not the single contact a CRM record names.

What does India’s DPDP Act require for personalized marketing? Consent is the primary legal basis for processing personal data, with no general legitimate-interest exception. Tracking, behavioural monitoring and targeted advertising aimed at children are barred outright, and serious breaches carry penalties starting at Rs 50 crore, effective under the Data Protection Board operational from November 2025.

My honest opinion is that the personalization industry sells specificity as if it were value, when the research it cites keeps showing the value sits in three decisions and specificity past those three mostly buys risk. Where has your own personalization spend stopped paying for itself, and did anyone go back to check?