The most widely distributed B2B NPS benchmark published in 2026 runs on survey data collected between January 2019 and July 2021, and the publisher discloses this in its own FAQ at the bottom of the page. The headline says 2026. The fieldwork ended five years earlier.
That is the state of B2B customer experience statistics. Most circulating figures are older than their publication dates suggest, most are consumer-derived, and several compare populations that were never measured against each other.
Current research does exist. Two studies published in the last ten months carry disclosed methodology, stated sample sizes, and fieldwork from 2025. This article leads with those numbers, then traces each of the older circulating figures to its source and shows what it actually measures.
Part one: what the current research measures
Forrester, The State Of Business Buying, 2026
Published 21 January 2026, based on Forrester’s Buyers’ Journey Survey of nearly 18,000 global business buyers fielded during 2025.
| Finding | Value |
|---|---|
| Internal stakeholders influencing a typical buying decision | 13 |
| External participants influencing the same decision | 9 |
| Buying group size when the purchase includes genAI features | 14 members, against 7 without |
| Buyers reporting AI use during the buying process | 94% |
| Buying cycles in which procurement acts as a decision-maker | 53% |
| Buyers purchasing some form of trial before committing | more than 60% |
| Buyers making purchases of $10 million or more who run a trial first | 78% |
| Trial users who planned to convert to paid with the same provider | just over one third |
| Buyers in groups of six or more reporting clear benefits | 94% |
Two findings here matter more than the rest for customer experience measurement. The first is that a buying decision now involves 22 people across internal and external roles. The second is that trials do not convert. Just over a third of buyers who ran a trial planned to move to a paid version with the same provider, which makes the trial a measured experience surface with a two-thirds failure rate attached to it.
6sense, 2025 Buyer Experience Report
Published 12 November 2025. Main survey n = 3,744 buyers across North America, EMEA and APAC, with a companion survey of 766 additional responses. Median purchase value between $200,000 and $300,000, minimum spend $25,000 in the prior two years to qualify. Sample composition, role distribution and regional splits are published in the report’s appendix.
| Finding | 2025 | Prior years |
|---|---|---|
| Average buying cycle length | 10.1 months | 11.3 months |
| Point of first contact with a seller, as share of journey | 61% | 69% |
| Winning vendor was on the Day One shortlist | 95% | 85% |
| Buying groups ranking their shortlist before contacting a seller | 94% | not asked |
| First conversations initiated by the buyer | 79% | 82% |
| Buyers using LLMs during the buying process | 94% | not asked |
| Interactions with the winning vendor, per person | 16 | 17 |
| Vendors evaluated | 5.1 | 4.5 |
| Vendors the buyer had prior experience with | 3.8 | 3.2 |
| Buyers with prior experience of the vendor that won | 85% | not asked |
| Prior purchase journeys in the same category, per buyer | 8.6 | not asked |
| Buyers engaging sellers earlier to evaluate vendor AI claims | 58% | not applicable |
| Buyers engaging sellers earlier because of economic conditions | 62% | not applicable |
The finding that carries the most weight for experience design is the control case 6sense ran on its own headline. Buyers contact the eventual winner first roughly 80% of the time. That could mean first conversations are persuasive, or it could mean buyers contact whoever they already preferred. 6sense separated the 6% of buyers who had not ranked their shortlist before speaking to a seller. In that group the first vendor contacted won 57% of the time, not 80%. The order of contact reflects a preference formed earlier. It does not create one.
Figure 1. Year-over-year movement in the 6sense series, measured on a consistent methodology.
The two studies disagree, and the disagreement is informative
Both measure B2B buying in 2025. They report different buying group sizes and different AI-usage effects. 6sense finds LLM use has not reduced vendor interactions, holding at 16 per person. Forrester finds AI is the starting point but produces mistrust that sends buyers to human validators.
The samples explain most of it. 6sense screened for purchases over $25,000 with a median between $200,000 and $300,000, and 42% of respondents came from technology firms. Forrester surveyed nearly 18,000 buyers across a far broader population with no comparable spend floor. Higher-value, technology-heavy purchases behave differently from the general business-buying population. Neither number is wrong. They are answers to different questions.
Part two: eight fractures in the circulating figures
Fracture one: publication date and fieldwork date are different objects
CustomerGauge’s B2B NPS benchmark page is titled “B2B NPS Benchmarks by Industry: 2026 Data and Revenue Guide” and dated 16 June 2026. It reports that healthcare leads with a median NPS of 58 and telecommunications sits lowest near 31. It states that “this year” healthcare comes out on top.
Its own FAQ, further down the same page, states that the benchmark combines self-reported program assessments from 776 B2B companies collected between January 2019 and July 2021 with anonymised platform data. The two charts on the page carry filenames dated February 2022.
This is not a hidden flaw. CustomerGauge discloses the collection window plainly. The disclosure sits roughly 4,000 words below the headline, and almost nobody who quotes the 58 and the 31 reads that far.
The same pattern runs through the two most-quoted figures in the category. The delivery gap statistic was published in 2005, making it 21 years old. The 17% supplier-time figure comes from a survey fielded in 2017, making it 9 years old. Both circulate in 2026 content with no date attached.
The question to ask of any customer experience statistic is not when it was published. It is when the fieldwork closed.
Figure 2. The same five sources, dated by close of fieldwork rather than by publication.
Fracture two: the 80/8 delivery gap is two samples, and it is not B2B
Bain & Company published Closing the Delivery Gap in 2005, authored by James Allen, Frederick Reichheld, Barney Hamilton and Rob Markey. The body text reports that Bain surveyed 362 firms, found 80% believed they delivered a superior experience, then asked customers, who said only 8% of companies were really delivering.
The phrasing implies one continuous study. The source line under Figure 1 of that report does not. It reads: Bain Customer-Led Growth diagnostic questionnaire, n = 362; Satmetrix Net Promoter database, n = 375.
Two samples. Two sample sizes. Two instruments. The 80% is self-assessment by 362 firms. The 8% is derived from a Net Promoter database covering 375 companies. The two bars in the most famous chart in customer experience are not measuring the same set of organisations, so the 72 percentage point gap between them cannot be interpreted as a gap within any company.
Figure 3. The delivery gap with the sample sources restored.
The underlying point survives. Companies almost certainly do overrate their own service. The specific number does not survive, because a within-company comparison was never performed.
Three further problems compound it. It is 21 years old. It is not a B2B study, and the worked examples in the report are Vodafone consumer mobile plans, Superquinn grocery stores, Intuit TurboTax retail software, eBay message boards and American Express card activation. And the genuinely useful numbers in the same report are almost never quoted: more than 95% of management teams claim to be customer focused, but only 50% tailor products and services to customer needs, only 30% organise functions to deliver superior customer experiences, and only 30% maintain effective customer feedback loops.
Those three figures are within-sample, describe operating practice rather than perception, and are considerably more actionable than the headline.
The gap statistic also gets rebuilt periodically with fresh numbers and the same structural flaw. An IDC white paper commissioned by Emplifi reported in January 2022 that 87% of companies stated they provide excellent customer experience while only 11% of customers agreed. Different decade, different figures, same two-population design.
Fracture three: NPS benchmarks are built from consumers, and the vendor says so
Net Promoter Score is the default B2B customer experience metric. The published benchmarks that B2B companies measure themselves against are largely not B2B.
Qualtrics XM Institute states the limitation directly in its own NPS benchmark documentation: the benchmark is composed of consumer-facing industries, and by the nature of the consumer-facing study it can only measure customer relationships with B2C industries. It adds that if a company’s customers are primarily businesses, or the customer base mixes B2B and B2C, this may not be the best benchmark to use. It also notes the benchmark is US-only.
That is an unusually clear disclosure from a vendor, and it is routinely ignored. A B2B SaaS company comparing its NPS of 35 against a published cross-industry benchmark is comparing itself to consumers rating airlines and supermarkets.
Where B2B-specific figures exist, the spread by sector is wide enough to make a single benchmark meaningless. Retently’s data puts B2C at an average NPS of 49 against 38 for B2B, an 11 point difference. In healthcare the same comparison shows a 32 point gap, B2C at 70 against B2B at 38. A benchmark spread that varies by 21 points between sectors is a reminder that the metric is sector-specific, not a benchmark.
The mechanism is structural rather than a matter of service quality. B2B relationships involve longer cycles, multiple stakeholders and complex support needs, and the person answering an NPS survey is rarely the person who signs the renewal.
Fracture four: the customer in B2B is between 6 and 22 people, depending on the source
Every B2B customer experience metric assumes a customer who can be surveyed. B2B buying decisions are made by groups, and published estimates of group size now disagree by more than 3x.
| Source | Fieldwork | Buying group size |
|---|---|---|
| Gartner B2B Buying Journey research | 2017 onward | 6 to 10 decision-makers |
| 6sense Buyer Experience Report | 2023 to 2025 | 10 or more at $250,000 average deal size |
| Forrester Buyers’ Journey Survey | 2025 | 13 internal stakeholders plus 9 external participants |
Figure 4. The spread is a definitional difference, not a measurement error.
The Forrester figure is the widest because it counts a category the others do not. External participants, meaning analysts, consultants, peers, distributors and resellers, are not employees of the buying organisation and cannot be surveyed by the vendor at all. 6sense found separately that 76% of buyers engaged at least one such outside party during the journey, up from 71% the year before.
A customer experience survey returns one response per account, usually from the day-to-day user or admin who opens the email. The renewal is decided by a group that includes procurement, finance, IT, security and the line-of-business owner, most of whom never see the survey. Forrester’s finding that procurement acts as a decision-maker in 53% of cycles puts a number on one slice of that invisible population.
An account can return a promoter score from its power user and churn anyway, because the economic buyer was never asked.
This is the central measurement problem in B2B customer experience and it has no clean solution. It does mean any single-respondent metric should be treated as a partial signal about one relationship inside an account, not a verdict on the account. Behavioural measures carry more weight in B2B than attitudinal ones for this reason. Renewal, expansion and product usage are observed at the account level, while satisfaction is self-reported at the individual level. B2B customer retention benchmarks measure what accounts actually did, which is why they belong alongside a customer experience programme rather than downstream of it.
Fracture five: the 17% supplier-time statistic is from a 2017 survey
The most-quoted B2B buying statistic is that buyers spend only 17% of their purchase time meeting with potential suppliers. It appears constantly in 2026 content presented as a current finding.
Its source is Gartner’s Digital B2B Buyer Survey, n = 750 B2B buyers, fielded in 2017.
The full distribution, which is almost never quoted, is more useful than the headline:
| Buying activity | Share of buying group time |
|---|---|
| Researching independently online | 27% |
| Meeting with the buying group | 22% |
| Researching independently offline | 18% |
| Meeting with potential suppliers | 17% |
| Other | 16% |
Read as a whole, the finding is not that buyers avoid suppliers. It is that 22% of buying time is spent internally with the buying group, nearly as much as is spent with all suppliers combined, and that independent research online and offline together account for 45%.
Gartner’s own framing supports this. The report describes the problem as customers struggling to buy rather than reps struggling to sell, and identifies consensus creation as one of the hardest of six buying jobs.
The commonly attached corollary, that a single vendor gets only 5% to 6% of buyer attention, is a derivation rather than a measurement: 17 divided by three competing vendors gives 5.7%, and by four gives 4.2%. It is arithmetic applied to a 2017 figure.
Fracture six: share of the journey before contact is partly a spread and partly a trend
The claim that buyers complete most of their journey before contacting a vendor appears with at least four values.
| Claim | Value | Source and fieldwork |
|---|---|---|
| Journey completed before vendor contact | approximately 80% | Gartner, widely attributed |
| Journey completed before vendor contact | 70% | Forrester, 2019 |
| Point of first contact, as share of journey | 69% | 6sense, 2023 and 2024 |
| Point of first contact, as share of journey | 61% | 6sense, 2025 |
Two separate things are happening in that table, and most articles that quote it treat them as one.
The first is definitional. Journey and evaluation are not the same object, and contact and engagement are not the same event. A buyer who downloads a whitepaper has contacted the vendor by some definitions and not by others.
The second is a real trend inside a consistent methodology. 6sense measured the same variable across three annual waves and found it moved from 69% to 61%, which it reports as buyers contacting sellers roughly six to seven weeks earlier than before. 6sense also tested nearly 30 variables against that shift and could explain only about 12% of it, roughly six days worth. The change is real and mostly unexplained by the purchase characteristics measured.
Quoting a single number from this cluster as a fixed fact is not defensible. The direction is consistent and useful: most B2B evaluation happens before a conversation, on the vendor’s website and on third-party surfaces the vendor does not control.
What follows from the direction rather than the number is that the experience a buyer has with a company’s public content is part of the customer experience, measured or not. Most customer experience programmes begin at the point of purchase and therefore never see it. That is the reasoning behind treating bottom-of-funnel content and comparison surfaces as pre-sale experience design rather than as lead generation.
Fracture seven: attitudinal metrics and behavioural outcomes are different objects
NPS, CSAT and CES are frequently used interchangeably in B2B reporting. They measure different things over different time horizons and are not convertible.
| Metric | What it measures | Scope |
|---|---|---|
| NPS | Stated likelihood to recommend | Relationship, cumulative |
| CSAT | Satisfaction with a specific interaction | Transaction, immediate |
| CES | Effort required to get something done | Transaction, immediate |
A company can post strong CSAT on support tickets and a poor NPS, because CSAT measures whether the last interaction went well and NPS measures how the customer feels about the relationship overall. Neither measures whether the account renewed.
Qualtrics XM Institute’s cross-industry research finds that how an experience makes people feel predicts loyalty more strongly than whether it worked or how easy it was, which is a direct statement that transactional metrics and the relationship metric capture different variance.
The same body of research reports that promoters are 4.2x more likely to trust the organisation, 3.6x more likely to purchase more, and 4.4x more likely to forgive a bad experience than detractors. Those are ratios between two groups of people, not effect sizes for an improvement programme. Moving a company’s NPS by ten points does not multiply purchasing by 3.6.
Fracture eight: response rates decide the number before the customer does
Every self-reported customer experience metric is computed on respondents, not customers. In B2B, where account counts are small and survey fatigue is high, the respondent group is a self-selected minority of an already small population.
Two structural biases apply and pull in opposite directions. Customers with an active complaint are more motivated to respond, which depresses scores. Customers with a strong account-manager relationship are more likely to respond as a favour, which inflates them. Which effect dominates depends on how and by whom the survey was sent, a variable that is never disclosed alongside a published benchmark.
This is why findings about operational drivers transfer better than score levels. A finding that faster ticket resolution correlates with higher NPS within a consistent methodology says something about mechanism. A cross-company score comparison across inconsistent methodologies says very little.
Part three: what this means for B2B SaaS planning
Seven questions replace the single question of what good B2B customer experience is.
When did the fieldwork close? Not when the page was published. A 2026 headline can sit on 2019 data, and at least one widely cited benchmark does.
Is the benchmark B2B? Most published NPS benchmarks are consumer-derived, and the largest vendor says so explicitly in its own documentation.
Was it measured within one sample? The 80/8 gap was not. Any perception-versus-reality statistic should be checked for whether the same organisations appear on both sides.
Who answered? One respondent per account cannot represent a group of 13 internal stakeholders and 9 external participants, and the respondent is rarely the renewal decision-maker.
Is it attitudinal or behavioural? NPS is a stated intention. Renewal is an observed outcome. Only one appears in revenue.
Does the measurement start before the sale? Buyers rank their shortlist before speaking to any seller 94% of the time, and the winner is on the Day One shortlist 95% of the time. Most customer experience programmes never measure the part of the experience that decides whether a company enters the shortlist at all.
Does the sample match the purchase? A study screened for $200,000 purchases in technology and a study of the general business-buying population will disagree, and both can be correct.
The pattern is the same one that runs through every marketing statistics category: a number is only as good as the population it was measured on, and the date on the page is not the date of the measurement. Verifying both before quoting a figure is unglamorous and it is the entire job. SERPsculpt applies the same standard to organic performance reporting, and the full statistics library documents the sources behind every figure published. The production method behind that standard, including how primary sources are fetched and logged rather than summarised, is documented at tanjagavrilovic.com.
Summary: what each circulating statistic actually measures
| Statistic | Circulating value | What it actually measures |
|---|---|---|
| B2B NPS by industry | 58 high, 31 low | 776 self-reporting companies, fieldwork closed July 2021, republished as 2026 data |
| CX delivery gap | 80% against 8% | Two non-matched samples, n=362 and n=375, from 2005 |
| CX delivery gap, restated | 87% against 11% | IDC and Emplifi, 2022, same two-population design |
| Cross-industry NPS benchmark | Varies | Consumer-facing industries, US only, per Qualtrics |
| B2C against B2B NPS | 49 against 38 | Sector-dependent; the healthcare gap is 32 points |
| Buying group size | 6 to 22 | Depends entirely on whether external participants are counted |
| Buyer time with suppliers | 17% | Gartner 2017 survey, n=750, nine years old |
| Journey before vendor contact | 61% to 80% | Part definitional spread, part a measured trend from 69% to 61% |
| Promoter against detractor behaviour | 3.6x to 4.4x | Group ratios, not programme effect sizes |
| Winner on the Day One shortlist | 95% | 6sense 2025, n=3,744, up from 85% the prior year |
| Procurement as decision-maker | 53% | Forrester Buyers’ Journey Survey 2025, n approximately 18,000 |
Methodology and verification
Verified by direct primary fetch, this revision
Forrester’s blog post The State Of Business Buying: Risk-Averse Buyers Demand Proof, Not Promises, published 21 January 2026, retrieved in full, containing the 18,000-buyer sample size, the 13 internal and 9 external figures, the 14-against-7 genAI buying group comparison, the 94% AI usage figure, the 53% procurement figure, the 60% trial figure and the one-third conversion figure. Forrester’s corresponding investor newsroom release of the same date, retrieved in full, containing the 78% trial figure for purchases of $10 million or more and the 94% large-group benefit figure.
The 6sense 2025 Buyer Experience Report, retrieved in full including the methods section and appendix, containing all cycle length, point-of-first-contact, shortlist, LLM usage, vendor count, prior experience and economic uncertainty figures, along with the sample composition and the 57% control-case finding.
The CustomerGauge B2B NPS benchmarks page, retrieved in full including the FAQ section disclosing the January 2019 to July 2021 collection window, the 776-company sample, the healthcare median of 58 and the telecommunications figure near 31.
The IDC and Emplifi white paper announcement of January 2022, retrieved in full, containing the 87% and 11% figures.
Verified by direct primary fetch in the original production session
Bain & Company, Closing the Delivery Gap, 2005, retrieved in full including the Figure 1 source line disclosing the two sample sizes, the 95%, 50%, 30% and 30% operating practice figures, and the list of worked examples. Gartner buyer enablement material including the complete distribution of buying group time and the provenance line reading n = 750 B2B buyers, Gartner 2017 Digital B2B Buyer Survey. Qualtrics XM Institute benchmark documentation containing the B2C-only and US-only caveats and the promoter-against-detractor ratios.
The Bain PDF host and the Qualtrics article URL were not reachable during this revision. The figures are carried forward unchanged from the original direct fetch and are flagged accordingly rather than re-verified.
Derived, with calculation shown
- Delivery gap magnitude: 80 minus 8 = 72 percentage points
- Sample mismatch: 375 minus 362 = 13, drawn from two different instruments
- Age of the delivery gap statistic: 2026 minus 2005 = 21 years
- Age of the supplier-time statistic: 2026 minus 2017 = 9 years
- Age of the CustomerGauge benchmark fieldwork close: 2026 minus 2021 = 5 years
- Gartner time distribution sums to 27 + 22 + 18 + 17 + 16 = 100%
- Non-supplier share of buying time: 100 minus 17 = 83%
- Per-vendor attention: 17 divided by 3 = 5.7% and 17 divided by 4 = 4.2%
- B2C to B2B NPS gap: 49 minus 38 = 11 points
- Healthcare B2C to B2B gap: 70 minus 38 = 32 points
- Total Forrester buying group: 13 + 9 = 22 people
- Buying group size spread: 22 divided by 6 = 3.7x
- 6sense point-of-first-contact shift: 69 minus 61 = 8 percentage points
- 6sense buying cycle compression: 11.3 minus 10.1 = 1.2 months
- Journey-before-contact spread: 80 minus 61 = 19 percentage points
- 6sense unexplained share of the point-of-first-contact shift: 100 minus 12 = 88%
Reported secondary
Retently’s B2C and B2B NPS averages and the healthcare sector split are cited as reported by secondary customer experience publications rather than from Retently’s own report, which was not retrieved. The Gartner buying group range of 6 to 10 is cited as reported rather than from a retrieved Gartner primary document.
Rejected and excluded
A claim that NPS explains roughly 1% of variance in customer spending, attributed to Marketing Science Institute research covering more than 250,000 ratings, could not be traced to a retrievable primary source and is not used, despite being directly supportive of the argument made here.
A claim that only 23% of enterprise customer experience leaders still use NPS as a primary performance metric, attributed to a TELUS Digital and Statista survey, could not be traced to a primary source and is not used.
Aggregated NPS benchmark tables blending Bain, Satmetrix, CustomerGauge and Retently figures were reviewed and excluded because the constituent methodologies are not comparable and the blending is undisclosed.
The Qualtrics XM Institute B2B NPS benchmark itself is available only to Qualtrics customers inside the benchmark editor and could not be retrieved, so no B2B benchmark figure is attributed to XM Institute.
Forrester’s 2024 figures that 86% of purchases stall and 81% of buyers are dissatisfied with their chosen provider were reviewed and excluded from the current-research section. They come from the Buyers’ Journey Survey 2024 and have been superseded by the 2026 edition, which does not restate them. Including 2024 figures in a section defined by 2025 fieldwork would repeat the error this article documents.
Known gaps
The full text of both Forrester reports sits behind client access. All Forrester figures used here come from Forrester’s own public blog post and newsroom release rather than from the report body, so question wording and sub-sample sizes could not be inspected.
No publicly retrievable, methodologically transparent B2B-specific NPS benchmark with fieldwork from 2024 onward was found. Every B2B NPS figure available in the open is either vendor benchmark data with undisclosed sampling or, in the case of the most-cited one, fieldwork that closed in 2021.
No source reviewed reported customer experience survey response rates for B2B programmes, so fracture eight is argued structurally rather than quantified.
No study was found that measures the same B2B accounts on both self-assessed and customer-assessed experience, which is the study the 80/8 statistic is widely assumed to be.
Figures carry the fieldwork dates of their original studies. In three cases those dates precede publication of this article by five, nine and twenty-one years.