ES Book a demo
Platform
Products
RAtrivity ReadinessAAtrivity Assistant
How it works
How Atrivity worksWhat reports you getAtrivity vs generic AICompliance and securityImplementation and service
Use cases
Launch readiness and supportWin against competitorsAssess and ensure readinessHCP and channel engagement
Evidence
What the users sayLife Sciences knowledge benchmarkResults from a sample of 100 programsMeasurement methods
Industries
PharmaMedtechDental CareAnimal HealthAutomotiveRetailTechnologyConsumer ElectronicsInsuranceBanking
Company
AboutBlogContact

Evidence · Life Sciences knowledge benchmark

56 million answers. One honest picture of what the field knows.

Every time a rep, a medical liaison or a pharmacist answers a question in Atrivity, we record whether they got it right first time, whether they got it right in the end, and which wrong answer they chose. This is what twelve years of those answers show.

56,564,235answers recorded in Life Sciences, September 2014 to August 2026
11.9
years of continuous data
3,847
campaigns in this benchmark
19
languages
5.6 M
knowledge gaps found at first attempt
2.0 M
gaps closed and verified in the same program
Readiness

Teams start lower than anyone assumes. They do not stay there.

Same people, same questions, same program: first attempt against last attempt. No comparison between different groups, no inference over time.

76.1%→84.7%

Accuracy across every answer, first attempt to last

+8.6 points
Medtech and diagnostics+9.5
Dental care+8.0
Animal health+8.6
Pharma+7.6
Consumer health+6.0
Medical and infant nutrition+6.1
70%75%80%85%90%
first attemptlast attempt

Campaign medians by industry. Medtech enters lowest and gains the most: the program works hardest where the gap is widest.

Readiness axes

Six readiness axes, across all Life Sciences programs.

Every field-facing question is assigned to one of the six axes Atrivity reports on. Clinical evidence starts with the lowest initial knowledge. On every axis, programs reduce gaps by about a third.

Product Readiness

Product, dosing, use, safety
75.6%→84.4%
initial knowledge
Initial gaps1.11M
Final gaps707k
Gaps reduction rate−36.2%
38.1k test questions
10.6M answers27.0% of the content

Market Readiness

Competition, positioning, access
79.0%→86.4%
initial knowledge
Initial gaps401k
Final gaps260k
Gaps reduction rate−35.1%
17.2k test questions
4.85M answers11.2% of the content

Medical Readiness

Disease, mechanism, science
77.7%→85.8%
initial knowledge
Initial gaps1.06M
Final gaps676k
Gaps reduction rate−36.1%
28.5k test questions
11.4M answers27.2% of the content

Clinical Readiness

Clinical evidence, guidelines, cases
73.2%→83.0%
initial knowledge
Initial gaps1.22M
Final gaps777k
Gaps reduction rate−36.5%
36.2k test questions
11.3M answers25.3% of the content

Customer Readiness

The right patient and customer profile
79.1%→86.6%
initial knowledge
Initial gaps146k
Final gaps93.6k
Gaps reduction rate−35.9%
4.7k test questions
1.75M answers3.9% of the content

Conversation Readiness

Objections and the sales argument
75.8%→84.4%
initial knowledge
Initial gaps238k
Final gaps153k
Gaps reduction rate−35.6%
5.5k test questions
2.54M answers5.3% of the content

In pharma, initial clinical knowledge is 70.0%; in medtech, 67.8%. Market, conversation and customer insight are a fifth of the content today, so they rest on fewer questions.

Benchmark

Where would your team land?

Move the slider to your team’s first-attempt accuracy on its own product and science. The band shows where Life Sciences programs started.

First-attempt accuracy
70%
Median 75.4%Bottom 10%Top 10%
50%60%70%80%90%100%

Within the program, the median team gains 7.5 points. The top quarter gains 12.9 or more.

Diagnosis

When people fail, they fail the same way.

Every wrong answer is recorded with the option the person chose. Across 243,328 questions asked in programs with five or more errors, one wrong option collects three out of four mistakes. A question used in several programs counts once per program.

75.2%of errors go to the same wrong answer, on average
58.8%of these questions have one wrong option taking 70% or more of the errors
14.2%of these questions have every error on the same wrong option
Illustrative example
A physician asks how long the treatment takes to show its full effect. What do you answer?
Two weeks75% of wrong answers
Six to eight weekscorrect
Three months25% of wrong answers

A concentrated error pattern identifies a specific point to address with targeted content, followed by a new knowledge check.

Share of these questions by how concentrated their errors are
18.9%50–60%
22.4%60–70%
16.9%70–80%
19.8%80–90%
22.1%90–100%
Errors taken by the most chosen wrong answer
Priority

Half the risk sits in 7% of the questions.

Each question is weighted by what a wrong answer puts at stake in front of a customer. Sorted from highest to lowest, the risk is anything but evenly spread.

You do not need to fix everything. You need to know which 7%, and that is what the data shows.

Stakes

Higher stakes do not make people more careful.

Each question is classified by the real consequence of getting it wrong. Questions with clinical or safety risk are failed at first attempt almost as often as any other.

Loss of credibility with the customer
66.0% of questions
24.0%failed at first attempt
Clinical or patient safety risk
6.0% of questions
22.9%failed at first attempt
Regulatory or compliance risk
4.3% of questions
22.5%failed at first attempt
Lost commercial opportunity
2.9% of questions
23.5%failed at first attempt
Internal inefficiency
4.7% of questions
25.7%failed at first attempt
Internal context only
16.0% of questions
23.5%failed at first attempt
What it measures

Strong on product and science. Honest about the rest.

68%

of questions cover product, science, clinical evidence and use protocol: what a field team must know before it enters a consultation.

80%

ask the person to recall a specific fact. It is the layer that moves most when a program reinforces the same questions.

3%

of questions, the ones clients reuse most, carry a quarter of all answers. Clients keep asking what matters to them.

See where your field stands.

In 30 minutes we show you sample questions for your therapy area and how your team would be measured against this benchmark.

Book a demoHow we measure

Source. Atrivity platform records for Life Sciences clients, extracted on 5 August 2026, with campaigns up to 3 August 2026: 438,290 question-in-campaign records, 56,564,235 answers, 231,333 questions, 3,847 campaigns in 19 languages, September 2014 to August 2026. Direct measurement on the full set, no sampling or simulation. Clients are not named.

Scope. Life Sciences only: pharma, medtech and diagnostics, medical and infant nutrition, animal health, dental care and consumer health. Figures elsewhere on this site that cover all industries or other samples can differ.

Readiness. Accuracy is measured over the people who answered each question. First and last attempt are the same people answering the same questions within the same campaign. Industry figures are campaign medians; 76.1% and 84.7% are pooled over every answer.

Readiness axes. Each field-facing question (73% of records) is assigned automatically to one axis from its topic classification: customer profile to Customer, objections and the sales conversation to Conversation, competition, positioning and access to Market, clinical evidence, guidelines and cases to Clinical, mechanism and disease to Medical, and product, dosing, use and safety to Product. Axis figures are pooled over every answer.

Errors. Concentration is computed on questions with at least five wrong answers at first attempt and two wrong options. Gaps closed: a gap counts as closed when the wrong answer at first attempt becomes correct by the last attempt within the campaign.

Risk weighting. Each question is classified by the consequence of a wrong answer, weighted from 16 (clinical risk) to 0 (internal context). It is the one judgement convention in the analysis; the risk curve excludes internal, corporate and general content.

Limits. The record holds how many people answered and failed each question, not who. It does not follow individuals across programs, so it makes no claim about forgetting between campaigns. Market, conversation and customer insight together represent about one fifth of the axis-classified content.