Every figure on this page links to the organization that produced it, and each was re-checked against that source before publishing. What organizations spend on employee training and how many hours employees get, how much of it runs online, in virtual classrooms and on video, what randomized trials say video does for learning, how accurate automatic captions are on instructional video, which training the law requires, who makes it, what the companies selling it report, and where the popular training numbers came from.
U.S. organizations with 100 or more employees spent an estimated $102.8 billion on training in 2025, or $874 per learner, by Training magazine's survey. (Training magazine)See details
Employees got 40 hours of training a year in Training magazine's 2025 survey but only 13.7 formal learning hours in ATD's, a 2.9-fold gap that comes from different samples and definitions. (Training magazine; ATD; our calculation)See details
That is about 46 minutes of training a week per employee on Training magazine's count, or 16 minutes on ATD's. (Our calculation)See details
ATD's formal learning hours per employee fell 61% from 34.7 hours in 2019 to 13.7 hours in 2024. (ATD; our calculation)See details
34 percent of U.S. training hours ran online or on a computer in 2025 and 24 percent in a virtual classroom, down from a 37 percent peak in 2021. (Training magazine)See details
92 percent of organizations deliver at least some compliance training online, and 45 percent deliver all of it online. (Training magazine)See details
Across 105 randomized trials, video in place of other teaching improved learning slightly (g = 0.28: the average student scored above 61% of the comparison group), and video added to teaching improved it strongly (g = 0.80). (Noetel et al., Review of Educational Research; our conversion)See details
In health-professions training, online instruction beat no instruction by a wide margin (effect size 1.00 on knowledge) but beat classroom teaching only slightly (0.12). (Cook et al., JAMA)See details
On recent educational YouTube videos, automatic captions missed a median 1.8% of words but 15.4% of names and other capitalized terms, 5.1 times the rate for other words. (Our measurement)See details
California requires 2 hours of harassment-prevention training for supervisors and 1 hour for other staff every two years; Illinois requires training once a year for every employee. (California Civil Rights Department; Illinois General Assembly)See details
The BLS counted 458,300 training and development specialist jobs in May 2025, with a median wage of $69,280. (U.S. Bureau of Labor Statistics)See details
LinkedIn's widely quoted '94% of employees would stay longer' comes from an online survey of its own members in 2017, with no sample size published. (LinkedIn Learning)See details
How much do organizations spend on employee training?
About $100 billion a year in the United States, by the training trade press's count. Training magazine's 2025 Training Industry Report puts U.S. training expenditures at $102.8 billion in 2025, up 4.9% on 2024 Training magazine, 2025. The figure covers organizations with 100 or more employees and is a projection: the magazine multiplies the average budget in its online survey by a weighted universe of 152,572 U.S. companies drawn from Dun & Bradstreet. Training staff payroll made up $64.7 billion of the total, outside products and services such as content, technology and consultants $16 billion, and travel, facilities and equipment $22.1 billion.
Per person, the same survey found organizations spent $874 per learner in 2025. Large corporations spent the least per learner, $468, and small companies the most, $1,091. The average organization put $290,987, 13% of its training budget, into learning tools and technologies, and gave its two largest budget shares, 13 percent each, to mandatory compliance training and to management and supervisory training. 41 percent of organizations said their training budget grew in 2025, and 41 percent named a lack of resources or staff as their biggest training challenge. Training magazine sells advertising and events to training vendors, and its 2025 page does not say how many organizations answered.
The Association for Talent Development, the profession's trade association, counts differently. Its 2025 State of the Industry report, built from data that 539 organizations chose to submit, found direct learning spending of $1,254 per employee in 2024, $29 less than in 2023, and a cost of $165 per learning hour used ATD, 2025. The median organization had 14 talent development staff. Five years earlier, 283 organizations reported $1,308 per employee for 2019, with 66 percent of direct learning spending going to internal services, 24% to outside suppliers and 10% to tuition reimbursement ATD, 2020.
Training spending per person, counted four ways. Each source uses its own population, year and currency, so the figures are not directly comparable; that is the point of the table.
Somewhere between 13.7 and 40 hours a year, depending on who counts. Training magazine's survey found employees received 40 hours of training on average in 2025, down from 47 hours in 2024 Training magazine, 2025. Organizations reporting to ATD counted far less: 13.7 formal learning hours per employee in 2024, down from 17.4 in 2023 ATD, 2025. Training magazine's average is 2.9 times ATD's, by our calculation. The gap comes from the samples and the definitions: Training weights its survey to U.S. companies with 100+ employees and counts all training hours, while ATD counts only formal, standalone learning at the organizations that send it data.
Spread over a year, 40 hours is about 46 minutes of training a week per employee, and 13.7 hours about 16 minutes a week, by our calculation. ATD's formal hours have fallen fast: from 34.7 hours per employee in 2019 ATD, 2020 to 13.7 in 2024, a drop of 61%. ATD suggests some of that time has moved to informal learning such as mentoring and coaching, which its hours do not count. In the UK, employers provided 6.0 days of training per person trained in 2022.
Training magazine's annual survey, 2019 to 2025: U.S. organizations with 100+ employees. Each edition is re-weighted to a new Dun & Bradstreet universe, which grew 15% in the 2021 edition alone, so year-to-year changes mix real change with a changing base.
Sources for the table: Training magazine's Training Industry Reports for 2019, 2020, 2021, 2022, 2023, 2024 and 2025. The 2021 edition says a major factor in that year's rise was a 15 percent jump in the number of companies in the database it projects onto Training magazine, 2021.
How is workplace training delivered, and how much of it is video?
About a third of U.S. training hours run online or on a computer, and a quarter in a live virtual classroom. In Training magazine's 2025 survey, 34 percent of training hours were delivered through online or computer-based technologies, 28 percent by an instructor in a classroom, 24 percent by virtual classroom or webcasting, and 22 percent through blended learning Training magazine, 2025. Mobile devices carried 6 percent of hours, up from 3%, and artificial intelligence tools 2 percent. The categories overlap, so they add up to more than 100%. Neither Training magazine nor ATD counts recorded video as a delivery method of its own: on-demand video sits inside "online or computer-based", and live video inside "virtual classroom".
The virtual classroom's share shows the pandemic in one line. It was 15.4 percent of hours in 2019, peaked at 37 percent in 2021 and has eased back to 24 percent. Online and computer-based delivery rose from 29.6 percent to a steady third and stayed there.
Tools follow the same pattern. 89 percent of organizations used a learning management system in 2025, and 77 percent used virtual classroom, webcasting or video broadcasting technology, down from 88 percent in 2021 Training magazine, 2021 and 85 percent in 2023 Training magazine, 2023, against 75 percent in 2019 Training magazine, 2019. Use of artificial intelligence as a learning technology jumped to 37 percent from 25%, and 42 percent used rapid e-learning authoring tools. Looking ahead, 25 percent of organizations expected AI training to get more money than the year before, second only to management training, while only 3 percent said that of diversity, equity and inclusion training, down from 24%.
Required training is the most likely to be online. 92 percent of organizations delivered at least some mandatory or compliance training online in 2025, and 45 percent delivered all of it online. Onboarding was among the least likely programs to be online, at 56 percent of organizations doing at least some of it that way, and executive development the least, at 48 percent. In the UK, 67% of employers that trained staff in 2022 used online training or e-learning, up from 51% in 2017.
Does video improve learning? What controlled studies show
Yes, modestly when it replaces other teaching and strongly when it is added to it, in the largest review of randomized trials. Noetel and colleagues pooled 105 randomized trials with 7,776 students in a 2021 systematic review in the Review of Educational Research Noetel et al., 2021. Swapping video for existing teaching methods produced a small improvement, an effect size of g = 0.28. Adding video to existing teaching produced a large one, g = 0.80. In plain terms, the average student taught with video in place of other methods scored above about 61% of the comparison group, and the average student given video on top of other teaching above about 79%, by our conversion. The trials studied college students, not employees, and the authors note possible experimental and publication bias.
For working professionals, the best evidence comes from health care. A 2008 JAMA meta-analysis of 201 studies of Internet-based instruction for physicians, nurses, pharmacists and other health professionals found a large effect against no instruction at all: 1.00 for knowledge across 126 studies, about the 84% mark, and 0.85 for skills Cook et al., JAMA. Against classroom or other non-Internet teaching, the edge almost disappeared: 0.12 for knowledge, about 55%, and 0.09 for skills, which was not statistically significant. Results varied widely between studies, with heterogeneity of 79% or more in every analysis. The lesson for training teams is that delivering the same content online works about as well as delivering it in person; the gain comes from training people who would otherwise get none.
Behavior modeling, in which trainees watch someone perform a skill, often on video, and then practice it, has its own meta-analysis of 117 studies in the Journal of Applied Psychology Taylor, Russ-Eft and Chan, 2005. Its effects were largest on learning, smaller on job behavior and smallest on business results, and transfer to the job was strongest when trainees set goals and their managers were trained too.
On length, the platform evidence points to short. Analyzing several million video sessions on edX, the average time learners spent on a course video topped out at about 6 minutes, however long the video was, and learners spent around 3 minutes on videos longer than 12 minutes edX, 2013. Those learners were taking open online courses, not mandatory workplace training.
How accurate are automatic captions on instructional video, and how fast do instructors talk?
Close, but not close enough for names. We re-cut the public dataset from our YouTube auto-caption accuracy study, which scored YouTube's automatic captions against the channels' own human captions on 299 recent videos from six educational channels: MIT OpenCourseWare, Stanford, Khan Academy, CrashCourse, TED and NASA. In the median video, automatic captions missed or got wrong 1.8% of words. Only 74 of 299 videos reached the 99% level, 1% of words or fewer missed. The errors cluster on the words training depends on: names and other capitalized terms, such as product names, places and people, were missed 15.4% of the time, 5.1 times as often as other words. A compliance or product course full of brand names and acronyms needs its captions checked by a person.
The gap between channels is wide. Khan Academy's scripted, narrated lessons had the fewest misses, a median 1.0%, and MIT OpenCourseWare's recorded classroom lectures the most, 2.9%. The median video in the set was narrated at 157 words a minute. Khan Academy's lessons ran faster, at 169 words a minute in videos with a median length of 3.7 minutes, and MIT's recorded lectures slower, at 136.5 words a minute in sessions that typically run over an hour. These are educational channels, not corporate training libraries, so treat them as a benchmark for recorded instruction in general. The full method is on the study page.
Which workplace training does the law require?
Some training is required by law. In the United States, federal safety rules fix minimum hours and refresher intervals for specific jobs, and several states require harassment-prevention training for most employers. This table compiles the requirements from the regulations and statutes themselves, as published on October 30, 2026. It is a summary, not legal advice.
Selected U.S. training mandates with numbers in the rule. Figures link to the regulation, statute or enforcing agency.
two hours of training, new hires within six months, refreshed at least every ten yearsConnecticut General Assembly
Delaware, 19 Del. C. §711A
Employers with 50 or more employees in Delaware
Interactive training for new employees within a year of hire, then every 2 yearsDelaware General Assembly
Maine, 26 M.R.S. §807
Workplaces with 15 or more employees
Training for all new employees within one year of hire Maine Legislature
California's rules show how these mandates create recurring demand. Every two years, each covered employee needs a fresh session, new supervisors and new staff must be trained within six months of starting, and the next statewide deadline is January 1, 2027, according to the state's Civil Rights Department. Mandates like these help explain why compliance training ties with management training for the largest share of the average training budget, and why it is the program most often delivered online.
Who makes workplace training?
In the United States, mostly training and development specialists, who design and deliver training programs. The Bureau of Labor Statistics counted 458,300 wage and salary jobs in the occupation in May 2025, earning an annual mean wage of $75,550 and a median of $69,280BLS OEWS. The lowest-paid tenth earned under $38,760 and the highest-paid tenth over $123,250. That is about 2.9 specialist jobs for every 1,000 jobs in the economy, measured against all 155,495,730 jobs the survey covers BLS OEWS, by our calculation.
Above them, 48,050 training and development managers earned an annual mean wage of $148,370 and a median of $133,000, with the top tenth over $233,600BLS OEWS. Managers' median pay was 1.9 times specialists'. A related occupation, instructional coordinators, held 227,760 jobs with a median wage of $77,440, but most of them work in schools and colleges rather than in companies BLS OEWS. These counts exclude the self-employed, so freelance instructional designers and video editors who build courses on contract are not in them.
Training and development specialists in five large states, May 2025 (BLS OEWS)
The tax code also shapes what employers pay for. Under section 127 of the Internal Revenue Code, an employer can exclude up to $5,250 a year of educational assistance per employee from the employee's taxable wages, and the Internal Revenue Service notes that 2025 legislation made that exclusion permanent IRS Publication 15-B. This is not tax advice.
What do the companies selling workplace learning report?
Their annual reports to the U.S. Securities and Exchange Commission show how large the business of on-demand corporate learning has become, and which way it is moving. Each figure below is the company describing itself in a filing, not an independent count.
Udemy's enterprise segment, Udemy Business, brought in $524.1 million in fiscal 2025, out of total company revenue of $789.8 millionUdemy 10-K. Coursera had more than 1,700 paid enterprise customers buying through its sales team at the end of 2025 Coursera 10-K, and its enterprise segment generated $255.3 million. Skillsoft served over 3,000 enterprise customers and 59 million registered learners in the fiscal year ended January 31, 2026, with a library of more than 431,000 learning assets, on revenue of $512.7 millionSkillsoft 10-K. Docebo's learning platform was used by nearly 3,500 companies Docebo annual information form, with annual recurring revenue of $238.1 million and an average contract value of $66,500 per customer at the end of 2025 Docebo results release.
Video platforms report the same customers. Kaltura, which sells enterprise video software used for training and internal communications, says its customers include about 30% of Fortune 100 companies Kaltura 10-K. LinkedIn says about 119 hours of learning content are consumed on its platform every minute LinkedIn.
How much workplace training happens outside the U.S.?
In the European Union, two in three businesses train their staff. Eurostat's Continuing Vocational Training Survey found 67.4 % of EU enterprises with 10 or more persons employed provided training in 2020, down from 70.5 % in 2015 as the pandemic disrupted courses Eurostat. Size matters: 92.8 % of large enterprises trained, against 63.5 % of small ones. The share ranged from 17.5% in Romania to 96.8 % in Latvia. 42.4 % of people employed in these enterprises took part in a training course, from 82.8 % in Czechia down to 11.8% in Greece.
Courses cost EU training enterprises 1,441 PPS (purchasing power standard) per participant in 2020, or 64 euros per training hour, one euro more than in 2015. The most common brake on training was time: 36.8 % of training enterprises said staff lacked time to take part. Among enterprises that trained no one, 74.7 % said their workforce's existing skills already met their needs.
In the United Kingdom, 60% of employers funded or arranged training in the 12 months before the Department for Education's 2022 Employer Skills Survey Department for Education. Of the employers that trained, 67% used online training or e-learning, and among the largest establishments, with 250 or more employees, 94% did. UK employers spent £53.6 billion on training, 7.7% less in real terms than in 2017, or £1,780 per employee, and provided 6.0 days of training per person trained. The survey interviewed 59,486 employers in England alone. Figures from different countries use different populations, years and definitions of training, so compare them with care.
What the popular training numbers really say
Four figures recur in training-video and e-learning vendors' blogs. None of them measures what it is used to claim.
"Viewers retain 95% of a message in video, 10% in text"
It is often credited to Forrester Research or to unnamed studies. Its earliest datable publication is a 2013 statistics list by Insivia, a marketing agency, which stated that viewers retain 95% of a message when they watch it in a video Insivia, 2013 list. Insivia now writes on its own site that marketers have repeated for years that viewers retain 95% of a video message Insivia, and says the figure came from a small in-house survey of around 200 B2B buyers in the late 2000s, not from a peer-reviewed study. Our video advertising statistics page traces it in full. Use instead: the randomized-trial evidence above, where video in place of other teaching had a small positive effect, g = 0.28.
"Learners forget 70% of training within 24 hours"
This is Hermann Ebbinghaus's forgetting curve, carried a long way from its source. In 1885 Ebbinghaus reported experiments on one subject, himself, learning series of 13 nonsense syllables. He measured memory as savings: how much less work it took to relearn a list. After 24 hours the saving was 33.7 per cent; in his summary, about one third was remembered after a day, and an hour after learning he already had to spend one half the original work again Ebbinghaus, 1885. The remaining 66.3% is close to the "70% forgotten" figure, by our calculation, but it describes the effort to relearn meaningless syllables, not the share of a training course an employee forgets. A 2015 replication at the University of Amsterdam also used one subject, who spent 70 hours learning and relearning lists Murre and Dros, PLOS ONE. Use instead: a course's own assessment results, or the controlled-study evidence above.
"94% of employees would stay longer if the company invested in their learning"
This comes from LinkedIn, which sells online learning. Its 2019 Workplace Learning Report says 94% of employees would stay at a company longer if it invested in their learning and development LinkedIn, 2019. The report's own source note says the figure comes from an online survey of employed LinkedIn members conducted in 2017, with no sample size, question wording or margin of error published. The report's main surveys, of more than 1,200 talent developers and over 2,100 workplace learners, are separate. It is a stated intention from a self-selected group, not measured retention. Use instead: describe it as a LinkedIn member survey from 2017, or leave it out.
"We remember 10% of what we read and 90% of what we do": the learning pyramid
The pyramid is usually credited to the NTL Institute. NTL has said it no longer has, and cannot find, the original research behind the numbers NTL Institute, via Learning Accelerators, and a 2018 peer-reviewed study found versions of the pyramid circulating for more than 160 years, concluding the models did not come from empirical research Letrud and Hernes, via ERIC. The full trace is on our online course video statistics page.
We started from the questions people search for about training video statistics and looked for the organization that actually produced each answer. Every figure on this page sits in an evidence ledger with its source, the exact table or paragraph it came from, the measurement period, the population measured, and the date we checked it. This edition holds 193 figures from 29 publishers.
On October 30, 2026, each of the 173 sourced figures on this page was re-fetched with the fetch cache cleared and matched against the recorded wording: 152 matched against the publisher's page or file or a dated Internet Archive capture, and the 20 Bureau of Labor Statistics wage and employment figures against the BLS data API. The remaining one, from a PubMed Central article that answered the script with a bot check on that pass, matched in a headless browser the same night. 69 figures cite Internet Archive captures, mostly Training magazine's annual reports, because trainingmag.com did not respond to automated requests that day, plus two ATD blog summaries and an edX blog post that has since moved. We left out what we could not read or check: ATD's full State of the Industry reports, which are sold, so only ATD's free summaries are used; California's statute site and New York's training pages, which refused automated requests, so California is cited through its Civil Rights Department's own guide and New York is not in the table; the Arthur (2003) and Sitzmann (2006) meta-analyses of workplace training, whose abstracts were not openly readable; and Bersin by Deloitte's widely quoted '24 minutes a week' figure, for which we found no original copy. The 20 figures marked as our calculations are listed with their formulas below.
We only cite the producer: government agencies, regulators, company filings and official posts, and the survey owner's own report. We do not cite statistics roundups, Statista-style aggregators, or salary sites without a public method. We did not bypass paywalls or sign-up forms; where a full report was gated, we used only the figures its publisher states publicly.
Figures marked "our calculation" are arithmetic on the cited numbers, and anyone can reproduce them:
median of missed_pct over all 299 rows of videos.csv (1.750). missed_pct = (substitutions + deletions) / words in the human track. Matches the study's published headline (98.2% of words right in the median video)
SPD-008 (40 hours, 2025) / SPD-065 (13.7 hours, 2024) = 2.92. Training counts all training hours at weighted U.S. organizations with 100+ employees; ATD counts formal learning hours at 539 self-selected organizations
This edition contains no Mark Studios customer data.
Limitations
The training industry counts itself. Training magazine sells advertising and events to training vendors and does not publish its 2025 sample size; ATD is the profession's trade association and its figures come from organizations that chose to send data; LinkedIn, Udemy, Coursera, Skillsoft, Docebo and Kaltura sell learning products.
The two U.S. surveys measure different things. Training magazine weights its survey to U.S. organizations with 100+ employees and counts all training hours; ATD counts formal learning hours at the organizations that report to it. Neither is a government statistic, and no U.S. agency has published a current count of employer training spending.
Training magazine's series is re-weighted every year. Each edition projects onto a new Dun & Bradstreet universe of companies, so year-to-year changes in total spending mix real change with a changing base.
No survey counts video on its own. On-demand video sits inside "online or computer-based" delivery and live video inside "virtual classroom", so this page cannot say what share of training is video.
The controlled evidence comes mostly from students and health professionals. The 105 trials in the largest review studied college students, and the largest online-learning meta-analysis for working adults covers the health professions.
Our caption and pace figures come from educational channels, not corporate training libraries: 299 videos from six YouTube channels, scored against those channels' own human captions.
Government job counts exclude the self-employed, so freelance instructional designers and editors are not in the BLS figures.
Rules are summaries, not legal advice, and cover only selected federal rules and states; each is dated to the version we read.
Figures from different countries are not comparable. Eurostat counts enterprises with 10+ persons employed in 2020, a pandemic year; the UK survey counts establishments with 2+ employees in 2022.
How to cite this page
For a single figure, cite the original publisher named next to it, and link here if the verification trail is useful to your readers. Each figure has its own anchor, so you can link straight to it, for example https://www.markstudios.com/research/training-video-statistics#DRV-003. To cite the page as a whole:
Mark Studios. "Training Video Statistics." Last verified Oct 30, 2026. https://www.markstudios.com/research/training-video-statistics
APA
Mark Studios. (2026, October 30). Training video statistics. https://www.markstudios.com/research/training-video-statistics
MLA
“Training Video Statistics.” Mark Studios, 30 Oct. 2026, www.markstudios.com/research/training-video-statistics.
Chicago
Mark Studios. “Training Video Statistics.” Last modified October 30, 2026. https://www.markstudios.com/research/training-video-statistics.
BibTeX
@misc{markstudios_training_video_statistics_2026,
author = {{Mark Studios}},
title = {{Training Video Statistics}},
year = {2026},
month = oct,
url = {https://www.markstudios.com/research/training-video-statistics},
note = {Last verified 2026-10-30}
}
Our calculations and the evidence table are free to reuse with attribution and a link to this page. Each underlying figure remains the claim of its original publisher, and some publishers ask for a link when their figures are reused.
Pipeline and Hazardous Materials Safety Administration (U.S. Department of Transportation), 49 CFR 172.704 Training requirements (1992-05-15). Accessed 2026-10-30.
We re-verify every figure quarterly and when a major source publishes a new edition. The next scheduled review is January 2027. Corrections go to contact@markstudios.com.
Oct 30, 2026: First published with 39 figures from 6 publishers, each checked against its original source.
Sep 28, 2026: weva.com no longer responds, so the three WEVA survey figures now link to Internet Archive captures of the same 2003 and 2006 releases, re-checked word for word.
Frequently asked questions
How much do companies spend on training per employee?
It depends on who counts. Training magazine's 2025 survey of U.S. organizations with 100 or more employees found $874 per learner. The Association for Talent Development's 2025 report, based on 539 organizations that sent data, found $1,254 per employee in 2024. UK employers invested £1,780 per employee in 2022, according to the Department for Education.
How many hours of training do employees get a year?
Training magazine's 2025 survey found an average of 40 hours a year, about 46 minutes a week by our calculation. ATD counted 13.7 formal learning hours per employee in 2024, down from 34.7 in 2019. The two surveys use different samples and definitions.
Is video training more effective than classroom training?
Slightly, on average, and much more when it is added to classroom teaching rather than replacing it. A 2021 review of 105 randomized trials found an effect size of g = 0.28 when video replaced other teaching methods and g = 0.80 when video was added to them. The trials studied college students. For health professionals, a 2008 meta-analysis found online instruction only slightly better than classroom teaching for knowledge, an effect size of 0.12.
How long should a training video be?
Shorter than most lectures. On edX, the average time learners spent on a course video topped out at about 6 minutes however long the video was, and they spent around 3 minutes on videos longer than 12 minutes. Khan Academy's lessons in our caption study had a median length of 3.7 minutes and ran at 169 words a minute.
Do employees really forget 70% of training within 24 hours?
No study of workplace training shows that. The figure traces to Hermann Ebbinghaus's 1885 experiments on himself with nonsense syllables: after 24 hours, relearning a list saved 33.7% of the original work, leaving 66.3% to redo. That measures the effort of relearning meaningless syllables, not how much of a course an employee forgets.
Are automatic captions good enough for training videos?
For general words, nearly: on 299 recent educational YouTube videos, automatic captions missed a median 1.8% of words. They missed 15.4% of names and other capitalized terms, which is where product names, acronyms and people's names in training content sit, so captions for training should be checked by a person.
How often is this page updated?
Every figure is re-checked against its original source each quarter, and sooner when a major source publishes a new edition, such as Training magazine's Training Industry Report each November, ATD's State of the Industry each spring, or the BLS occupational wage estimates each spring. The Last verified date changes only after that check, and each change is recorded in the change log.
