If you are building a business case for a customer academy, you will find the same dozen numbers on every vendor blog in this category. Revenue up 6.2%. Retention up 7.4%. Support costs down 15.5%. Lifetime value up around 34.6%.
They get repeated because they are useful. They are also, in most cases, repeated without a source, a sample size, or a date, and a decimal point makes a number look measured whether or not it was.
This page lists the figures you will actually encounter, and what you need to know before you put one in a slide your CFO will read.
The numbers most often cited
Revenue and retention lift. Formalized customer education programs are associated with roughly a 6.2% increase in revenue and a 7.4% increase in customer retention. This pair travels together across dozens of pages, which is a clue: they come from the same original study, not from many independent ones.
Return on investment. Somewhere between 90% and 96% of organizations report recouping their customer education investment, depending on which write-up you read. Note the phrasing in every version of this stat. It is organizations reporting, which is a survey of opinion about ROI, not a measurement of ROI.
Product adoption. An average increase of about 38.3% in adoption for products targeted by training. Adoption is defined differently by almost every company that measures it, so this one is directionally interesting and operationally close to meaningless without the definition attached.
Support deflection. Around a 15.5% cut in support costs and roughly a 16% drop in inbound questions. Of the commonly cited figures, this family is the most likely to be measurable in your own instance, because ticket volume is something you already count.
Customer satisfaction. Roughly a 26.2% improvement in CSAT among trained customers. Carries the standard selection problem below.
Lifetime value. Around a 34.6% rise for customers who complete training.
Spend. Businesses have been reported as planning to nearly triple customer education spending across 2024 to 2026.
Three questions to ask before you use any of them
Who was surveyed, and by whom? Most of these figures originate in vendor-run industry surveys. That does not make them false. It does mean the respondents are companies that already invested in customer education, which is not a neutral sample.
Is it a measurement or a self-report? "96% of organizations recouped their investment" is a self-report. Nobody audited their books. Treat self-reports as evidence about sentiment.
Does the causation run the way the sentence implies? This is the one that matters most and the one nobody mentions. Customers who complete training have higher lifetime value. Customers who are getting value from a product are also the ones most likely to finish a course about it. Both are true at once, and no roundup on page one of this search separates them.
That last point is not a reason to ignore the numbers. It is a reason to expect a smaller effect than the headline, and to build your case on the mechanism rather than on the multiplier.
The stat that would be worth having, and nobody publishes
Every figure above measures completion. None of them measure retention of what was learned, which is the thing the training was for.
That gap is not an accident. Completion is trivial to log and retention is not. An LMS knows who finished a module. Very few know whether the person could still perform the task six weeks later, because measuring that requires testing them again after a delay, and most systems have no reason to.
The research on the underlying question is much older and much sturdier than any of the vendor surveys. Spaced retrieval beats massed study for durable recall. Successive relearning, meaning retrieve to a criterion and then relearn on a schedule, is among the best-supported protocols in the learning-science literature, and it has been replicated for decades rather than surveyed once.
If you are building the business case, that literature is stronger ground than a 6.2% from a vendor deck, and it is the ground your program actually stands on.
Numbers you can generate yourself in a quarter
The most defensible statistics in your business case will be the ones with your own company's name on them. Four that most teams can produce inside one quarter:
Ticket deflection on a specific topic. Pick the single most common support question. Build one lesson for it. Compare ticket volume on that topic before and after, for trained and untrained accounts. This is the cleanest natural experiment available to you.
Time to first value. Median days from signup to the first meaningful action, split by whether the account touched onboarding education.
Delayed retrieval, not completion. Ask the same five questions six weeks after a module and record the score. Almost nobody does this, which is why almost nobody knows whether their training worked.
Expansion rate by training exposure. Split accounts by education engagement and compare expansion. Watch the causation direction here as carefully as you would want a vendor to.
The first time you run the six-week check, the number will probably be worse than you expect. That is the finding. A program that knows its real retention figure is in a better position than one quoting somebody else's 38.3%.
How Elite Recall fits
Elite Recall exists because of the gap in the middle of this page. It runs spaced repetition and staged retrieval underneath branded learning programs, so the thing being measured is whether a learner can still produce the answer later, rather than whether they clicked to the end of a module.
The review queue sorts a cohort by what each person is closest to forgetting, which means the retention number is a byproduct of how the system works instead of a separate research project.
If you want the mechanism rather than the pitch, start with why employees forget training.