Blended Learning Explained

By Brian Tr · Updated Oct 7, 2026

Blended learning is a way of teaching that combines online instruction with in-person or live instruction, and connects the two on purpose. The online part usually carries content and practice that learners can take at their own pace. The live part, whether in a classroom or a video call, carries discussion, feedback and application.

The term gets stretched to cover almost anything with a screen in it, which makes it slippery. This article pins down what it means, lays out the standard models, looks honestly at the research, and then turns to the question that matters if you build or sell learning products: what does blending do to your product, your delivery costs and your pricing?

Key Facts

• The Christensen Institute (then the Innosight Institute) defines blended learning as a formal education program in which a student learns at least in part through online delivery of content and instruction, with some element of student control over time, place, path or pace, and at least in part at a supervised brick-and-mortar location away from home (Staker and Horn, Classifying K-12 Blended Learning, 2012).
• The same paper says a blended program's modalities are usually connected: what students learn online informs what they learn face to face, and the reverse (Staker and Horn, 2012).
• A US Department of Education meta-analysis found an average effect size of +0.35 for studies comparing blended instruction with purely face-to-face instruction, against +0.14 for studies comparing purely online with face-to-face (US Department of Education, 2009).
• The same report warns that blended conditions often included additional learning time and instructional elements the control groups didn't get, so the advantage should not be attributed to the media itself (US Department of Education, 2009).
• The Christensen Institute's current model list has seven items under four families: Station Rotation, Lab Rotation, Individual Rotation, Flipped Classroom, Flex, A La Carte and Enriched Virtual (Blended Learning Universe, Christensen Institute).

What blended learning is, and what it isn't

Two ideas sit inside the definition above, and both are worth keeping.

First, the online part has to be real instruction. Staker and Horn specify online delivery of content and instruction, and add that the learner has some control over time, place, path or pace. That phrase is there to separate blended learning from technology-rich teaching, where a teacher projects slides or students type on laptops but nothing is actually delivered online with learner control (Staker and Horn, 2012).

Second, the parts should connect. A course that offers a video library on one side and a workshop on the other, with no link between them, is two products sold together. The paper treats the connection between modalities as a common feature, and notes some researchers think it should be written into the definition itself.

Corporate trainers use the idea more loosely: self-paced e-learning plus instructor-led sessions plus on-the-job practice. The core logic travels fine. What changes is who controls the schedule and where the "live" part happens, which is often a video call rather than a room.

The authors also left quality out of the definition on purpose: like a hybrid car, blended learning can be efficient or a clunker and still be what it is (Staker and Horn, 2012). A blend is a design pattern, not a guarantee.

The common models

The best-known taxonomy comes from Michael Horn and Heather Staker's work for the Innosight Institute, now the Christensen Institute. Their 2012 paper sorted programs into four families: Rotation, Flex, Self-Blend and Enriched-Virtual, with four versions of Rotation underneath (Staker and Horn, 2012). The Institute's current model page uses the label "A La Carte" where the 2012 paper said "Self-Blend", and lists seven models in all (Blended Learning Universe).

The descriptions above follow the Institute's own wording closely. A few details are easy to blur.

Station Rotation differs from Individual Rotation because every student rotates through every station, not only those on a custom schedule. Flipped Classroom is classed by Staker and Horn as a rotation, since students rotate on a fixed schedule between teacher-guided practice at school and online instruction elsewhere. They also stress that the primary delivery of content has to be online, which separates a true flip from ordinary online homework practice (Staker and Horn, 2012). Enriched Virtual differs from the flipped classroom because learners seldom attend the physical site every weekday, and it differs from A La Carte because it describes a whole program, not one course (Staker and Horn, 2012).

The taxonomy was built from school programs, and the authors say the models describe particular programs, not entire institutions. Many schools run more than one at once. It's the same in companies, where a single academy might run a flipped model for onboarding and a flex model for continuing education.

Corporate learning versus school and university use

The models above grew out of classrooms, but the idea has long been used in adult and professional training. The Department of Education's meta-analysis, which we'll turn to next, found that the learners in its studies were about evenly split between college or community college students and graduate students or adults receiving professional training. The most common subject was medicine or health care, with business among the other areas (US Department of Education, 2009).

Still, the settings differ in ways that matter for design and for selling.

That last row connects to a theme running through this collection. A corporate buyer doesn't usually care about the blend as such. They care about whether people can do the job better, which is the territory of proving training ROI. The blend is a means, and a buyer who runs a training needs analysis first will often find the right blend falls out of the gap they've identified.

What the evidence says

The research most people cite is the US Department of Education's meta-analysis of online learning studies, prepared by Barbara Means and colleagues at SRI International. Here is what it did and what it found, with the caveats the authors attached.

The method. The team searched studies from 1996 through July 2008 and screened 1,132 abstracts, ending with 176 studies that used experimental or quasi-experimental designs and measured learning objectively. Of those, 99 had at least one contrast between an online or blended condition and a face-to-face condition. Forty-six studies provided enough data for 51 independent effect sizes, 28 of them purely online versus face to face and 23 blended versus face to face (US Department of Education, 2009).

The headline. Students in online conditions performed better on average than students in face-to-face instruction, with an average effect size of +0.24. The advantage was larger for blended conditions (+0.35) than for purely online ones (+0.14) (US Department of Education, 2009). By Cohen's convention, which the report cites, an effect of .20 is small and about .50 is medium, so these are small to modest effects.

The caveats. The report is unusually direct about them, and they matter more than the headline:

• Blended conditions often included extra learning time and instructional elements that the control groups didn't get. The authors conclude the positive effects "should not be attributed to the media, per se."
• Studies where online learners spent more time on task showed a larger benefit (+0.46) than studies where face-to-face learners spent as much time or more (+0.19).
• Studies in which the curriculum and instruction were identical or almost identical across conditions had smaller effects (+0.20) than those where the conditions differed in several ways (+0.42).
• Many studies had small samples, didn't report retention, or had the same people acting as experimenter and instructor.
• Most of the evidence came from older learners. The mean effect was not significant for the seven K-12 contrasts, and the authors say the number of K-12 studies was too small to support much confidence.

All of those points are in the report's own summary (US Department of Education, 2009). The authors' conclusion is that blended instruction was more effective in the studies reviewed, which gives a rationale for the effort of designing it, but that the studies don't show online learning to be superior as a medium.

Read honestly, the finding is narrower than the way it's often quoted. Blended designs tended to do well because they tended to add time, materials and interaction, and online delivery makes adding time easier. The report notes that online learning is much more conducive to expanding learning time than face-to-face instruction. Nothing in it says a blend of two ordinary lectures would beat one.

One more detail from the review of studies on practices: blended and purely online conditions inside the same study generally gave similar learning outcomes (US Department of Education, 2009). So the strongest comparison in the report is blended against fully face-to-face, and it carries the time-on-task caveat. The report is also from 2009, so it can't speak to the video-call era or to current corporate platforms.

Design choices that decide whether a blend works

If the evidence says design and time matter more than medium, then a blend lives or dies on a handful of choices.

What goes online and what stays live. A common starting rule is to put the parts that scale well online (explanations, demonstrations, knowledge checks) and keep the parts that need people live (practice with feedback, debate, judgment calls, relationships). The flipped model is the cleanest version of that rule.

How tightly the pieces connect. Staker and Horn describe connected modalities as a common feature. In practice that means the live session builds on the online work and uses its results, such as quiz errors that shape the discussion, instead of repeating it.

Who sets the pace. The definition requires some learner control over time, place, path or pace. A fixed cohort calendar gives more accountability, and a flexible path gives more convenience. The trade-off is explored in cohort-based vs self-paced learning, and most blends sit somewhere between the two.

How long each piece runs. Short online units suit busy employees, and microlearning covers where those small units help and where they fall short. The sequencing and objectives side of the job belongs to instructional design basics.

Who delivers the live part. A trainer, a manager, a peer coach or a community. This is also where the cost sits, which brings us to the business side.

What blended learning means for an EdTech company

For founders and operators, blended learning is mostly a question about what you actually sell. It forces three decisions.

Product and services mix

A purely digital product is software. A purely live product is a service. A blend is both, and the two have different economics. The Department of Education report frames it as a cost question too: blended learning, it says, needs to be more effective than face-to-face instruction to justify the additional time and costs it entails, while fully online delivery is attractive on cost and convenience as long as it's as effective as classroom instruction (US Department of Education, 2009). Buyers who've read any of this will ask what the live part adds.

A useful way to plan is to list every element of your program and tag it as software, content, live delivery or support. Then ask which tags are costly to scale. Live delivery scales with the number of facilitators, and content and software scale with almost no extra cost per learner. That asymmetry is why many companies treat the live layer as the premium tier.

Delivery costs

Live sessions bring costs that software companies aren't used to: facilitator time, scheduling, cohort management, venues or video-call logistics, and quality control across trainers. Margins on a blended offer therefore sit between a software margin and a services margin, and they depend on facilitator utilization. The sensible response is to design the live component to be as small and as valuable as the learning needs, and to say what each session is for. For the wider picture of how companies earn from training, see corporate training business models.

Pricing implications

Blended offers are hard to price per seat alone, because a seat in a cohort and a seat on a platform don't cost the same to serve. Common patterns are a platform fee plus a per-cohort or per-session fee for live delivery, or a tiered package where the live layer is an add-on. The choice between per-seat and per-learner pricing gets sharper here, because live capacity is limited and a cohort has a hard maximum size.

The platform question matters too. A blend needs somewhere to host online content, track progress and schedule live sessions, which is where the difference between a system of record and a discovery layer comes in. See LMS vs LXP for how those categories split.

Where blending fits in your segment

Not every segment values a blend equally. Consumer learners often want convenience and may resist fixed live sessions, while enterprise buyers often want the accountability a live element provides. Check which segments your own customers sit in before committing to a model.

Related reading

• Cohort-Based vs Self-Paced Learning
• LMS vs LXP
• Corporate Training Business Models
• What Is a Training Needs Analysis?