One common question is “how long will it take to create this elearning?” It’s important to estimate the effort and time required for different tasks. Because I work for myself, I have to create good time estimates. If I don’t, I either bid low and lose money or bid high and lose a project.
In this post, I walk through my thought process for using industry benchmarks to create a time estimate for elearning development. I published the original version of this post over 10 years ago. In additional to updating links and other revisions, I’m adding notes on how AI may change your time estimates.
Benchmarks
The two primary sources I use for benchmarks are Bryan Chapman’s research and the ATD research.
Of the two sources, I usually go for Chapman’s data, since it’s broken down with more detail. I find it helpful to refer clients to these sources, especially if they think training development should take barely any time. Chapman’s research is from 2010, so it’s an older source. However, it’s still likely the most commonly referenced benchmark in the field.
The ATD research is updated every few years, but it also shows more variability and is trickier to compare. In the 2023 update of the research, Robyn Defelice makes a key point about these benchmarks:
The largest conclusion we can draw comfortably from the data is that development time still varies considerably for each type of learning product—and no matter what, the variables of the training development for each organization greatly differ. We do not advise that you use the results as stand-alone pieces for calculating project estimates because context is necessary to understand respondents’ situations.
–Robyn Defelice
IconLogic also has benchmarks for development, including breakdowns for different tasks. The biggest difference in their benchmark is the time to develop in an authoring tool, which is significantly higher than if you use Chapman’s data alone. This estimate says that Captivate and Storyline development generally take 2 hours per finished minute to produce, or a ratio of 120:1. That’s not including writing a script or recording audio. Chapman’s estimate is closer to 20:1 for the authoring/programming section, so there’s a pretty large discrepancy. That’s partly due to how tasks are classified, but I can’t completely reconcile the numbers. My own data matches IconLogic’s data more closely.
I also track my own time for every project I create so I can compare my actual numbers to the benchmarks. I use a time tracking template that lets me analyze my time on different tasks and projects. That’s the best situation, but it takes time to build up enough data to create your own personal benchmarks. Use the published benchmarks as a starting point, but start tracking your own time too if you haven’t yet. As Robyn noted, there are too many variables for any set of industry benchmarks to be as accurate as your own data.
Applying the Benchmarks
As an example, let’s say a client asks me to convert an existing full day training program to self-paced elearning. The existing training is working well, but the client needs it to scale to a larger audience.
The elearning version will be mostly linear with 25% interactivity but no branching scenarios. A “full day” of training means 6 hours of actual content. The content itself is in pretty good shape; there’s slides, a participant guide, and a facilitator guide, and it’s all fairly complete. The elearning version will not use video. The narration will be generated with an AI voice tool.
(In my original estimate, I specified professional voice over. Even a few years ago, I never used computer-generated voices for anything other than alpha versions of content. Now, they’ve improved enough that more of my clients are shifting to AI voices.)
I’m going to assume this can be compressed to about 3 hours of e-learning. That’s 50% of the original time, which is a standard estimate backed up by research.
This project a Level 2 by Chapman’s study, so the ratio for development is 184:1 (that is, 1 hour of e-learning takes 184 hours to develop). For 3 hours, that’s 184 * 3 or 552 hours total work. That’s the work for everyone on the team, not just me.
Chapman’s study provides this breakdown of tasks and the percentage of time for each (see slide 18).
Front End Analysis: 9%Instructional Design: 13%Storyboarding: 11%Graphic Production: 12%Video Production: 6%Audio Production: 6%Authoring/Programming: 18%QA Testing: 6%Project Management: 6%SME/Stakeholder Reviews: 6%Pilot Test: 4%Other: 1%
Thinking Through the Numbers
I always weigh different factors to tweak these benchmarks. Chapman’s numbers are for everyone on the team, not just my role, so many of these estimate should be lower.
Analysis, Design, and Storyboarding
Analysis
Front end analysis is 9% of 552 or about 50 hours. The analysis involves other stakeholders, so it’s not just my time. This should also be much lower since it’s based on an existing training that is already working well. I’ll call my part of this analysis 15 hours. That analysis will include reviewing all of the existing training, feedback, and metrics, as well as meeting with SMEs and stakeholders.
My original estimate here was 30 hours, but that’s higher than I’d estimate now. I can use AI to help speed up the process of reviewing feedback and metrics. I also use AI to take notes during stakeholder calls, which reduces my time typing up minutes and noting key points.
Note that I’m being a bit idealistic for this example. I’m assuming that the existing course is working well. If the existing face-to-face training wasn’t working, this would require more analysis. I’m also agreeing with the client’s request for elearning as a solution and that 3 hours of elearning is actually the right solution. Real life is probably messier than this. A full-day training would probably be better converted to some form of blended learning rather than fully self-paced elearning. Just go with it for the purposes of the thought exercise here, OK?
Design
Instructional design is 13% or about 72 hours. I’ll call this 40 hours for me to create a detailed design document. Some of the additional time will be for the SME to answer questions and review the design document.
My original estimate here was 60 hours, which I think was too high. I would need to spend time going through the content in detail, and planning how to transform existing face-to-face activities to online activities takes time. AI can help brainstorm online activities, which speeds up that part of the process. The content is already organized for the face-to-face training, but I often move elements around for better flow when I convert training to online. A 3-hour elearning program would need to be broken into smaller chunks, so one goal of the design document is to identify where those divisions will happen.
Storyboarding
Storyboarding is 11% or about 61 hours. I’ll estimate 45 for my portion. I know from my own personal data that I tend to write faster than this benchmark.
I’m faster at storyboarding than I was over a decade ago. My original estimate was 55 hours, but between my increased experience and some improved efficiency with AI co-writing, I can reduce that time.
Development
Graphics
For graphics, the Chapman data assumes you spend 66 hours, more time on graphics than you do on storyboarding. Personally, I never end up spending that much time on visuals. The amount can vary based on the project though. If they have an existing template and a library of well tagged images, the amount is lower. If I have to spend a lot of time searching for stock images or creating custom graphics, the estimate needs to be higher. On the other hand, if I can generate AI images, the time is lower than searching for stock images (at least for many topics).
For this example, let’s assume I need to create a custom template and style. They have some existing images I can use, but many of the images in the existing slides are awful and need to be updates. I predict I can generate a lot of the images with AI, which reduces my time. I’ll need to create some custom diagrams. I decide to use 35 hours for creating the template and generating images.
My original estimate was 66 hours, the full amount from Chapman’s data. Like with storyboarding, I’m more experienced and efficient now than I was a year ago. I’ve also been using AI image generation tools long enough now to know how to get usable images with consistent styles.
Video and Audio
Since there’s no video, I use 0 for that value.
Audio will take some time to generate. I also need some time to generate closed captions. I’ll use 6 hours for this estimate (2 hours of work per hour of audio needed). That’s a number I would flag to revisit in the future as my experience and workflows with AI audio tools improve. I think it could be much lower than 6 hours.
My original estimate was 0, leaving it to the professional voice over artist. I also didn’t include time for closed captions in my original estimate, but I never deliver elearning without it now. AI makes creating captions faster. Chapman’s estimate is 33 hours for audio, which reflected the time for professional voice over and editing.
Building and testing
Authoring/Programming is 18% or 99 hours in the Chapman estimate. That seems low for building in Storyline, based on my experience, even assuming that we rely heavily on templates. IconLogic’s estimate is 2 hours per finished minute (120:1), or 360 hours. That’s a big discrepancy between the benchmarks. For my work, there’s some overlap between creating the template, generating graphics and audio, and authoring, so I can probably reduce this from the IconLogic estimate. I’ll split the difference and call this 180 hours.
QA Testing is 6% or 33 hours. Again, I think this is part of the difference in the IconLogic benchmark, since it doesn’t split testing out as a separate task. Generally a full review of a course takes me 2-3 times the length of the course, plus testing interactions throughout the process. For a 3-hour elearning course, that means I need 6-9 hours to do a single full review. I would do an alpha and beta review and revisions for the course, so I’ll estimate 20 hours for this section. (My original estimate was 17 hours; I would add more padding for reviews now.)
Project Management and Pilot
Project Management is also 6% or 33 hours. How much project management I do varies depending on the project and who else is on the team. I’ll assume 20 hours for this example.
The Pilot Test is 4% or 22 hours. I assume other people will be involved in that test, so I’ll estimate 6 hours for my part.
Total Hours: 367
Adding it all up, it’s 367 hours. How much padding I add to that estimate depends on a number of factors. If I’ve worked with the client before and I know they’re always responsive and very clear with feedback, I might use that estimate as is or round up to 370 hours. A 10% buffer (400 hours) or 20% (440 hours) is reasonable and typical for projects. A total of 400 hours would be a ratio of 133:1, which feels reasonable for this kind of conversion.
If the client seems unclear about what they want or I suspect that reviews and revisions will be time-consuming, I’ll add a larger buffer.
The above breakdown also helps me determine an estimate if I’m not creating the entire course. I often work in teams with other multimedia developers, so I might only be doing the analysis, design, storyboarding, and project management. Using this process, it’s easy to take those components and come up with a rough estimate for my portion of the course.
Adjustments from my original estimate
My original total was 450 hours, or up to 500-550 hours with a buffer. That means my updated estimate is 80-110 hours less than my estimate from 10 years ago. I have more experience now, but I also have more data from my own projects that lets me adjust from the Chapman benchmarks. I’m also starting to see improvements in efficiency by using AI for certain tasks.
Resources
Use these resources to create and compare your own time estimates (or to show stakeholders why you can’t create 5 hours of elearning in 3 weeks!).
Originally published 3/18/2014. Updated 5/16/2019, 3/3/2021, 4/17/2025.
