How Active Installs Are Calculated

Introduction: How WordPress.org Tracks Plugin Installations

WordPress.org has access to remarkably precise data about plugin installations across the web. This is possible because WordPress installations regularly check for available updates by contacting WordPress.org servers. During these update checks, WordPress installations send information about their active plugins, including plugin names and versions.

This automatic update mechanism allows WordPress.org to count exactly how many times each plugin is installed and active. The data is highly accurate, with the only exceptions being WordPress installations that cannot reach WordPress.org servers for various reasons — typically sites with no traffic or those behind restrictive firewalls.

The Challenge: Rounded Public Data

While WordPress.org has precise internal data, they only publish rounded thresholds based on the first significant digit × powers of ten — e.g. you’ll see 10+, 20+, 30+, … up to 90+; then 100+, 200+, 300+, … 900+; then 1,000+, 2,000+, 3,000+, … and so on.

For example, a plugin with exactly 1497 active installations would be listed as 1,000+ on WordPress.org:

Additional Data: Version Distribution

For most plugins WordPress.org also publishes detailed version distribution data. This information shows what percentage of users are running each version of the plugin. You can view this data on any plugin page under the “Advanced View” section:

There you will see a diagram where you can hover over the parts and see the percentage values.

For instance, a plugin might show:

• Version other: 6.7%
• Version 2.3: 8.4%
• Version 2.4: 18.2%
• Version 2.5: 66.7%

This percentage data is crucial for our calculation method.

Note: WordPress.org does not publish these percentages for brand new plugins or those with very few installations. However, we currently (at the time of writing) do not know the exact criteria. We have also seen plugins with several hundred or even thousands of installations for which this data is not available. In most cases, however, this data should be available for plugins starting from around 10–30 installations.

The Calculation Method We Use: Reverse Engineering from Percentages

The core insight behind our calculation method is that we can work backwards from the published percentage data to determine the actual number of installations. Here’s how it works:

The Basic Principle

When WordPress.org publishes that 66.7% of users are running version 2.5, this percentage is calculated by dividing the number of installations running version 2.5 by the total number of active installations.

Formula: (Installations of Version X / Total Installations) × 100 = Published Percentage

By reversing this formula, we can find the total installations if we can determine how many installations are running each version.

However, this only works when there are at least two different versions in use, as we need multiple data points to create a system of equations that can be solved. Plugins with only one active installation cannot be calculated this way. Therefore, having more than one active version is a prerequisite.

The Calculation Process

1. Start with the rounded range: We know the plugin has 1000 or more installations, but fewer than 2000 (based on the 1,000+ label).
2. Test different total values: We systematically test different total installation numbers within that range.
   1. Calculate expected version counts: For each test value, we calculate how many installations would be running each version based on the published percentages.
   2. Round to integers: Since you can’t have a fraction of an installation, the calculated version counts must be rounded to whole numbers.
   3. Back to percentages: We then convert the integers back into percentages and round these to the same number of decimal places as the original, published percentages.
3. Find matches: We look for total values where all calculated percentages, match the published percentages exactly.

Example Calculation

Let’s say a plugin has a published value of 1,000+ active installations and these published percentages:

• Version other: 6.68%
• Version 2.3: 8.42%
• Version 2.4: 18.17%
• Version 2.5: 66.73%

Fortunately, the WordPress API provides the values with two decimal places, which makes our calculations even more successful.

We test different total values from 1000 to 1999 – for example:

Testing 1497 total installations:

• Version other: 1497 × 0.0668 = 99.9996 → 100 installations
• Version 2.3: 1497 × 0.0842 = 126.0474 → 126 installations
• Version 2.4: 1497 × 0.1817 = 272.0049 → 272 installations
• Version 2.5: 1497 × 0.6673 = 998.9481 → 999 installations

Converting back to percentages:

• Version other: (100 ÷ 1497) × 100 = 6.68002672 → 6.68% ✓
• Version 2.3: (126 ÷ 1497) × 100 = 8.41683367 → 8.42% ✓
• Version 2.4: (272 ÷ 1497) × 100 = 18.16967268 → 18.17% ✓
• Version 2.5: (999 ÷ 1497) × 100 = 66.73346693 → 66.73% ✓

Since all calculated percentages exactly match the published percentages, 1497 could be the correct number of active installations.

Handling Multiple Possible Values

Sometimes, multiple total values can produce matching percentages. This happens more frequently with larger plugins because the rounding effects become less significant.

In our previous example, the percentage values match not only for the value 1497, but also for the values 1662 and 1960. So we know one of these 3 numbers is very likely the correct one, but we don’t know which one.

In our diagrams for active installations, we represent such cases as scatter points to illustrate the uncertainty. Often, a pattern emerges from the historical trend from which the correct values can still be guessed.

For more information on interpreting the charts and what the data means, see our article How to Interpret Active Installs Charts.

Further Challenges

There are still a few detailed challenges in the calculation, such as false positives, rounding inaccuracies, data synchronization, and more. These would go beyond the scope of this article, but we wanted to mention their existence here for the sake of completeness.

More Versions Improve Accuracy

The calculation becomes more reliable when a plugin has many different versions in use. Here’s why:

Minimum requirement: At least two different versions must be in use for the calculation to work at all. With only one version (100% usage), any total number would produce the same result, making it impossible to determine the actual count.

• Minimum constraint: At least two versions create the minimum system of equations needed for a solvable calculation
• More data points: Each additional version provides an additional constraint that must be satisfied
• Reduced ambiguity: With more versions, fewer total values will satisfy all percentage requirements
• Better precision: The rounding effects are distributed across more data points

For example, a plugin with 5 different versions in use is much more likely to have a unique solution than one with only 3 versions. However, even with just 2 versions, we can often find valid solutions for plugins with less than 100 active installations.

If a plugin is in use in 7 or more versions, we can sometimes still find unique solutions even with 2,000 to 3,000 installations. Beyond that, however, it’s extremely unlikely that we’ll find unique matches. From that point on, one could only try to identify a pattern on our active installs diagrams based on the scatter points and historical trends, from which the correct values might be guessed. But this also has its limits, of course.

Conclusion

While WordPress.org only publishes rounded installation ranges, the additional version distribution data allows us to calculate the actual number of active installations for many plugins.

These precise numbers can help plugin developers track growth and evaluate marketing measures, users make installation decisions, and researchers study ecosystem trends.

While the method works best for plugins under 2,000 installations with multiple active versions, it provides unprecedented visibility into the WordPress plugin ecosystem.

Our personal opinion: We consider it a stroke of luck that more accurate data is available for smaller plugins, as new plugin developers often struggle with growth anyway, and this can be a nice advantage for up-and-coming plugin developers over established competitors.

For information on how to interpret the active installs charts and understand what the data means, see our article How to Interpret Active Installs Charts.

---