Adoption of Bootstrap design

Analysis
1

I’m working on a State of the Web episode about design and pulling stats for the popularity of design systems like Bootstrap. Here’s the analysis:

SELECT
  SUBSTR(_TABLE_SUFFIX, 0, 10) AS date,
  IF(ENDS_WITH(_TABLE_SUFFIX, 'desktop'), 'desktop', 'mobile') AS client,
  SUM(IF(app = 'Bootstrap', 1, 0)) AS urls,
  ROUND(SUM(IF(app = 'Bootstrap', 1, 0)) / COUNT(DISTINCT url), 4) AS pct
FROM
  `httparchive.technologies.*`
GROUP BY
  date,
  client
ORDER BY
  date,
  client

Raw results

As a percent of all pages tested:

image

Absolute count of pages detected with Bootstrap:

image

So until 2018, our Alexa websites were increasingly adoption Bootstrap. Then when we switched to CrUX, we starting looking at a wider and distinctly different population and the adoption rate dropped.

As of today, Bootstrap is detected on nearly a million websites, representing ~25% of the web!

1 Like
2

How did you get to the last statement in this post? There’s over 1.5 billion websites active online right now and 1 million websites doesn’t represent 25%

1 Like
3

Weclome DeanoLingardo! :slight_smile:

If you consider the URLs in the HTTPArchive dataset to be a representative sample of the web and you find that 25% of those URLs have Bootstrap, then it’s a reasonable extrapolation that ~25% of the web uses Bootstrap.

There’s another argument to be made here about the percentage of web traffic that HTTPArchive examines that uses Bootstrap which might be a separate interesting question, but I’m not sure that’s what rviscomi was referring to in this case.

1 Like
4

What @patrickhulce said!

The second to last image shows the relative percent, which we can extrapolate to the entire web, assuming that our sample set is representative (I believe it is).

image
image1182×728 43.2 KB

5

@rviscomi When you say the data is representative of the web, is it a representative sampling of all sites, giving equal weight to obscure abandoned sites as to the most popular active sites today? Or is it a representative sampling of page loads?

Reason: I’d like to look at, eg, Bootstrap usage as a percent of page loads, not as a percent of urls, in case the top sites are very different from the long tail of the web. Does that make sense?

6

Yes, I believe equal weight is given to all sites. If you are interested in weighting by domain polularity, you could join to httparchive.urls.20170315 and use the Alexa_domain field as a rank.

2 Likes
7

Here’s an example of joining to that Alexa data: Getting domain rank with the new rank-less Chrome UX Report corpus

2 Likes