Skip to content

Instantly share code, notes, and snippets.

@sylar
Forked from pocha/weather-com-changes-mmt.md
Created July 21, 2015 12:47
Show Gist options
  • Save sylar/4078009fba5dadd3728f to your computer and use it in GitHub Desktop.
Save sylar/4078009fba5dadd3728f to your computer and use it in GitHub Desktop.

Research

To improve weather.com, broadly the intent of the users need to be matched with what the site provides. Additionally, new market/user trends can be predicted. Putting trend prediction outside the scope of this doc currently.

What site provides

User acquisition could be direct, search or social. Weather.com has a pretty high direct visits so people are largely landing on the home page itself. Hence the home page seems to be the correct indicator of what the site thinks users want.

  1. Your city weather is the most prominent offering but without a call to action button.
  2. A sensational weather news story & call to action button to read more.
  3. Apart from the top links, getting local forecase is another prominent engagement mode for Weather.

Limiting research to above the fold as it is the most important part here.

What people want from Weather.com

Similarweb shows Audience Interest as Weather related news which is a bit suprising as the name suggest that people might just be coming for checking weather information.

This is further coroborated by the fact that Wunderground which is exclusively forecast & weather information site (also listed as the most prominent similar site) does not have much traffic. Disclaimer: need to check how old wunderground.com is. Making assumption that it is reasonably old.

Conclusion0

A good engagement to the users is coming from the weather related news. Since a lot of traffic is from the US, it makes sense to keep information for in/around US. A suggestion for better engagement here is - rotate the main news when a user lands on the site again. Assumption here is - the user is news hungry & comes back to site often.

Update- it looks like they do rotate the main news although not on refresh exactly. But guess after some time delay.

Once user moves to the main story, the Recommended story is probably a bad job. It should show up similar stories, but it showed me varied stories. Assumption here is - the user is interested in similar stories.

The home page can be more customized to user's taste for better user engagement. The 'Recommended Story' algo can be picked up & applied on home page to provide more contexual story links. The stories that are showing up right below the fold is where user's taste stories can show up.

Putting up recommended stories & most popular stories side by side & right below the fold could be an interesting hack to increase user engagement. Hypothesis here is - user has been served his best & global best, so its like best of both world.

Looking at news oriented user behavior on the site, it could be the default destination for its users to check any weather related news (hypothesis). A generic search bar should be provided on the home page for search of such news.

Fixing too many contexts by removing the two bar layout. The sidebar is primarily used for ads at various places on the site. While scrolling, it puts up two 'uncorrelated segements' to the user. Rather, a more youtube like layout with just one main layout interspersed with tile based layout for regions that dont need to span the whole width could fix things.

5 changes

  1. Rotate main news on page refresh.
  2. Stories crafted to user's taste below the main story.
  3. Putting recommended & top stories side by side or one below another.
  4. A generic search bar on top that can search for news article from archives
  5. Single column layout or pinterest like tile based layout to avoid too much context problem

5 metric to measure for home page optimization (in order of importance)

  1. Click through rate (decrease in bounce/exit rate). Increase in click through may result in decrease in avg time on the home page.
  2. direct traffic
  3. social & referral traffic (total number not %)
  4. Avg time on home page

Impact of features on metric

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment