[carlosjobim]

There is almost no real business reason to use analytics, it's more of a cargo cult. If you want to check your sales channels, you make separate promo codes for each channel and see what codes are being used by paying customers. Because paying customers is the only metrics that matter if you're selling products online.

If you're curios about site visitors, you can use Clicky.com or Beamanalytics.io to get some basic stats that are enough to satisfy your curiosity, without spying on people.

[nicbou]

Which article brings you the most customers? Do people use that feature? Which landing page fails? Do you lose a lot of users during checkout? Which step is to blame? Should you care about Turkish visitors? What's the lowest resolution you should support? Can you use features that are missing from Safari 14? What sells the most on Sundays? What do people search for? Do they find it?

You are right, there are no reasons to use analytics.

[carlosjobim]

> Which article brings you the most customers?

Discount code in the article, now you know exactly, with hard evidence. Plus, you know if people actually got the message or if it just passed by. This works everywhere: Print ads, outdoors, Spotify, radio, social media, Google Ads, TV, you name it.

Most of the other stuff you listed is just common sense, if you have the slightest empathy for the customer when designing your website and purchase flow. The 99% of the websites that do use analytics, also have awful usability for visitors who want to become paying customers – so maybe these people putting themselves in the shoes of the customer could help better than trying to divine their analytics voodoo.

You have to do usability testing yourself from start to finish, but it seems a very tiny minority of website publishers actually do this. Then they say they need analytics to figure things out.

Take your example of check-out, which is usually the most worst user experience. What good is it to know you're losing customers at check-out, if you keep insisting on stuff like "Adress Line 2", pre-formatted phone number inputs, forced account creation etc.

I know your examples are a bit rhetorical, but I'll take them one by one, since I think the discussion is interesting:

> Which landing page fails?

The landing page that loads slowly and gets in the way of your sales with scroll-jacking, e-mail newsletter popup, cookie banners, irrelevant art and copy, no prices listed. Show people what you're selling and what is your price and let them purchase it easily. It is a solved problem. Which brick-and-mortar storefront fails? The one that doesn't have a door for customers to enter. But we don't need A/B testing for that.

> Should you care about Turkish visitors?

No need for advanced analytics to know which countries your site visitors come from. A simple counter without spying will do this. But this is a chicken and egg-problem: If you don't have material in their language, they won't become visitors at all, because search engines won't send them to you. But if you're already advertising to this public, of course you then have material suited for them.

> What's the lowest resolution you should support?

Your website should gracefully degrade, so that it supports as many devices as possible. Brick-and-mortar example again: What's the tallest person that should be able to enter our store? Make the door tall enough so that anybody can enter. Again, I say this is a solved problem.

> Can you use features that are missing from Safari 14?

These features should not be required for the customer to make a purchase, the website should gracefully degrade.

> What sells the most on Sundays?

You always know what you're selling, no need for analytics. Sales can't be made online without a record of it being made.

> What do people search for? Do they find it?

Considering the abysmal state of on-site search on most commercial sites, I'd say it doesn't seem like they use analytics to try to solve these questions. Most people have gotten used to only using Google, because on-site search always nets errors or irrelevant results, with few exceptions. But I agree with you on this point, this is where analytics can be useful. In my opinion only for the last mile, after you've done all the other essentials, which less than 10% actually bother with.