What "Free" Analytics Actually Costs You

Google Analytics is free in the sense that no invoice arrives. It is not free in the sense that it costs you nothing, and the costs are spread across four line items that never appear on a pricing page.

If you are running a small content site, the honest comparison is not "free versus paid". It is "what am I spending to keep this free thing working".

Cost one: the consent banner, and the data you lose

GA4 sets cookies. Under European rules that requires consent before the script runs, which means a consent management platform, which means a banner on every page.

The banner costs three things:

Conversion. Banners reduce signups. The measured effect varies, but nobody reports an increase.

Page weight and speed. A consent platform is another third-party script, typically 30–80 KB, loaded before anything else.

Data accuracy. People who decline are not counted. That group is not a random sample: it skews toward privacy-conscious, technically literate visitors, who are often the audience worth understanding. Your traffic numbers are systematically wrong in one direction.

Cookieless analytics avoids all three because there is nothing to consent to. That is the entire argument, and it is a strong one.

Cost two: the retention limit

GA4's free tier keeps event-level data for a limited window (two months by default, extendable to fourteen). After that, the detail is gone and only aggregates remain.

For most sites this never matters, because nobody looks at six-month-old event data. For a site trying to understand a seasonal pattern or investigate a slow decline, it matters a lot, and the fix is exporting to BigQuery, which is a project.

Cost three: the learning curve

GA4 is a professional tool. It was rebuilt around events and parameters, which is more correct and considerably less intuitive than the old Universal Analytics. The interface assumes you know what an exploration is, what a dimension scopes to, and why "users" and "active users" differ.

The realistic estimate for a non-analyst to become productive in GA4 is a few days. Those days are the cost. If you are a solo founder, they are days not spent on the product.

Compare: a cookieless tool with four metrics takes about five minutes to understand, because there are four metrics.

Cost four: the configuration tax

This is the one nobody mentions. GA4 does not work correctly out of the box for most sites. You need to:

• Define and mark conversion events
• Set up a data stream per platform
• Configure internal traffic filters, or your own visits pollute the data
• Configure cross-domain tracking if you have more than one domain
• Set retention explicitly
• Wire up Search Console
• Decide what to do about the default "enhanced measurement" events, some of which double-count

Each of these is small. Together they are an afternoon, and the maintenance is ongoing, because Google changes defaults.

What you actually need

For a content site, the questions you act on are about six:

1. How many people visited this week, and is that up or down?
2. Which pages get read?
3. Where does traffic come from?
4. Which pages lead to a signup or a purchase?
5. What device and country mix am I serving?
6. What search terms bring people in?

Five of those six are answered by a one-page dashboard. The sixth comes from Search Console, not from analytics at all.

GA4 answers all six and also answers four hundred questions you will never ask, at the cost of the four items above.

The comparison that matters

The last two rows are the real trade. If you need cross-session funnels, GA4 is the right tool and the costs are worth paying. If you are running a content site and the answer to "do I need funnels" is "I have never built one", you are paying four costs for two features you do not use.

The version I would run

Collection: self-hosted Umami, one container, no cookies. Or Plausible if you would rather pay than maintain.

Weekly review: one page, five minutes, same time every week. Traffic, top pages, sources. Write down one observation. That is the entire analytics practice.

Deeper questions: when you have a real question that a dashboard cannot answer, export events to a database and query them with Metabase. This happens maybe twice a year, and it is cheaper than paying for a tool that anticipates it.

Search data: Search Console, which is free, has no consent implications, and tells you what people typed to find you, which analytics never will.

The total cost of that stack is a few dollars a month and about ten minutes of setup. The total cost of the free alternative is a banner, a lost afternoon, and a number you cannot fully trust.