Website Analytics and Reporting

Two free tools, two different jobs

Almost every argument about website numbers comes from treating two different instruments as though they measure the same thing.

Search Console tells you what Google did with you. Analytics tells you what people did on your site.

Google Search Console reports on your presence in Google Search results: how often you appeared, what people typed, how often they clicked, and which of your pages Google has indexed. It is measuring Google, not your website.

Google Analytics reports on what happens after somebody arrives: which pages loaded, how long they stayed, what they clicked, whether they submitted a form. It measures behavior on your site, and only when its tracking code manages to run.

Neither is a measurement of traffic in any absolute sense, and asking which is correct is the wrong question. They are two partial views with different blind spots. Both are free, and there is no business for which Search Console is not worth having.

Why the numbers never match

This is usually presented to a business owner as somebody's mistake. It is not. There are at least seven structural reasons the two disagree, and every one is expected behavior.

1. Different populations. Search Console counts clicks on Google Search results. Analytics counts page views where its tag fired. Neither is a subset of the other.
2. Tag loss. Ad blockers, privacy browsers, refused consent banners, people who leave before the script runs, and ordinary script errors all remove hits from analytics. They do not remove clicks from Search Console, so analytics is usually the lower number.
3. Redirects. A click on a result that redirects can land in analytics as a different page, or as direct traffic with no source at all.
4. Google Search only. Search Console knows nothing about Bing, Facebook, direct visits, the QR code on your truck, or the Website button on your Google Business Profile. That last one matters, because for many local businesses it is the largest single source of website visits.
5. Different clocks. Search Console reports in Pacific Time and lags by a couple of days. Analytics uses whatever time zone the property was configured with.
6. Privacy filtering. Search Console deliberately omits rare queries, which is why the list of search terms never adds up to the total. Documented behavior, not a bug.
7. Attribution. Analytics assigns a visit to a channel using a lookback model. Search Console attributes a click to the click.

The practical rule: never compare the two numbers to each other. Compare each to itself over time.

What Search Console shows that analytics cannot

Search Console has exactly four headline metrics, and Google defines them precisely: clicks, the number of times a user clicked through to your site from Google Search results; impressions, how many times your site appeared in results; click-through rate, clicks divided by impressions; and average position, the average position of the topmost result from your site. The definitions are published at the Search Console performance report documentation.

Several things there have no equivalent in website analytics at all.

• Impressions. How often you appeared without being clicked. This is the only free source of demand data about your own business, and it reframes problems. If you appeared nine hundred times and got a dozen clicks, the issue is your title, your reviews or your relevance to that search, not your position.
• The actual search terms. Analytics has not shown organic keywords since Google encrypted them more than a decade ago. Search Console shows what people typed, subject to that privacy filtering.
• Indexing status. Which pages Google has, which it does not, and why. The message Crawled – currently not indexed is probably the single most useful diagnostic in small-business search work, and nothing else will tell you.
• Manual actions. Whether a human being at Google has penalized your site. No other source reports this.
• URL inspection. What Google actually sees when it renders one of your pages, which is often not what you see.

On the same documentation page Google warns that what you see in Search Console will not match what you see when you search, because results are specific to the time, place, device and recent history of the person searching. That sentence is also the reason a monthly ranking report is worth less than it looks.

What GA4 is, and what happened to your old data

The current version of Google Analytics is built on a different model from the one it replaced. The old system counted sessions and pageviews. The current one is event-based: every interaction, including a page view, is an event with parameters attached. That is why numbers for the same month never match between the two, and why a year-over-year comparison spanning the changeover is not a valid comparison at all.

The transition was not a soft one. Standard properties of the old system stopped processing data on 1 July 2023. The paid tier stopped on 1 July 2024, and from that week onward access to current and historical data through the interface and the reporting interface ended, with Google's documentation stating the data would be permanently deleted and would not be recoverable. Google's own notice is here: Universal Analytics sunset.

The consequence is worth saying plainly, because a lot of local businesses do not know it yet. If nobody exported the old data before the middle of 2024, your website history before 2023 no longer exists anywhere. Not in an archive, not on request. That is a reason to export what you have now, on a schedule, rather than trusting that a platform will hold it.

One more configuration problem is common enough to name. A great many small sites have analytics installed twice, once in a theme setting and again through a plugin or a tag manager. The result is doubled page views, an oddly flattering engagement rate, and a year of reports that mean nothing. It is worth checking once, and it takes a few minutes.

The report that is actually useful at this size

For a small local business, a useful report is four numbers and one sentence.

1. Calls. For a trade business the conversion is a phone call, and calls are invisible to website analytics unless a tap-to-call link is instrumented or a tracking number is used. A report that counts sessions and ignores calls is measuring the wrong thing entirely.
2. Form submissions. Counted properly, and checked occasionally by actually sending one, because the most common failure in this category is a form that silently stopped delivering months ago.
3. Business Profile calls and direction requests. Free, already collected by Google, and closer to revenue than anything on the website. For many businesses around the lake this is the single most important line on the page.
4. Search Console clicks. Not impressions. Clicks.

Then one sentence describing what changed and what it appears to have done. That is the report.

Two habits to avoid, because both are common in this industry. The first is reporting impressions as an achievement. An impression is your listing being present on a screen somebody may never have scrolled to; it is a measure of appearing, not of being chosen. The second is reporting rankings as a result. They are personalized by location, device and history, Google says so in writing, and a screenshot from an office in Kaufman County describes one person's screen.

If a monthly report runs to twenty pages and does not contain the word calls, it is being padded.

What is measurable about AI search, and what is not yet

This deserves a short, honest section because it is being sold hard right now.

Until recently, appearances in Google's AI-generated answers were invisible to site owners: they were folded into ordinary Search Console totals with no way to separate them. On 3 June 2026 Google launched generative AI performance reporting in Search Console, covering impressions in AI Overviews and AI Mode. That is a genuine step forward, and it is announced on Google's Search Central blog.

Read the limits carefully before anyone builds a monthly line item on top of it. As launched, it reports impressions only. No clicks, no click-through rate, no query detail, no indication of where in an answer you were cited. Reporting on it monthly means reporting a single number with no denominator and no outcome attached to it.

There is a related point about all the click-loss statistics circulating on this subject. Independent panel research has found that people click a traditional search result substantially less often when an AI summary is present, and Google has publicly disputed that study's methodology. Both of those facts are true, and neither tells you what happens on a search like plumber near Mabank. Most published measurements are drawn from keyword sets skewed toward informational questions, which are exactly the searches AI answers absorb. Local commercial searches may behave completely differently, and nobody has published credible data on them. Measure your own Search Console rather than importing somebody else's headline.

When you do not need this

A site with forty visits a month cannot learn much from analytics, and it should not be paying for a reporting retainer. At that volume a five percent improvement is two visits. Nothing reaches statistical significance: not a comparison test, not a conversion rate, not a claim about which page performs best. Reading daily numbers at forty visits a month is reading noise and calling it insight. Read months, not days, and expect even a month-to-month comparison to be weak.

What a site at that volume can honestly learn is four yes-or-no answers, and they are worth having:

• Is anyone tapping the phone number?
• Which handful of pages get any traffic at all?
• Are people finding you by your business name, or by the service you provide? Those imply completely different problems.
• Is the site broken on a phone?

Those are four answers, not a monthly deck, and the difference between forty visits and fifty-five is not an increase. It is noise.

Related products sold at this scale that should not be: heatmaps, session recordings and split testing, which need roughly a few thousand sessions a month before they say anything reliable; a long monthly PDF, because length is not diligence; and arguably analytics at all, when the whole funnel is the Google Business Profile. If customers find you in Maps and tap Call, website analytics measures none of it, and the profile's own report plus some form of call tracking are the right instruments. That is a common shape for trades in this market.

Where this work does earn its place: a site running advertising, where cost per lead has to be real; a store, where the funnel has steps that can be measured and fixed; a site with enough traffic that a difference is a difference; and any site at all having Search Console set up properly, because it is free and tells you things nothing else will.