Search Console position or GA4 organic conversions: what do you trust for SEO?

I keep seeing SEO reports lead with average position from Search Console, and I think that number misleads more often than it helps. It is impression-weighted, so if you start ranking for many new long-tail queries at position 30, your average can fall even while clicks and conversions rise. A page that moves from position 12 to 8 on a query with 10,000 impressions can matter more than a jump from 40 to 5 on a query nobody searches.

GA4 has its own problems. Organic conversions depend on the channel grouping, last-click or data-driven attribution, and how much traffic lands as direct because of missing UTMs or dark social. So the two sources can disagree for reasons that are both legitimate.

My current habit is to use clicks and conversions by landing page as the headline, and treat position only as a diagnostic for specific queries. I am less sure how to handle the cases where GA4 shows organic conversions falling while Search Console clicks look flat. How do you decide which source to believe when they split, and what gap between them would make you investigate tracking before blaming the content?

Kai, I agree that average position is a poor headline, but I'd go a step further: GA4 and Search Console aren't measuring the same thing, so the first check is whether they are even counting the same visits.

Search Console clicks come from Google's own logs. GA4 organic sessions depend on the browser loading the tag, consent choices, and the referrer surviving redirects. On a shop, the usual breakage is at checkout, where a payment provider or a cross-domain hop strips the referrer and the session gets reclassified as direct.

So before debating which number to trust, I'd run this for the last 16 weeks, per landing page:

1. Divide GA4 sessions where source/medium is google/organic by GSC clicks for the same page and dates.
2. Plot that ratio.

A gradual drift usually means real demand or ranking change. A sudden step, especially on a specific date, usually means something changed in your setup: a GTM publish, a consent banner update, a new domain or subdomain. If the ratio is stable but conversions fall, look at the checkout path and purchase event deduplication before touching content.

Position stays useful, but only per query and only for queries whose impressions held steady. Otherwise the query mix is doing the moving.

Iris, the ratio check is the right first step, but for local services I'd question the denominator. A clinic or plumbing firm often converts through a phone call that starts on the Business Profile listing, not on a landing page. Neither GA4 organic sessions nor GSC clicks see that path. So a falling GA4 organic conversion line can reflect real demand moving into the map pack and calls while the site looks flat.

I'd add the Business Profile as its own line in the comparison: calls, direction requests and website clicks from the listing, over the same 16 weeks. Also use a separate call tracking number for the website and for the listing, so a call from a service page isn't lumped in with one from the map.

On the step-change test, check the breadth of the drop. If the ratio falls across every page on the same date, suspect measurement, such as a tag change or a consent update. If only one template or directory moves, suspect your setup on those pages.

Before building the ratio, I'd check the Search Console side. The Performance report combines all search types by default, and Discover and Google Images clicks land in the same totals. Discover can swing hard on a single article, and those visits may show up in GA4 under a different channel or as a referral, depending on your rules. A page whose clicks rose because of Discover will look like a conversion problem in the ratio. Filter to the Web search type first, then compare.

Marco, I like the separate call numbers, but calls from the listing only reach your reporting if you import them somewhere. If they don't, the 16-week comparison sets a site line against a line the stack can't see, and the "falling" GA4 line may be a measurement gap rather than demand moving to the map pack. A quick test is to check whether listing call volume changed on the same dates as the GA4 drop.

Also, Search Console data from the last two or three days is still settling. Don't read a fresh dip as a step change until it has stopped moving.

Kai, the call import point is fair, and it cuts against my earlier framing. I also think the GBP visits can push the ratio the other way. A website link in a Business Profile is a Google referral, and GA4 often files that under google / organic, even though Search Console never counts it as a search click. So a page can look like it converts better in GA4 than GSC supports, because some of its "organic" sessions came from the map pack.

A practical check: add a UTM to the website URL in the listing, such as utm_source=google&utm_medium=business_profile. Then in GA4, split the sessions with that medium out before you build the ratio. If the GBP sessions are a large share of a page's google / organic traffic, the denominator problem is bigger than the tracking problem.

It's also worth checking whether the listing's website URL points to the same pages you're comparing. For a clinic that links the homepage, most of that traffic lands on one page and can distort a per-page ratio.

Marco, the UTM tag may do more than split the data for analysis. GA4's default Organic Search channel matches medium "organic", so a link tagged utm_medium=business_profile moves out of google / organic on its own, from the day you add it. That makes the tag a filter on new traffic, not only a diagnostic. Log the date in your annotations, or the organic line will look like a measurement break.

Before you trust the split, check two things. Click the listing link in DebugView and confirm the session appears under a different source/medium row. Then check that the UTM survives any redirect on the target URL. A 301 that drops the query string sends those visits back into google / organic.

I'd also run the per-page ratio on the period before the tag exists. That tells you how much of each page's "organic" traffic came from the map pack, which is the number you actually need before you change anything.