Keywords Not Provided: What Google Hid and What to Do About It

“Keywords not provided” is the label Google Analytics assigns to organic search traffic when the originating keyword is withheld, which has been the default for virtually all signed-in Google searches since 2013. It means you can see that organic traffic arrived, but not what search term brought it there.

For most sites, this accounts for 85 to 95 percent of all organic keyword data. The visibility you think you have into organic search performance is largely an illusion built on the small fraction Google still passes through.

Why Google Removed Keyword Data in the First Place

In October 2011, Google began encrypting searches for users who were signed into a Google account. It was framed as a privacy measure, which it partly was. But it also had the convenient effect of pushing advertisers toward paid search, where keyword data remained fully visible. That tension was not lost on anyone paying attention at the time.

By late 2013, Google extended encryption to all searches, regardless of whether the user was signed in. The referrer string, which had previously passed the search query to Analytics, was stripped. What arrived in your reports instead was a session tagged as organic with the keyword field set to “(not provided)”.

I remember the reaction in agency circles at the time. There was genuine frustration, some performative outrage, and a lot of vendors rushing to sell workarounds. What there was not, in most agencies I saw, was a serious rethink of how organic search was being measured. Teams just accepted the gap and kept reporting on what was left.

Forrester has written about the risks of treating analytics as a black box, and the not-provided shift is a good example of that dynamic in practice. When a major data source disappears and the reporting methodology does not change, you end up measuring the shadow of the thing rather than the thing itself. The Forrester perspective on analytics black boxes is worth reading if you are responsible for how organic performance gets reported upward.

The short version: Google made a decision that suited Google, framed it as user protection, and the marketing industry largely adapted without asking hard questions about what it had actually lost.

What Data You Can Still Access

The keyword data did not disappear entirely. It moved. Google Search Console still holds query-level data, including impressions, clicks, click-through rate, and average position. What it does not hold is anything that happens after the click: session depth, time on site, pages visited, conversions. That post-click behaviour stayed in Analytics, attached to sessions that no longer carry keyword attribution.

This creates a split that most teams manage badly. Search Console tells you what queries are driving clicks. Analytics tells you what organic visitors do once they arrive. Connecting the two requires either the native GA4 and Search Console integration or a manual process of matching landing pages to queries and inferring intent from there.

The GA4 integration with Search Console is worth setting up if you have not already. It surfaces a “Queries” report inside GA4 that shows the search terms driving organic sessions, alongside engagement metrics. The data is sampled and subject to Search Console’s own thresholds, which means low-volume queries get filtered out, but it gives you a workable view of what is driving organic traffic at the keyword level.

If you are still finding your way around GA4’s reporting structure, the Semrush breakdown of GA4 user metrics is a useful reference for understanding how sessions, users, and engagement are defined in the current version of the platform.

For a broader view of how organic analytics fits into the wider measurement picture, the Marketing Analytics hub at The Marketing Juice covers GA4, attribution, and performance reporting in more depth.

The Landing Page Workaround That Actually Works

When keyword data was available, the typical workflow was: identify high-converting keywords, optimise the pages those keywords landed on, repeat. Without keyword data, most teams stalled. But the logic does not actually require keyword data to function.

A landing page that receives organic traffic is, by definition, ranking for something. If you know the page, you can infer the intent. A page about enterprise accounting software pricing is not receiving traffic from people searching for recipe ideas. The keyword uncertainty is real, but the intent range is narrow.

The practical approach is to segment organic traffic by landing page in GA4, then evaluate each page against engagement metrics: session duration, pages per session, goal completions, and conversion rate. Pages that perform well on these metrics are working, regardless of which specific queries are feeding them. Pages that pull organic traffic but show high bounce rates and zero conversions are a signal that the content is not matching the intent of the visitors arriving.

I ran this analysis across a retail client a few years ago, looking at roughly 400 organic landing pages. About 60 pages were driving the majority of organic-attributed revenue. Another 80 or so were pulling decent traffic volumes but converting at near zero. The keyword data would have been useful context, but the landing page performance data was enough to make the optimisation decisions. We redirected resource from the high-traffic, low-conversion pages and put it into the pages that were already working. Organic revenue went up. No keyword data required.

This is not a perfect substitute for keyword-level insight. There are cases where knowing the specific query matters, particularly when you are trying to understand whether a page is cannibalising another or when you are building out a new content cluster. But for the majority of ongoing optimisation work, landing page performance is a more actionable signal than keyword data ever was.

How Google Search Console Fills the Gap

Search Console’s Performance report is the closest thing to what was lost. It shows you the actual queries triggering impressions and clicks, broken down by page, device, and country. The data covers the last 16 months and is updated with a delay of a few days.

There are limitations worth knowing. Search Console applies anonymisation thresholds that filter out queries with very low impression volumes, so long-tail terms often disappear from the data. Average position is a mean across all positions in all sessions, which can be misleading if a query triggers results at different positions depending on personalisation or device. And the data reflects what Google chose to show, not necessarily what users searched for in the way you might expect.

Despite those caveats, Search Console is the most reliable source of actual query data available to site owners. The workflow I use is straightforward: export query data at the page level, match it against GA4 organic landing page performance, and look for patterns. Pages ranking well for queries with clear commercial intent but converting poorly are candidates for conversion rate work. Pages with high click-through rates but low average positions are candidates for content depth or authority building.

If you are running A/B tests on organic landing pages, the Semrush guide to A/B testing in GA4 covers how to set up experiments that will show you whether changes to organic landing pages are actually moving the metrics that matter.

Third-Party Tools and What They Actually Tell You

The not-provided problem created a market for tools that could estimate keyword rankings and traffic from outside Google’s walls. Semrush, Ahrefs, and Moz all crawl the web, track ranking positions, and model traffic estimates based on click-through rate curves. They are useful. They are also not what they are sometimes sold as.

These tools show you estimated rankings and estimated traffic. The estimates are based on their own crawl data and modelled assumptions about how much traffic a given ranking position receives. For well-crawled, high-volume keywords, the estimates are reasonably reliable as directional signals. For niche terms, new content, or heavily personalised search results, the gap between the estimate and reality can be significant.

I have sat in client meetings where a tool showing a ranking improvement was used as the primary evidence of SEO success, while the actual organic traffic in Analytics was flat or declining. The tool was not wrong exactly, it was just measuring something slightly different from what was happening in reality. Ranking position and traffic are correlated but not the same thing. A position-one ranking for a query no one searches is worthless. A position-four ranking for a high-volume query with strong intent can be significant for a business.

Use third-party tools for what they are good at: competitive analysis, keyword opportunity identification, and tracking ranking trends over time. Do not use them as a substitute for actual traffic and conversion data from your own Analytics instance.

If you are considering supplementing or replacing Google Analytics entirely, the Crazy Egg overview of Google Analytics alternatives covers the main options and what each one does differently.

Combining Qualitative Data With the Quantitative Gap

One of the underused responses to the not-provided problem is qualitative data. If you cannot see what search terms brought visitors to a page, you can observe what those visitors do once they arrive, and you can ask them directly.

On-site surveys, exit intent prompts, and session recording tools all provide context that keyword data never could. A visitor who searches for “enterprise accounting software comparison”, lands on your pricing page, reads it for four minutes, and then exits without converting is telling you something specific. The keyword data would have confirmed the search term. The behavioural data tells you the content is not closing the gap between interest and decision.

Tools like Hotjar sit alongside Analytics specifically to surface this kind of behavioural context. The Hotjar and Google Analytics integration is designed to let you pull session recordings and heatmap data into the same workflow as your quantitative metrics, so you can see not just that a page has a high bounce rate but where and why visitors are leaving.

Early in my career, when I was building websites on a zero budget because the MD said no to external spend, I learned to read user behaviour from whatever signals were available. Web analytics in 2000 was primitive. You had raw server logs and not much else. The discipline of making decisions from incomplete data, rather than waiting for perfect data that never arrived, has stayed with me. Not-provided is a version of the same problem. The data you want is not available. The question is whether you adapt your methodology or just complain about the gap.

What This Means for Organic Reporting

The not-provided problem has a knock-on effect on how organic search performance gets reported to stakeholders, and most teams have not resolved it cleanly.

The typical organic report shows sessions, maybe engagement rate, maybe goal completions attributed to organic. What it rarely shows is a coherent account of which content is driving which outcomes, and why. Keyword data used to paper over some of that gap. Without it, the weakness in the reporting logic becomes visible.

The better approach is to build organic reporting around outcomes rather than traffic. Organic-attributed conversions, revenue, and leads are the metrics that connect to business performance. Traffic volume is a leading indicator at best. If organic traffic grows but organic-attributed revenue stays flat, that is a problem. If organic traffic stays flat but organic-attributed revenue grows, that is a success. The keyword data would have been interesting context, but it would not have changed the commercial conclusion.

I spent several years running performance marketing across a client portfolio that included e-commerce, lead generation, and subscription businesses. The teams that reported on outcomes consistently made better decisions than the teams that reported on activity. Not-provided accelerated that lesson for organic search, because it removed one of the activity metrics teams had been leaning on. What remained was the outcome data, which was always the more important signal.

Filtering your Analytics data correctly is a prerequisite for any of this to work reliably. The Crazy Egg guide to Google Analytics filters covers the configurations that keep your organic data clean and separated from direct, referral, and paid traffic.

The Honest State of Organic Analytics in 2025

Organic search analytics has never been as clean as paid search analytics. Even before not-provided, keyword data in Analytics was subject to sampling, session attribution quirks, and the usual gaps between what a tool reports and what actually happened. Not-provided made those limitations more visible, which was uncomfortable but arguably useful.

The honest position is that organic search is partially measurable. You can see traffic volumes, landing page performance, engagement behaviour, and outcome attribution. You cannot see the exact query that triggered each session. For most optimisation decisions, that is enough. For some decisions, particularly around content strategy and keyword targeting, you need to supplement Analytics with Search Console and third-party tools, accepting that the picture you get is directional rather than definitive.

When I was judging the Effie Awards, one of the consistent weaknesses in entries from digital-first brands was the conflation of data availability with measurement quality. Teams with access to granular click-level data often made worse strategic decisions than teams working from rougher but better-framed signals. More data is not the same as better insight. The not-provided problem is a useful reminder of that.

Marketing analytics is a perspective on reality, not reality itself. The keyword that brought someone to your site is one data point in a sequence of decisions and behaviours. Losing it is inconvenient. It is not the catastrophe it was sometimes described as in 2013, and it does not prevent you from running a well-measured organic search programme if your methodology is sound.

For more on building an analytics framework that holds up under these kinds of constraints, the Marketing Analytics section of The Marketing Juice covers measurement strategy, GA4 configuration, and performance reporting across the full channel mix.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions