Analytics

Foundations

Analytics

• Overview
• Traffic Analysis
• Search Performance
• Content Performance
• BotsSoon
• AudienceSoon
• ConversionsSoon

Foundations

Analytics

• everything together
• visitors, including those arriving from AI assistants
• Google Search Console data
• how content you published is doing

Overview#

Seerly spends most of its time measuring whether AI answers mention you. Analytics is where that connects to what happened afterwards: traffic arriving from AI referrers, your Google search performance, and how content you published has actually done.

Visibility is a leading indicator, and this is the screen where you find out whether it led anywhere.

Data sources#

Google Analytics picks out visitors arriving from AI assistants and separates them from ordinary search and direct traffic, which makes it the closest thing you have to a direct answer on whether any of this is producing visits. Google Search Console sits alongside it, putting your Google performance next to your AI visibility so a single page can be read on both surfaces at once. Content Performance covers how content created in Seerly has done since it was published, and Web Presence tracks the sites linking to you, which turns out to be closely related to which sources end up being cited. Traffic Analysis and Search Performance can both be narrowed to one country, as described under Filter by location.

Traffic Analysis

AI Sources

which assistants actually sent visitors

Top Queries

what those visitors had asked

Filter by location#

Traffic Analysis and Search Performance each have their own filters, in one row at the top of the page: the location picker first, with the date range beside it. The row stays at the top as you scroll, so an active filter is never out of sight. The picker reads All locations until you choose a country, then shows that country's flag and name with a small button to clear it. You filter by one country at a time.

Open the picker to pick a country from the list or search for one. After All locations, the list holds every country the page has data for in the chosen date range, each with its share, ranked by clicks on Search Performance and by sessions on Traffic Analysis. The two pages draw on different sources, so their lists can differ.

The choice is kept in that page's link, together with the date range, so reloading the page or sharing the link shows the same view. It belongs to that page only. Opening Traffic Analysis or Search Performance from the sidebar always starts on All locations and the default date range, whatever you had chosen on the other. If a link names a country the page has no data for in the period, the picker still lists it as the selected country, without a share, and the page shows a single message in place of its figures, as described below. The Analytics overview has no picker and opens unfiltered.

While no country is selected, a Top locations card ranks every country. On Search Performance it sits below the key figures and ranks by clicks; on Traffic Analysis it sits below the traffic overview chart and ranks by sessions. Clicking a row filters the whole page to that country, and the card is then removed, because the picker is how you switch or clear.

Top locations

Search Performance

Country	Clicks	Share	vs previous period	Impressions	CTR	Position
India	1840	46%	+12%	52K	3.5%	14.2
United States	1210	30%	-4%	61K	2.0%	11.8
United Kingdom	460	11%	New	9.4K	4.9%	9.6

Search Performance gives each country its Clicks, Share, vs previous period, Impressions, CTR and Position. Traffic Analysis gives each its Sessions, Share, vs previous period, Engagement rate and AI-referral sessions, so you can see which countries your AI visitors come from rather than only how many there are. Most columns sort when you click the header. Eight countries are listed at a time; the page controls at the bottom left of the card step to the first, previous, next or last page. Visitors whose country was not reported are not listed, in the card or in the picker, though they still count toward the total each share is measured against. A site with traffic from only one country gets a single sentence in place of the table.

Once a country is selected, the Clicks figure on Search Performance and the Sessions figure on Traffic Analysis add the country's share, for example 41% of all Google clicks or 41% of all sessions. The keyword drilldowns leave out their Top countries section, since it would be a single slice, and give the width to the sections that remain. A country with nothing to show gets one message instead of a page of zeros: Not enough data on Traffic Analysis, and "No Google Search data for" the country on Search Performance, each with a View all locations action. If the filter cannot be applied to a request, the page says The location filter could not be applied. and offers Try again and View all locations, so figures for all locations are never shown under a country.

Bing does not report search data by country, so a country and Bing never combine. Choosing a country on Search Performance shows Google only, and a note under the page heading says so with View all locations to undo it. Choosing All or Bing while a country is active clears the country instead, and the note offers a way back to it. When Bing is connected the Top locations card carries a Google Search only badge, and on the Bing source it is replaced by the line "Location breakdown is not available for Bing." If Bing is your only connected source, the picker explains that Search Console is needed to filter by location, and a country named in a shared link is kept in the link without being applied.

Some figures are never narrowed by country, and each is marked All locations while a country is selected: the Lost, Fading and New views of Your pages on Google, since indexing is not specific to a country, and the AI side of Search & AI Topic Convergence and Top Insights, since AI visibility is not measured per country here. The search side of those two cards is filtered.

Landing pages#

At the bottom of Traffic Analysis, the Landing Pages card lists every page a visit started on, most sessions first. It follows the date range and the country chosen at the top of the page.

The Source menu narrows the list to visits from one kind of source: All traffic, AI assistants, Google search, Other search, Direct, Paid, Social, Referral or Other. Each visit is counted under one of them only. A visit from an AI assistant is always counted as AI assistants, and Google's unpaid search results are kept apart from other search engines so you can read them on their own.

Choosing AI assistants adds an AI source menu listing each assistant that sent visits in the period. Leave it on All AI sources to see every page AI visitors arrived on, or pick one assistant to see only the pages it sent people to.

Landing Pages

AI assistants

Page	Sessions	Top Source	Engagement	Pages / Session	Avg Time
/pricing	184	ChatGPT 62%	71.4%	3.1	2m 12s
/blog/buyers-guide	96	Perplexity 48%	64.0%	2.2	1m 40s
/compare	41	ChatGPT 55%	58.5%	1.9	1m 05s

Sessions counts the visits that match the current filter and shows the change against the previous period, or New when the page had none then. Top Source names the source that sent the most of them and its share. Click a row to open the breakdown for that page, which lists its five biggest sources with their sessions and share and marks the AI ones, so a page reached from several assistants shows how they compare. Click a column header to sort, and use the filter box to find a page by its path. The list pages through its first 10,000 rows; on a site with more, a note under the table says so, and the filter box or a narrower Source brings the rest within reach. If the list cannot be loaded, Try again asks for it once more.

Pages that changed#

The Your pages on Google card also tracks which of your pages are appearing in search from one period to the next, because a page that stops appearing does not show up as a bad ranking. It simply leaves the report, and nothing else on the screen would tell you it had gone.

Alongside Opportunities and Best performing, three more views answer what changed. Lost lists pages that were showing in the previous period and are not showing now. Fading covers pages that still show, but far less often than they did. New covers pages that started showing during this period. The number on each of those three is the real total for the period, and the card says which two periods it is comparing.

Your pages on Google

Lost

showed before, not showing now

Fading

still showing, far less often

New

started showing this period

Where a page has been lost, Seerly looks at your most recent completed Smart Audit and tries to say why: the page returns an error, it carries a tag telling search engines not to list it, it points at a different page as the canonical version, or it now redirects somewhere else. When the audit found nothing wrong, it says that too, rather than inventing a cause.

When the page itself is fine, Seerly says what became of its searches instead: another page of yours now serves most of them, the searches stopped being made, or they are still being made and somebody else answers them. Two of your own pages competing for the same searches is the most useful of those to catch, because it is invisible from either page's own numbers.

Some lost pages will say the last audit did not cover them. Audits crawl to a set depth and a set number of pages, so a page buried several clicks deep may never have been fetched. That is not a finding about the page, and Seerly says so rather than implying it checked and found nothing.

Lost pages also show the searches they used to bring in, so you can see whether a page took a whole topic with it rather than a stray query or two.

Interpreting results#

A page that stopped appearing has not necessarily been removed from Google's index. Seerly can see that it stopped showing up for the searches it used to show up for, which is a different and weaker claim, and the wording on the screen stays inside what the data supports.

Reasons come from your last completed Smart Audit, so they describe your site as it was when that audit ran. If the audit is older than the period being reported on, the screen says so. Running a fresh audit before acting on a reason is usually worth the few minutes.

Two situations are reported rather than guessed at. If most of your pages stop appearing at once, that points at one site-wide cause, and Seerly says so instead of listing every page separately. If your site has more pages than Search Console will return in one request, lost and new pages are not shown at all for that period, because a page nobody could fetch is not the same as a page that went missing.

Very large sites and very quiet ones both need patience here. A page needs a small amount of real search activity in the earlier period before it can be called lost, which keeps the list free of pages that only ever appeared once.

A country here means where the searcher or visitor was, not which Google domain they used, and for traffic it is worked out from the visitor's IP address. Query-level tables leave out the queries Google anonymises, so with a country selected they can total less than the headline clicks and impressions, and the gap is wider for smaller countries. The headline figures are unaffected. Google Analytics also withholds figures for very small audiences, which is why a small country over a short date range can show Not enough data. A longer range or a larger country usually clears it. For the same reason the sessions listed under Landing Pages can add up to slightly less than the headline figure for the same source. The list also leaves out visits Google Analytics recorded without a page, since there is no page to show for them.

Expect AI referral traffic to be small, and do not read that as failure. A great many AI conversations end with the person satisfied and no click to anybody, so low referral volume sitting alongside strong visibility is a perfectly normal shape. Being the brand that gets recommended has value whether or not it produces a session that week.

Read direction rather than absolute numbers while you are here. Attribution of AI referrers is imperfect across the whole industry, not just in this product, so growth in AI-referred sessions tells you something real while the precise count does not.

The most useful thing you can do on this screen is compare the same page across both surfaces. A page that performs well in Google search and never gets cited by an AI engine is a specific and diagnosable situation rather than a mystery, and the cause is usually structural. Smart Audit is where you go to look at it.

Best practices#

• AI-referred sessions trending upward as visibility improves, even at small volume.
• Published content showing up in both search performance and citations as time passes.
• A sensible lag between the two. Visibility moves first and referrals follow; the two moving together is unusual enough to be worth a second look.

Next steps#

1. Connect Google Analytics early, so you have a baseline in place before you start changing anything.
2. Judge this over months rather than weeks. AI referral volumes are small enough that a weekly reading is mostly noise.
3. Go after the mismatches. Pages that rank in search but never get cited are the most informative rows on the screen.
4. Resist judging Seerly on referral traffic alone, since a good deal of the value sits in being recommended during a conversation that never produces a click at all.

What to keep in mind#

Analytics can only show you what your connected tools can see. Conversations that end without a click leave no trace anywhere, and that describes a large share of AI usage.

Attribution of AI referrers also depends on what referrer information the assistant chooses to pass along, and that varies between platforms and changes over time without notice.

Location filtering depends on what each source reports. Google Search Console and Google Analytics provide countries and Bing does not, so Bing data is only ever shown unfiltered.