Search Console Average Position vs Rank Tracker Positions: Why They Differ
Michael Carter · · 11 min read · Last updated

Key takeaways
- Search Console average position is the topmost position your site held in each search with an impression, averaged over impressions, across all countries and devices unless you filter.
- A position is recorded only when your link gets an impression, so a result on page 3 counts only for searchers who open page 3, and days spent on page 2 barely move the average.
- In Search Console every result element with a link takes one position, including AI Overviews and video carousels, while ads take none, so an organic-only rank tracker can show a better number.
- In our illustration, a tracker's daily US desktop checks had a median of 4 and a mean of 6.3, while Search Console showed 4.5 for all traffic and 4.4 filtered to US desktop.
- To reconcile the two, filter Search Console to the tracker's exact query, country and device, drop days without impressions, and compare medians over the same 28 days.
Search Console average position is the topmost position your site held in each search where it got an impression, averaged over all those impressions, across every country and device unless you filter. A rank tracker records one results page from one place, on one device, at one moment. The two measure different things, so they rarely match.
This guide is for SEO teams and developers who must explain the gap in a report: Google's counting rules, a worked week of numbers, a real tracker snapshot, which number to report and how to reconcile the two.
Why Search Console average position and rank trackers disagree
They answer different questions. Google Search Console answers "where did real searchers see my site, on average?" A rank tracker answers "where does my site sit on this page, in this market, right now?" Google itself calls position "a complex metric that can be misleading if you don't understand the subtleties."
Our guide to keyword rank tracking covers Google's definition and its 2.5 example. This post goes deeper: which searches count, how SERP features take positions and how the report's tabs change the number.
How Search Console average position is calculated
Search Console takes your highest place in each search where you got an impression and averages it over all impressions in the date range and filters. Google's help page on impressions, position and clicks, read on November 17, 2026, states both halves:
- The value is "the topmost position occupied by a link to your property or page in search results, averaged across all queries in which your property appeared."
- The metric "averages the position value for all impressions (because the position of the link will be different each time it is seen)."
So the average is weighted by impressions: sum of topmost positions ÷ number of impressions. A query seen 1,000 times weighs a hundred times more than one seen 10 times.
What counts as an impression
An impression is counted "whenever an item appears in the current page of results, whether or not the item is scrolled into view." In carousels and expanding blocks, the item "must typically be scrolled into view within the carousel, or expanded by a click, to register an impression."
The rule that matters most: "A link must get an impression for its position to be recorded." Google's example is a result on page 3 when "the user only views page 1"; then "its position is not recorded for that query." A tracker checking the top 30 records a position 25 every day; Search Console records it only for searchers who open page 3.
How SERP features take positions
Search Console numbers result elements, not just blue links: "Each element in Search results occupies a single position, whether it contains a single link or many different links or child elements." It counts down the main column, then the side column. How Google's help page treats common blocks on the SERP:
| Block on the results page | Position in Search Console |
|---|---|
| Organic result | One each |
| Featured snippet | One, "(typically)" position 1 |
| AI Overview | One for the block, shared by every link in it |
| Carousel of pages, such as videos | One for the whole carousel |
| Carousel of links that open a new Google search | None |
| Knowledge panel on the right | One, after the main column |
| Ads | None: "ads do not occupy a search position" |
Blocks above you push your number down: under an AI Overview and a video carousel, the third organic result is position 5. A link inside a block takes the block's position, so a citation in an AI Overview at the top counts as 1.
Property vs page: why the number changes with the tab
Search Console aggregates by property or by page, and Google's About the data page spells out the difference:
- By property: "All results for the same query that point to the same Search Console property are counted once in total," and "The topmost position of your property in Search results is reported."
- By page: "Each unique URL in a Search result is counted separately," and "The topmost position of that specific page in Search results is reported."
The chart always aggregates by property; in the table, the Pages and Search appearance tabs aggregate by page and the others by property. In Google's example, three pages from one pet store fill the results for "fun pets for children": by property, one impression at position 1; by page, one per URL, averaging (1 + 2 + 3) ÷ 3 = 2.
Filters, freshness and missing queries
Web search is the default; Image, Video and News are separate search types, and Search Analytics "does not combine the data across search types." Country means "the country where the search originated" and devices are desktop, tablet or mobile (dimensions and data groupings). Three limits matter when you compare:
- Freshness. "Collected data is usually available in 2–3 days" (data discrepancies), the newest days can still change, and days run in Pacific Time.
- Anonymized queries. Queries "that aren't issued by more than a few dozen users over a two-to-three month period" never get a row, though they stay in chart totals (Search Central blog). A rare keyword you track may be missing.
- Row limits. The report's table shows at most 1,000 rows.
Worked example: one week, two instruments
The numbers in this section are an illustration, not real site data. One page, one keyword: a tracker checks it daily at 06:00 UTC from the US on desktop, while Search Console sees every search that week, from every country and device, if it produced an impression.

The tracker's seven checks read 4, 4, 11, 5, 4, 12 and 4: mean 6.3, median 4. Search Console for the same query and week:
| Who searched | Searches | Impressions | Topmost position |
|---|---|---|---|
| US desktop, 5 days on page 1 | 300 | 300 | 4.3 |
| US desktop, Wednesday and Saturday (page 2) | 120 | 6 | 11.5 |
| US mobile | 700 | 700 | 5.0 |
| United Kingdom, all devices | 250 | 250 | 2.0 |
| Other countries | 120 | 120 | 7.0 |
| All | 1,490 | 1,376 | 4.5 |
The report shows (300 × 4.3 + 6 × 11.5 + 700 × 5.0 + 250 × 2.0 + 120 × 7.0) ÷ 1,376 = 4.5.
- The page-2 days nearly vanish. They are 2 of 7 tracker checks but 6 of 1,376 impressions: 114 of 120 US desktop searches on those days never reached page 2.
- Other markets pull both ways. The UK at 2.0 improves the average; mobile and other countries worsen it.
- Filtering brings the numbers together. Filtered to US desktop, Search Console shows (300 × 4.3 + 6 × 11.5) ÷ 306 = 4.4: close to the tracker's median of 4, far from its mean of 6.3.
How rank trackers measure position
A rank tracker runs the search itself with fixed settings: one keyword, one country (some tools also a city), one device, no search history, a scheduled time. It stores the topmost position of your domain or of one URL within a set depth, or "not found". Some tools number organic results only, others every block.
To show what a tracker stores, we ran a fresh check with BestSERP, a Google SERP API, on November 17, 2026: US, English, desktop, top 10, tracking google.com and its subdomains. It was one `/search` request with noCache (1 credit, about 4 seconds). The trimmed output:
2026-11-17 16:07 UTC | 'search console average position' | us, en, desktop, top 10
blocks: 8 organic, 3 videos, 0 top stories
google.com at 3: https://support.google.com › webmasters › thread › aver... | Average position in search console
google.com at 5: https://support.google.com › webmasters › thread › goog... | Google Search Engine Console Average Position and ...
google.com at 7: https://support.google.com › webmasters › thread › is-av... | Is average position being sunset for Search Console?
tracker records: 3 (topmost)
What the record shows, and what it leaves out:
- The topmost result wins. support.google.com held 3, 5 and 7; the tracker stores 3. Search Console's property view would also count one impression, at the topmost position; its Pages tab would list three URLs. When the domain is absent, the record is "not in top 10".
- Blocks change the count. BestSERP numbers organic results on their own and returns the video carousel as a separate list without its place on the page; it does not return ads, AI Overviews, featured snippets or the knowledge panel. If the carousel sat above position 3, Search Console would record position 4 for this search.
- One check is a sample. A check of the same query five minutes earlier shared only 2 of its 7 sites with this one and had no video carousel, though support.google.com was again topmost at 3. One check cannot tell whether Google served different results or the collector got an altered page.
Five reasons the numbers differ
Each reason pushes the numbers apart in a known direction:
| Reason | Search Console | Rank tracker | Usual effect |
|---|---|---|---|
| 1. Who searched | Everyone, all countries and devices unless filtered | One country, one device, no history | Either way |
| 2. Which searches count | Only those with an impression; page 2 or 3 counts only when opened | Every check, to its depth | Search Console looks better when you sit on page 2 part of the time |
| 3. When | Every search in the date range | One check a day | The tracker swings more |
| 4. What takes a position | Every element with a link, including carousels and AI Overviews; not ads | Often organic results only | Search Console looks worse with blocks above you |
| 5. What is grouped | Topmost of the property, or of each URL on the Pages tab | Topmost of your domain, or one URL | The property view looks better when several pages rank |
Data lag and anonymized queries add noise on top. Location deserves its own check: checking rankings in another country shows how far results move between markets.
Which number to report
Report Search Console for real visibility and traffic, a tracker for exact positions in one market and against competitors, and label each with its settings. They are two measurements, not two estimates of one truth.
| Question | Number to use |
|---|---|
| Is our search visibility growing? | Search Console clicks, impressions and average position per query or page, over 28 days or more |
| Where do we rank for money keywords in one market? | Tracker: median of daily checks for that country and device |
| Are we gaining on competitors? | Tracker: positions and SEO share of voice |
| Did something break this week? | Tracker for the drop, Search Console clicks two to three days later |
| Keywords with few searches | Tracker: few impressions make an unstable average |
Three rules for client reports:
- Never mix the two in one column or average them. Label each: "Search Console average position: web, all countries and devices, 28 days" and "Tracked position: US, desktop, median of 28 daily checks."
- Do not report the site-wide average as a KPI. New visibility makes it worse: a site with 2,000 impressions at position 3 that gains 1,000 impressions at position 45 averages 17.0 without losing anything (an illustration). Use per-query rows, position bands, clicks and click-through rate.
- Watch movement. Google recommends that you "monitor change in position over time, particularly sudden position changes, as well as absolute position."
How to reconcile Search Console with a rank tracker
Give Search Console the tracker's settings, then compare medians over the same 28 days:
- Note the tracker's settings: country, device, depth, check time, and whether it counts every block.
- Open the Performance report, keep Search type on Web and choose the last 28 days.
- Filter to the exact query: Query, Custom (regex),
^your keyword$. Without the anchors, the expression matches every query containing the phrase. - Filter to the tracker's market: its country and device, for example United States and Desktop.
- Export daily positions from the Dates tab with Daily granularity. Delete zeros: values shown as "~" or "-" "will be zeros in the downloaded data," and mean a day without impressions.
- Line up the tracker's checks for the same dates; Search Console's days run in Pacific Time.
- Compare medians, for example
=MEDIAN(FILTER(B2:B29, B2:B29>0))in Google Sheets. A median ignores the page-2 days that impressions hide and the odd bad tracker page, which is why the worked example's 4 and 4.4 agree. - Investigate a gap of more than a position or two: check the Pages tab for another of your URLs, look for blocks above you on the live page (the free SERP Checker shows the top 20 for any country), and count the impressions behind the median.
To automate this, the Search Console API returns the same rows (query, country, device and date, with clicks, impressions and position).
Next steps
Search Console average position shows what real searchers saw; a tracked position gives a fixed, repeatable view that includes competitors. Both are useful once each carries its settings.
- Store your own daily checks: how to build a rank tracker in Python.
- Run tracking on BestSERP: the rank tracking use case has the setup and costs.