Why Your Google Ranking Is Different from Mine

Google rankings vary by query, location, device and search context. Learn how to compare observations and diagnose SEO problems with better evidence.

A client sees their retailer ranking in position 2. A colleague searches what appears to be the same phrase and sees position 7. The agency report says position 5.

That does not automatically mean someone has made a mistake. The three observations may simply have been taken under different conditions.

A Google ranking is not a fixed property of a keyword and a website. It is an observation made for a particular query, market, location, language, device, search environment and point in time. Google explains that search results can vary with factors including location, language, device, recent searches and query context. Google’s documentation on search-result differences describes these as ordinary influences on results.

Once rankings are treated this way, conflicting observations become easier to interpret. This article explains which differences are normal, why repeated manual searches are weak evidence and how to create a comparison that marketing, SEO and commercial teams can use.

A ranking is an observation, not a permanent number

When someone says, “We rank fifth for that keyword”, they usually mean that a particular page appeared fifth in a particular search result. The missing detail is everything around that search.

Where was the search made? Was it on a phone or desktop? Which country and language settings were active? What was the exact wording? Were shopping results, maps, videos or other search features present? When was the result collected?

Those details can affect both the results Google selects and the way they appear on the page. So “position 5” is only useful when you know what it describes.

It is a little like reporting a temperature without saying where or when it was measured. The number may be accurate, but it cannot explain the whole situation.

Why two people can see different results

Several variables can make apparently identical searches different. They do not all have the same effect, and none should be blamed automatically. The useful first step is to separate them.

1. Searcher location

Google can use location to interpret what is relevant. That is obvious for searches such as “accountant near me” or “Italian restaurant”, but location can also affect broader commercial searches where regional availability, delivery or service coverage matters.

Imagine a national retailer selling outdoor clothing. A desktop search from London might show one category page near the top, while a mobile search from Manchester produces a slightly different ordering. The difference could reflect regional relevance, competition, local result features or ordinary variation. It is not automatically evidence of a technical problem.

Location differences deserve particular attention for searches with local or regional intent. A difference between two cities is likely to be more consequential for “dentist” than for a broad product or information query, although neither search is guaranteed to produce identical results everywhere.

2. Device type

Desktop and mobile results are not perfectly interchangeable. Device context can affect what is useful, while the page layout changes substantially between devices.

A mobile result may be pushed further down by larger images, product carousels or expandable elements. A desktop result may appear alongside a different arrangement of shopping, video or knowledge features. Even where the underlying organic ordering is similar, practical visibility can differ.

That means a mobile position 5 and a desktop position 5 do not necessarily give a user the same experience. The number is the same. The page is not.

Google’s visual-elements documentation is useful when identifying the features occupying a results page, although it does not provide a universal measure of how prominent each feature is to users.

3. Language and country settings

Language, country and search settings help Google interpret which results are relevant. Someone searching in the UK with English-language settings may receive different results from someone using English in another country, even when the query wording is identical.

Small differences matter here. A spelling variation, a country-specific product name or a different currency can change the interpretation of a query. Before comparing rankings across markets, confirm that the search was made in the same country and language environment.

4. The exact query and its context

People often describe different searches as “the same keyword”. They might search running shoes, best running shoes and running shoes UK and treat the results as one observation.

They are related searches, but they may express different needs. One person may be browsing products, another may be looking for advice and another may be trying to find a specific retailer. Google may respond differently to each.

Check the exact words, spelling and modifiers before looking for a technical explanation. Also check whether recent searches or query context may have influenced the result. Related keywords are not interchangeable measurements.

5. Search features change what “position” means

A results page may include advertisements, shopping results, maps, images, videos, featured snippets, discussions or other elements. These features can appear above or between traditional organic listings.

As a result, a numerical position does not fully describe visual prominence. An organic listing in position 4 might be immediately visible on one results page but below several large features on another. A result counted in the same position can therefore receive a different amount of attention.

Search Console also has its own counting rules for impressions and position. Its average position is an aggregate across recorded impressions. It is not a live rank that every searcher sees. Google’s Search Console performance documentation explains how position is calculated and why the metric should be interpreted as a report-level measure.

That does not make Search Console less useful. It answers a different question: how a site appeared across its actual recorded exposure over a period, rather than what one person saw at one moment.

6. Timing and ordinary result variation

Search results change. Pages are published, competitors update content, links appear and disappear, ranking systems are adjusted, and search features are tested or rolled out. Google’s overview of how Search works provides the broader context: crawling, indexing and serving results are separate processes, and the web itself is continually changing.

Google has also documented that changes may take time to reach all of its systems. Small differences can therefore occur during updates or rollouts. That is a possible explanation, not a diagnosis. It is risky to label every disagreement “data-centre variation” without a controlled pattern showing that it is happening.

A result collected on Tuesday morning may not be identical to one collected on Friday afternoon, even when every other setting appears unchanged.

What about personalisation?

Personalisation exists, but it is often given too much credit.

Google can use elements of search history, recent searches and query context to make results more relevant. That does not mean every person receives a heavily customised organic result for every search. Location, language, device, query wording, SERP features and timing are often more straightforward explanations.

In practice, “Google personalised it” can become a convenient answer when nobody has recorded the search conditions. It may be possible, but it is not evidence on its own.

The sensible position is cautious: personalisation is one possible variable, not the default explanation for every difference. If two people see different results, first compare the observable conditions. If a difference remains after those conditions are controlled, personalisation and other forms of result variation may be worth considering. The available research on personalisation is also query- and study-specific, so historical percentages should not be presented as a current universal rate.

A worked example: one retailer, three rankings

Suppose a national retailer is checking the query waterproof walking jacket. Three observations are reported:

• A marketing director in London sees the retailer’s category page in position 2 on desktop.
• A colleague in Manchester sees a product page from the same retailer in position 7 on mobile.
• The agency’s rank tracker reports position 5 from a defined UK tracking location.

These are not yet contradictory measurements of one ranking. They differ in at least four ways: location, device, target URL and collection method. The exact query may also differ if one person searched waterproof walking jackets or used a different country setting.

The next step is not to decide which person is right. Record the conditions and repeat the comparison consistently. If the tracker always uses the same location and mobile device, its position 5 can be compared with future tracker observations. It should not be treated as a promise that every user in every city will see position 5.

Search Console can then show whether the site receives impressions and clicks for the query across relevant countries and devices, where the data is available. The business can also check whether the page appearing is the intended category page and whether visibility is producing useful visits, sales or other commercial actions.

Use a like-for-like comparison checklist

Before treating two rankings as comparable, record:

• Exact query: Include every word, spelling, modifier and location term.
• Market: Record the country and, where relevant, the city or region.
• Location: Note whether the search came from a specific physical location, a neutral location or a tracking location.
• Language and settings: Check language, country, currency and relevant search settings.
• Device: Separate desktop, mobile and tablet observations.
• Date and collection window: Record when the result was collected. For monitoring, use a defined and consistent schedule.
• Search environment: Note whether the search was logged in, whether unusual recent-search context may have influenced it and whether the query was entered directly or reached through a suggestion.
• SERP features: Record shopping, maps, videos, featured snippets and other prominent elements, rather than reporting only the blue-link position.
• Target URL: Check which page from the site appeared. A different URL may indicate an intent, relevance or site-architecture question even when the domain is visible.

You do not need to record every variable for every keyword forever. The point is to make the important conditions visible before drawing a conclusion.

Why repeated manual Googling creates confusion

Manual searches are useful for inspecting a results page. They can show what a customer might encounter, which competitors appear and whether a feature has changed.

They are a poor primary measurement method when repeated searches are used as a scoreboard. Each search may change the location, device, time, query context or result layout. The sample is also tiny: a handful of searches cannot represent everyone searching across a country.

There is a behavioural problem too. Teams can start optimising around whichever result they happened to see that morning. A single low position creates unnecessary alarm; a single high position creates false reassurance. Neither tells you whether the pattern is persistent or commercially important.

Use manual searching to investigate and understand a result, not to settle every ranking dispute.

For a broader discussion of what rank tracking can and cannot tell you, see Does keyword rank tracking still matter?

What evidence should you use instead?

Start with controlled rank tracking

A rank tracker is useful for identifying repeatable visibility patterns across a defined keyword set. Set the location, device, market and collection schedule, then compare like with like over time.

This can reveal a persistent gap between markets, a repeated change across many queries or a consistent shift in the page Google selects. It still represents a sample, though. A tracker cannot reproduce every individual search, and providers may count SERP features differently.

Use Search Console for actual site exposure

Search Console adds a different kind of evidence. It can show impressions, clicks, queries and pages, with dimensions such as country, device, date and search appearance where data is available. Google’s Performance report documentation describes these dimensions and the report’s aggregation limits.

Use it to ask whether the reported difference appears in the site’s broader exposure. Is there a sustained drop in impressions from one market? Does mobile receive fewer clicks? Is a different URL attracting impressions than the page the business expects?

Do not use average position to settle a disagreement about one manual search. Treat it as an aggregate signal that helps reveal patterns.

SEO reports should explain what their numbers represent before presenting them as simple performance scores.

Finish with commercial relevance

A visibility difference matters when it affects something the business values. That might be qualified visits, enquiries, applications, sales or revenue.

Commercial outcomes cannot identify an exact ranking position, and they are affected by factors beyond SEO. They are still the right final test for deciding whether a difference deserves urgent attention.

A lower numerical position for a high-intent query may matter more than a higher position for a broad query that rarely produces useful visits. The number needs business context.

Which differences are worth investigating?

A one-off difference of one or two positions may be normal. It could reflect timing, averaging, layout changes, a small sample or an unrecorded change in search conditions.

Investigation becomes more justified when you see a repeatable pattern, such as:

• a persistent visibility gap in one country, city or device;
• the wrong page appearing across many related searches;
• a change reproduced across a controlled dataset and several collection dates;
• a corresponding shift in Search Console impressions or clicks;
• a difference that aligns with a meaningful change in qualified traffic or commercial performance.

These signals justify diagnosis. They do not identify the cause by themselves. The cause could involve search intent, competition, content relevance, technical delivery, site architecture or measurement settings.

The practical rule: compare like with like

No measurement method reproduces every individual Google search exactly. A manual search reflects one person’s conditions. A rank tracker samples a defined environment. Search Console aggregates the exposure Google recorded for a site.

The goal is not to discover one perfect universal position. It is to create consistent, decision-useful evidence.

Before diagnosing an SEO problem, compare the exact query, market, location, language, device, date, search environment, SERP features and target URL. Then look for a pattern across controlled tracking, Search Console and commercial outcomes.

When differences persist across markets, devices or large keyword sets and the cause remains unclear, the next step is structured diagnosis rather than another uncontrolled search.

For the wider question of how ranking evidence fits alongside traffic and business outcomes, read How to measure SEO beyond rankings.