How to track rankings without fooling yourself

If you want to track rankings without fooling yourself, treat position as one clue, not the verdict. A ranking report can tell us whether a page is appearing for a query, roughly where it tends to appear, and whether visibility is moving over time. It cannot, on its own, tell us whether the page earned useful clicks, whether the visitor found what they needed, or whether the business gained an enquiry.

That matters because rank tracking is easy to over-read. A move from position 8 to 5 can look dramatic in a report and mean very little in practice. A fall from 2 to 4 can feel alarming while clicks hold steady because the search results page changed. Good SEO rank tracking is less about watching numbers twitch and more about checking whether the evidence supports a decision.

What rank tracking can tell you and what it cannot

SEO rank tracking is the practice of recording where a page or domain appears in search results for selected queries over time. Usually that means monitoring a defined keyword set, by location and device, to spot movement.

That is useful, but only within limits.

A ranking position is not the same thing as traffic. Search volume differs widely between queries. A page in position 3 for a low-demand term may bring fewer visits than a page in position 9 for a broader term. Position is also not the same thing as enquiries. A page can rank well, attract clicks, and still fail to help the user take the next step.

This is one reason we prefer to review rankings alongside impressions and clicks in Google Search Console, rather than as a standalone scorecard. Google itself does not promise rankings, and its own starter guidance focuses on useful, accessible content rather than guaranteed positions (Google SEO Starter Guide).

There is another complication. Search results pages are not ten plain blue links in a fixed order. They can include local packs, shopping results, image blocks, video results, AI features, People also ask, and other elements. So a reported “position 1” may not mean first thing a searcher sees. Equally, a “drop” may reflect a changed layout rather than a page becoming less relevant.

For local searches, the problem is sharper. In the UK, local visibility depends in part on relevance, distance and prominence, not just page optimisation (Google local ranking). So if you track “accountant near me” from an office in Ipswich and then again from a home in Woodbridge, you may not be measuring the same contest at all.

The practical lesson is simple. Rankings help answer questions such as:

• Are we eligible to appear for this query?
• Is our visibility broadly rising, flat, or falling?
• Did a page change coincide with a movement?
• Are we stronger on desktop than mobile, or vice versa?
• Are branded and non-branded searches behaving differently?

They do not prove:

• that more people visited
• that the right people visited
• that more leads came from search
• that the ranking change caused any business outcome

If you want a more grounded measurement framework, our Search Console evidence and limitations guide and content and measurement guide set out what each signal can and cannot support.

Which rankings are worth tracking in the first place

Most ranking dashboards are too large to be useful. They track every phrase a tool suggested, then bury the important movements under noise. We get better decisions from a short list of queries tied to page purpose.

Start with the page, not the keyword list.

Ask what the page is for. Common cases include:

• a service page intended to attract commercial enquiries
• a category or product page intended to support transactions
• an article intended to answer a question and earn relevant visits
• a location page intended to show local relevance
• a brand page intended to protect navigational visibility

Then choose a small set of queries that match that purpose.

For a service page, track the main service term, one or two close variants, and if relevant one location-qualified version. For an informational article, track the question or concept the piece is actually written to answer, not every adjacent phrase. For a location page, track queries that a real customer in that place would plausibly use.

Four filters help keep the list honest.

First, search intent. If the query suggests the user wants definitions, comparisons, reviews, maps, or direct suppliers, do not assume one page can satisfy all of those intents. A guide page and a service page often need different query sets.

Second, business relevance. A phrase may have attractive search volume and no practical value to the business. If ranking for it would not help the right audience find the right page, leave it out.

Third, location. Use location modifiers only where location meaningfully changes the result. “Solicitor Ipswich” and “emergency plumber near me” are location-sensitive. “How canonical tags work” usually is not. For UK businesses, this is often the difference between useful tracking and false precision.

Fourth, page ownership. One query should normally have one primary landing page in your tracking notes. If three internal pages all “target” the same query, your report can become impossible to interpret. You are no longer measuring one page’s visibility. You are watching internal competition and result rotation.

A practical starter set for a small site might be:

• branded queries
• 5 to 10 core commercial queries tied to important service or product pages
• 5 to 10 informational queries tied to important articles or guides
• a handful of location-sensitive queries, if local search matters

That is enough to spot patterns without creating a spreadsheet nobody reviews properly.

How to check whether a reported drop is real

When a rank tracker reports a fall, do not start rewriting the page immediately. First, check whether the drop is real, whether it affects impressions or clicks, and whether the page is still indexable.

A calm diagnostic usually goes in this order.

1. Check the page and query in Google Search Console

Open Performance and filter by the affected page, then by the query if you have one. Compare a recent period with a previous equivalent period. Look at:

• impressions
• clicks
• average position
• date trend

If position worsened but impressions and clicks are steady, the “drop” may not be commercially important. If impressions fell sharply as well, that is more likely to reflect a real visibility change.

Be careful with average position. It is an average across many impressions, potentially across different devices, locations and result layouts. It is a clue, not a precise placement log.

2. Confirm the URL is indexable and canonical

Use URL Inspection in Google Search Console for the exact URL you expect to rank. Check whether it is on Google, whether the selected canonical is the one you intended, and whether the page was crawled recently.

If the URL is excluded or another URL was selected as canonical, investigate before touching copy. Google’s Page indexing guidance explains that excluded pages need a reasoned review, and not every excluded URL should be indexed (Google Page indexing report).

Typical causes include:

• accidental noindex
• canonical pointing elsewhere
• redirect changes
• duplicate or near-duplicate pages
• soft 404 behaviour
• broken internal links
• significant rendering problems

It is also worth checking server logs if you have access. A drop after repeated crawl errors is a different problem from a drop after a content rewrite. When reviewing logs, distinguish between crawlers. Googlebot is Google Search’s crawler. OAI-SearchBot and GPTBot are separate from Google’s systems and have their own access and training implications. OpenAI documents that search access and model training controls are separate choices (OpenAI crawler documentation). Do not confuse those controls with Google Search indexing.

3. Look for changes on the page

Check what changed shortly before the drop. Useful questions include:

• Did the title tag change?
• Was the heading structure rewritten?
• Did internal links to the page reduce?
• Was important copy removed or moved behind tabs?
• Did the page template change on mobile?
• Was structured data added, removed, or broken?
• Did performance or rendering worsen?

This is where a simple page change log pays for itself. Without it, people tend to attribute every movement to “the algorithm” and miss their own edits.

4. Look at the search results page itself

Search the query in a clean browser, using the relevant location and device where possible. You are not trying to get an exact rank number. You are checking what the page is competing with now.

Look for:

• new local packs
• more video or image results
• AI features
• forum or community results
• stronger brand-dominated results
• fresh news or seasonal pages
• a different dominant intent

Google’s documentation on AI features states that existing search fundamentals still apply, and no special AI schema or text file is required for those features (Google AI features). So if a results page now shows AI elements, that does not automatically mean a technical omission on your site. It may simply mean the page layout changed.

5. Check whether the change is query-specific or wider

If one page drops for one query but holds for closely related queries, the issue may be intent alignment rather than a general quality problem. If several pages decline at once, look for sitewide causes such as indexing, internal linking, template changes, or tracking artefacts.

For a fuller step-by-step process, see our guide on how to investigate a change in search performance.

Why different tools show different positions

Different tools disagree because they are measuring different versions of the search experience.

Location is the obvious reason. Results can vary between towns, postcodes, or even searchers a short distance apart, especially for local intent queries.

Device also matters. Mobile and desktop layouts differ, and result features occupy different amounts of space. A page that appears “high” on desktop may be much lower on a mobile screen once maps, images or AI features are present.

Personalisation still plays a part in some contexts, particularly where previous activity, language settings or account state influence what is shown. Even when tools try to neutralise this, they do not reproduce every real-user condition.

Timing matters too. Search results are not frozen all month waiting for your rank check. Tools may collect at different times of day, on different dates, with different recrawl schedules and different data centres.

Result features complicate the count. Some tools count only organic listings. Some attempt to include SERP features. Some report absolute position, some “organic position”, some local pack visibility separately. If you compare those as if they were identical, disagreement is guaranteed.

Keyword sampling is another reason. A tool may report on your chosen query set, while Google Search Console reports on the actual queries that generated impressions. Those are not the same universe. One is a sample you asked to watch. The other is a record of what happened.

None of this means tools are useless. It means they are approximations. We should use them to spot patterns, not to claim exact truth. If one tool says position 6 and another says 9, the sensible response is not to argue about which one is “right” in the abstract. It is to ask whether the page’s visibility trend, impressions, clicks and page state support a decision.

A simple reporting method for monthly decisions

A monthly ranking review should reduce drama, not create it. We find a short evidence table works better than a glossy dashboard.

For each important page or query group, record:

• primary query or query cluster
• target page
• current average position or tracked range
• previous month comparison
• impressions
• clicks
• notable page changes
• SERP changes observed
• confidence level
• next action, if any

The confidence line matters. Sometimes the honest note is “uncertain, mixed signals”. That is better than forcing a neat conclusion from weak evidence.

A simple monthly review might read like this:

• Service page A, core commercial query, average position slightly down, impressions flat, clicks flat. Title tag changed mid-month. Local pack now appears more often. No action yet, monitor another month.
• Guide page B, informational query cluster, impressions up, clicks up, average position slightly improved. Added clearer subheadings and internal links. Keep structure, review related pages.
• Location page C, reported rank drop in tracker, Search Console stable, URL indexed, no page changes. Likely tool or location variation. No action.

This approach keeps rankings in context. It also creates a useful record when someone asks, three months later, “when did this start?”

If you are building this process from scratch, our search measurement baseline workbook is designed to help readers set down the first useful version rather than the most complicated one.

A final caution. Do not let monthly reporting become a search for certainty that the underlying data cannot provide. A crawler visit, a citation, an impression, a click and an enquiry are different kinds of evidence. Ranking data belongs in that chain, but it is not the whole chain.

That is the core habit we try to teach at The SEO Blog, published by Fleeta Limited in Ipswich. Learn what each metric can support, check the page before rewriting it, and leave room for uncertainty when the evidence is mixed. If you later need implementation help, that is the practical service side handled separately by Suffolk Web under the same owner. But the editorial standard here stays the same. Measure calmly, verify what changed, and do not mistake a neat chart for proof.