Checking your own rankings by searching Google is the fastest way to get a wrong answer. Your results are shaped by your location, your device, your history and whether you are signed in, so the position you see is yours and nobody else's.
This is the method that gives you numbers you can act on: what to track, how to set it up free, how often to look, and how to read the movement. If what you need is the report you send to a client afterwards, that is a different job and it is covered in our guide to the SEO keyword ranking report.
First, why manual checking misleads you
Four things distort a manual check, and knowing them is most of the skill.
Personalisation. Google adjusts results based on what you have clicked before. You visit your own site constantly, so Google shows it to you higher than it shows it to anyone else.
Location. Results differ by city and often by neighbourhood, especially for anything with local intent. Checking from your office tells you about your office.
Device. Mobile and desktop results differ, sometimes substantially, and mobile is the majority of searches for most categories.
Timing. Results shift through the day and vary between data centres. Two checks an hour apart can legitimately disagree.
If you must check manually, use an incognito window, add &gl=us&hl=en and a location parameter to the search URL, and treat the result as one sample rather than a fact. Better: use the free automated methods below, which check under fixed, stated conditions.
The four ways to track, compared
| Method | Cost | Accuracy | On demand | By city | Real clicks |
|---|---|---|---|---|---|
| Manual search | Free | Poor, personalised | Yes | No | No |
| Google Search Console | Free | Averaged but real | No | Country only | Yes |
| Dedicated rank tracker | Paid | High, fixed conditions | Yes | Yes | No |
| Grid-based local tool | Paid | High for map pack | Yes | By coordinate | No |
The column that decides it for most people is the last one. Only Search Console tells you whether a position produced anything, which is why the answer for nearly every site is Search Console first, and a tracker added on top when you need daily or city-level positions.
Step 1: choose 20 to 50 keywords, not 500
More keywords is not more information. A tracked set of 500 produces an average that moves for reasons you cannot act on, and most tools charge by keyword.
Pick your set from four groups. If the starting list is large, clustering it first makes the groups obvious, and we compared the keyword clustering tools that do it.
- Your money terms. The five to ten phrases that describe what you sell. These matter even when they are unwinnable this year, because you need to see the trend.
- The terms you already rank 5 to 20 for. Pull these from Search Console. They are your realistic upside and they belong in the tracked set for exactly that reason.
- The questions your customers ask. Informational terms that bring people in before they buy.
- Two or three competitor comparison terms. "Your category alternative", "competitor vs", and your own brand plus "review".
Exclude your plain brand name from the tracked set, or the average will look excellent forever while everything else declines.
Step 2: set it up free with Search Console
Search Console will do most of this job at no cost, and it is the only source that reports real clicks rather than a modelled position.
Connect it and wait. It backfills nothing, so the day you verify is the day your history starts. If you have not done this yet, it is the highest priority item on this page.
Use the Performance report with the right filters. Select Search results, set the date range to the last 28 days, and turn on Average position alongside clicks and impressions. Then filter by query to see individual terms.
Compare periods rather than reading absolutes. The comparison view against the previous 28 days is where the signal is. A position of 14.2 tells you little. A move from 18.9 to 14.2 tells you what to do next.
Export weekly. Search Console keeps 16 months, but exporting to a sheet lets you build the view you actually want, including position by page rather than by query.
The two limitations worth knowing. Positions are averaged across everyone who saw the result, so a term ranking 3 in London and 30 elsewhere reports as something in between. And you cannot ask it for today's position on demand, which is what a dedicated tracker is for.
Step 3: add a tracker when you need daily or local positions
Wincher is the affordable entry point, with scheduled email reports and enough accuracy for a small business. If you have never paid for tracking, start here.
AccuRanker is the accuracy option, with on-demand refreshes rather than a daily cycle. Priced per keyword, which rewards a focused set.
Nightwatch handles segmentation and location-specific tracking well, which matters if you serve several cities.
BrightLocal or Local Falcon if your visibility is in the map pack. Map pack position changes with the searcher's physical coordinates, so a grid-based check is the only honest measurement for local businesses.
Semrush and Ahrefs include position tracking inside a wider toolkit. Use it if you already pay for one; it is rarely worth buying for tracking alone.
Whatever you choose, fix the settings and leave them alone. Location, device and search engine must stay constant, because changing them mid-quarter makes your own history meaningless.
Step 4: pick a cadence and hold it
Weekly is right for most sites. Enough to catch a real drop within days, infrequent enough that you are not reacting to noise.
Daily only if you are in a fast-moving category, running a migration, or recovering from an algorithm update. Otherwise daily checking is a way of feeling busy.
Monthly for anything you report to other people. Positions swing enough week to week that a monthly view is the honest unit for a conclusion.
Step 5: read the movement, not the number
Four patterns cover almost everything you will see.
Positions 11 to 20 rising. Your best opportunity, every time. These pages are one genuine improvement away from page one, and the traffic difference between position 11 and position 9 is not incremental, it is a step change.
A sharp drop on one page. Usually a competitor updated their page, or you changed something. Check what ranks above you now and read it. Recovering a page you used to own is far cheaper than building a new one.
A gradual decline across many pages. An algorithm update or site-wide decay. Do not rewrite individual pages in response. Look for the pattern first, because the fix is usually structural.
Rankings flat, clicks falling. Increasingly common and widely misdiagnosed. It usually means an AI Overview or answer panel now sits above you, so the same position produces fewer clicks. Rewriting the page will not fix it, and the next section is what will.
Step 6: track your visibility in AI answers too
Position tracking assumed a page of ten blue links. A rising share of queries are now answered by ChatGPT, Perplexity, Gemini or an AI Overview, with sources cited and no click required. Ranking first for a query nobody clicks is not a win.
You can check this manually. Ask the assistants the ten buying questions in your category and record who gets named and which pages are cited. It takes an afternoon and it is usually a sobering exercise.
To track it properly, tools like Profound and Peec AI monitor how often your brand appears across assistants and which sources they pull from. That second output is the actionable one, because it turns an invisible problem into a list of third-party pages you can go and get mentioned on.
Treat this as a second tracked set beside your keywords, not a replacement. Both matter, and only one of them is currently being measured by most of your competitors.
Common mistakes
Tracking too many keywords. The average becomes meaningless and the bill becomes annoying.
Including your brand name. It ranks first, it never moves, and it drags every average upward.
Changing the tracking location mid-quarter. Your own history stops being comparable.
Reading positions without clicks. Position is a diagnostic. Clicks are the outcome. A report with only the first is how a programme looks healthy while it fails.
Comparing two tools' numbers. They check under different conditions and will always disagree. Pick one and stay with it.
Reacting to a single week. Below the top few positions, a three or four place swing week to week is usually nothing.
Frequently asked questions
How do I check my keyword ranking for free? Google Search Console gives you average position, clicks and impressions for every query you appear for, at no cost. The free rank checkers from Semrush, Ahrefs and Seobility give an on-demand position for a single keyword, which is useful for spot checks.
Why does my ranking look different in my tool than in my browser? Your browser personalises results based on your history and location, and you visit your own site more than anyone. The tool checks under fixed conditions, which is why it is the more reliable number even when it feels wrong.
How many keywords should I track? Twenty to fifty for a small site, a few hundred for a large one. Choose them deliberately from your money terms, your positions 11 to 20, your customers' questions and a few competitor comparisons.
How often should I check keyword rankings? Weekly for your own monitoring, monthly for anything you report to someone else. Daily only during a migration or a recovery.
Does Google Search Console show real rankings? It shows the average position across all the impressions your page received, which is a real measurement but not a single position. It is more honest than a tracker and less precise, and the two answer different questions.
Can I track rankings by city? Yes, with a tracker that supports location targeting, and you should if you serve several areas. For map pack visibility specifically, use a grid-based local tool, because position varies across a few miles.
How do I track whether ChatGPT mentions my site? Manually by asking it your category's buying questions, or with a dedicated AI visibility tool that monitors citations across assistants. Referrer data will not tell you, because most of these answers produce no visit at all.
Set it up this week
Connect Search Console if you have not, pick twenty keywords from the four groups above, and export the performance report every Monday. That is an hour of setup and it covers most of what a paid tracker would tell you.
The tracking is the easy half. The part that produces results is acting on the positions 11 to 20, every month, without fail. Distribb reads your Search Console data to find exactly those pages and then handles the rewriting, internal linking and publishing. More on the research side in our guides to keyword research automation and keyword gap analysis.