Query your Google Search analytics data

You can run queries over your Google Search data to see how often your property appears in Google Search results, with what queries, whether from desktop or smartphones, and much more. You can use the results to improve your property's search performance, for example:

  • See how your search traffic changes over time, where it's coming from, and what search queries are most likely to show your property.
  • Learn which queries are made on smartphones, and use this to improve your mobile targeting.
  • See which pages have the highest (and lowest) click-through rate from Google search results.

Search query data is exposed using the searchanalytics.query() method. The query() method exposes all the data available in the Performance report in Search Console. Before running any queries, you should read the Performance report documentation to learn what data is exposed and what it means.

This page shows how to perform common queries with different request parameters.

Getting started

Verify the presence of data

Before running a query, you should first test for the presence of data in that time range. Omit filters, sorting, row limits, and any other parameters except start date, end date, and "date" as the only dimension.

Code

request = {
      'startDate': flags.start_date,
      'endDate': flags.end_date,
      'dimensions': ['date']
  }

Output

python search_analytics_api_sample.py 'https://www.example.com/' '2015-05-01' '2015-05-15'
Available dates:
Keys                              Clicks         Impressions                 CTR            Position
2015-05-01                       22823.0            373911.0     0.0610385893969        8.1829472789
2015-05-02                       16075.0            299718.0     0.0536337490574       8.14173322924
2015-05-03                       18794.0            337759.0      0.055643224903       8.07772405769
2015-05-04                       31894.0            468076.0     0.0681385074219        7.4104611217
2015-05-05                       34392.0            482919.0      0.071216912153       7.20689805123
2015-05-06                       35650.0            484353.0     0.0736033430164       7.11683214515
2015-05-07                       33994.0            465812.0     0.0729779395979       6.91755472165
2015-05-08                       27328.0            413007.0     0.0661683700276       7.22172747677
2015-05-09                       16637.0            297302.0     0.0559599329974       8.01876206685
2015-05-10                       19167.0            332607.0     0.0576265682923       7.87882696395
2015-05-11                       35358.0            499888.0      0.070731843933       7.11701821208
2015-05-12                       35952.0            486583.0      0.073886675038       6.80677294521
2015-05-13                       34417.0            480777.0      0.071586203167       6.86552185317
2015-05-14                       32029.0            457187.0     0.0700566726525       6.92575904389
2015-05-15                       27071.0            415973.0     0.0650787430915       7.27105605412

Try different dates

We see that we have data for that segment of time, so it's safe to move forward. It's important to do this before running your actual query. For example, running this same query for a different range returns this:

python search_analytics_api_sample.py 'https://www.example.com/' '2015-06-01' '2015-06-15'
Available dates:
Keys                              Clicks         Impressions                 CTR            Position
2015-06-01                       31897.0            468486.0     0.0680852789624       6.81207122518
2015-06-02                       32975.0            460266.0     0.0716433540605       6.62655942433
2015-06-03                       32779.0            459599.0     0.0713208688444       6.58126758326
2015-06-04                       30116.0            435308.0     0.0691831990223       6.71409668557
2015-06-05                       25188.0            380444.0     0.0662068530454       7.00998570092
2015-06-06                       14829.0            272324.0     0.0544535186028        7.6309910254
2015-06-07                       17896.0            318094.0      0.056260099216       7.56606223318
2015-06-08                       33377.0            487274.0     0.0684973957158       6.77552260125
2015-06-09                       33885.0            484241.0     0.0699754874123       6.70545451542
2015-06-10                       32622.0            466250.0     0.0699667560322       6.64417372654
2015-06-11                       31317.0            447306.0     0.0700124746818       6.61534832978
2015-06-12                       25932.0            393791.0      0.065852190629       7.15718998149
2015-06-13                       15451.0            275493.0     0.0560849095984       7.69994518917
2015-06-14                       18358.0            318193.0     0.0576945438775       7.34048517724

Look carefully, and you'll notice that the data ends on the 14th; no data for the 15th.

You might find it useful to use the APIs explorer in free-form edit mode to test your queries quickly (click the dropdown arrow on the side of the request body field and click "Freeform editor").

After you've verified the range of valid dates, you can start grouping by other dimensions, adding filters, row count limits, and so on:

Top 10 queries, sorted by click count, descending

Code

request = {
    'startDate': flags.start_date,
    'endDate': flags.end_date,
    'dimensions': ['query'],
    'rowLimit': 10
}

Output

Top Queries:
Keys                              Clicks         Impressions                 CTR            Position
seo                               3523.0            270741.0     0.0130124362398       5.86615252215
hreflang                          3207.0              5496.0      0.583515283843       1.10080058224
robots.txt                        2650.0             23005.0      0.115192349489       4.30367311454
301 redirect                      2637.0              7814.0      0.337471205529         1.621192731
googlebot                         2572.0              6421.0      0.400560660333       1.15823080517
google seo                        2260.0             11205.0      0.201695671575       1.38295403838
google sitemap                    1883.0              4288.0      0.439132462687       1.21175373134
canonical url                     1882.0              3714.0      0.506731287022       1.12762520194
sitemap                           1453.0             22982.0       0.06322339222       3.78074144983

Top 10 pages, sorted by click count, descending

Code

request = {
    'startDate': flags.start_date,
    'endDate': flags.end_date,
    'dimensions': ['page'],
    'rowLimit': 10
}

Output

Top Pages:
Keys                              Clicks         Impressions                 CTR            Position
https://www.example.com/21       10538.0             62639.0      0.168233847922       3.63031019014
https://www.example.com/65        9740.0             82375.0      0.118239757208       5.61003945372
https://www.example.com/15        9220.0            128101.0     0.0719744576545       5.32300294299
https://www.example.com/41        8859.0            426633.0     0.0207649197319       1.62309057199
https://www.example.com/53        8791.0            829679.0     0.0105956641062       14.4941887164
https://www.example.com/46        7390.0             82303.0     0.0897901656076        5.7723290767
https://www.example.com/27        7169.0             64013.0      0.111992876447       4.98709637105
https://www.example.com/80        6047.0             84233.0     0.0717889663196       4.10592048247
https://www.example.com/9         5886.0             59704.0     0.0985863593729        4.0897594801
https://www.example.com/8         5043.0             66869.0     0.0754161120998       4.57651527614

Top 10 queries in India, sorted by click count, descending

Note that the filter operator "equals" is omitted, as it is the default operator.

Code

request = {
    'startDate': flags.start_date,
    'endDate': flags.end_date,
    'dimensions': ['query'],
    'dimensionFilterGroups': [{
         'filters': [{
              'dimension': 'country',
              'expression': 'ind'
          }]
      }],
      'rowLimit': 10
  }

Output

Top queries in India:
Keys                              Clicks         Impressions                 CTR            Position
googlebot                          250.0               429.0      0.582750582751                 1.0
search console                     238.0             34421.0    0.00691438366114       1.00101682113
dns error                          189.0               850.0      0.222352941176       1.38470588235
google seo                         165.0               552.0      0.298913043478       1.04166666667
canonical url                      141.0               282.0                 0.5                 1.0
301 redirect                       132.0               557.0      0.236983842011       1.78276481149
google search console              126.0             16898.0    0.00745650372825       1.03929459108
robots.txt                         117.0              1046.0      0.111854684512        3.9206500956
canonical tag                      111.0               223.0      0.497757847534                 1.0

Top 10 mobile queries in India, sorted by click count, descending

Code

request = {
    'startDate': flags.start_date,
    'endDate': flags.end_date,
    'dimensions': ['query'],
    'dimensionFilterGroups': [{
        'filters': [{
            'dimension': 'country',
            'expression': 'ind'
          }, {
            'dimension': 'device',
            'expression': 'MOBILE'
       }]
    }],
    'rowLimit': 10
}

Output

Top mobile queries in India:
Keys                              Clicks         Impressions                 CTR            Position
search console                      26.0              1004.0     0.0258964143426       1.00298804781
dns error                           24.0               111.0      0.216216216216       1.27927927928
google seo                          18.0                69.0      0.260869565217       1.02898550725
eliminar                            16.0               134.0      0.119402985075                 1.0
googlebot                           11.0                24.0      0.458333333333                 1.0
404                                  9.0               214.0     0.0420560747664       8.64018691589
robots.txt                           9.0                40.0               0.225               4.025
google search console                8.0               438.0     0.0182648401826       1.04337899543
seo                                  8.0               111.0     0.0720720720721       4.96396396396

Query a slice of rows

You can query for a specific slice of rows by specifying a (zero-based) start row number and the number of rows to return. Specifying an invalid start row number will return an error, but specifying more rows than are available will return all available rows.

Top 11-20 mobile queries for the date range, sorted by click count, descending

Code

request = {
      'startDate': flags.start_date,
      'endDate': flags.end_date,
      'dimensions': ['query'],
      'dimensionFilterGroups': [{
          'filters': [{
              'dimension': 'device',
              'expression': 'mobile'
          }]
      }],
      'rowLimit': 10,
      'startRow': 10
  }

Output

Top 11-20 Mobile Queries:
Keys                              Clicks         Impressions                 CTR            Position
dns error                         1220.0             15064.0        0.0809877854       3.13448726206
google seo                        1161.0              7923.0         0.146535403       2.31479556195
sitemap                            926.0             12478.0        0.0742106107        5.8130025067
googlebot                          903.0              7822.0         0.115443621        4.6910285792
robots.txt                         799.0             24868.0        0.0321296445       5.92759215963
404                                520.0             12777.0        0.0406981295       5.80352636506
seo                                506.0              2925.0         0.172991453       2.50413960996
search console                     487.0               981.0         0.496432212       1.00036102455
canonical url                      326.0              4087.0        0.0797651089       3.23664971157
301 redirect                       261.0              3165.0         0.082464455       3.63074363869

Getting more than 25,000 rows

If your query has more than 25,000 rows of data, you can request data in batches of 25,000 rows at a time by sending multiple queries and incrementing the startRow value each time. Count the number of retrieved rows; if you get less than the number of rows requested, you have retrieved all the data. If your request ends exactly on the data boundary (for example, there are 25,000 rows and you requested startRow=0 and rowLimit=25000), on your next call you will get an empty response.

Top 1-25,000 mobile queries for the date range, sorted by click count, descending

Code

request = {
      'startDate': flags.start_date,
      'endDate': flags.end_date,
      'dimensions': ['query'],
      'dimensionFilterGroups': [{
          'filters': [{
              'dimension': 'device',
              'expression': 'mobile'
          }]
      }],
      'rowLimit': 25000,
      'startRow': 0
  }

Top 25,001-50,000 mobile queries for the date range, sorted by click count, descending

Code

request = {
      'startDate': flags.start_date,
      'endDate': flags.end_date,
      'dimensions': ['query'],
      'dimensionFilterGroups': [{
          'filters': [{
              'dimension': 'device',
              'expression': 'mobile'
          }]
      }],
      'rowLimit': 25000,
      'startRow': 25000
  }

Getting all your data

See Query all your search traffic.