\n<\/p>\n
Data sampling is\u00a0widespread\u00a0in\u00a0analytics\u00a0and is\u00a0standard practice for\u00a0several of the major players.\u00a0It\u2019s important to be aware of the potential dangers of sampling and\u00a0negative\u00a0financial impact it can have on your organisation.<\/em>\u00a0<\/p>\n \u00a0<\/p>\n Data sampling is the\u00a0practice\u00a0of\u00a0analysing\u00a0a subset of your traffic data,\u00a0which\u00a0is used to estimate the overall results.\u00a0Instead of\u00a0gathering\u00a0all the data,\u00a0you only get access to a limited\u00a0sample,\u00a0meaning that\u00a0any\u00a0analysis\u00a0you carry out\u00a0after that will be an assumption based on\u00a0existing\u00a0patterns.\u00a0\u00a0<\/p>\n The aim of\u00a0data sampling is to\u00a0speed up reporting time while still being able to\u00a0uncover\u00a0all\u00a0the meaningful\u00a0and valuable\u00a0information in the larger data set.\u00a0<\/p>\n \u00a0<\/p>\n Data sampling\u202fmethods\u00a0are split\u00a0into two categories: probability and non-probability\u00a0sampling:\u00a0<\/p>\n \u00a0<\/p>\n The free version of\u00a0Google Analytics\u00a0uses probability sampling,\u00a0and\u00a0your data is aggregated and delivered to you as a random data set.\u00a0This means that the standard reports they provide, including the Audience, Acquisition,\u00a0Behaviour\u00a0and Conversion reports,\u00a0are all based on sampled data.\u00a0GA data is also sampled when you create a custom report.\u00a0It\u2019s impossible\u00a0be sure if your reports are\u00a0displaying the overall traffic and\u00a0any\u00a0meaningful trends, or if the selected set is\u00a0giving you\u00a0accurate\u00a0information.\u00a0And downstream\u00a0the lack of visibility hinders decision making and has a direct impact on\u00a0business\u00a0efficiency\u00a0\u2013 especially for\u00a0larger organisations.\u00a0This is\u00a0also\u00a0why\u00a0Google\u00a0encourages\u00a0users to upgrade to their premium offer.\u00a0\u00a0<\/p>\n \u00a0<\/p>\n In statistics, the\u00a0standard rule is that\u00a0whenever a population of\u00a0behavioural\u00a0data is studied, a sample must be\u202frepresentative.\u00a0If you limit that sample,\u00a0you might\u00a0not be able to see real patterns occurring due to the data already being predicted\u00a0and\u00a0could\u00a0miss out on opportunities you would\u00a0otherwise have\u00a0noticed if you were given the whole\u00a0picture.\u00a0\u00a0<\/p>\n Example<\/strong>:\u00a0If your\u00a0site generates 50 million hits\u00a0on average\u00a0per month and 50,000 visits a day, sampling\u00a0can\u00a0limit\u00a0you\u00a0to 10 million hits per month and\u00a010,000 visits a day\u00a0or less.\u00a0This makes it impossible to\u00a0obtain a decent\u00a0representation\u00a0of all the data,\u00a0and\u00a0the\u00a0more\u00a0your website grows, the more inaccurate your reports will become.\u00a0<\/p>\n \u00a0<\/p>\n Sampling\u00a0also\u00a0doesn\u2019t account for cumulative data as the sample is different every day. This means that cumulative results are not displayed\u00a0either\u00a0for the month, quarter or year.\u00a0Here\u00a0are\u00a0a couple of\u00a0practical\u00a0examples:\u00a0<\/p>\n Imagine your production department releases updates on Wednesday and Friday at 5pm including flash offers. On Wednesday, if you reach your\u00a0sample\u00a0quota at 6pm, your updates will only partly be taken into consideration. On Friday, if you reach your quota\u00a0at 4pm, your updates\u00a0will\u00a0not\u00a0be\u00a0considered at all,\u00a0even though the Internet\u00a0behaviour\u00a0of visitors to your site at 5pm is considerably different to those who visit it at 4pm.\u00a0<\/p>\n If you then publish your\u00a0sales newsletter on\u00a0a\u00a0Tuesday morning,\u00a0it\u00a0will be\u00a0impossible\u00a0to compare\u00a0Tuesday\u2019s sample (reached at 11am) with\u00a0(and added to)\u00a0Wednesday\u2019s\u00a0or Friday\u2019s sample.\u00a0You simply won\u2019t be able to draw any meaningful\u00a0insight based on:\u00a0\u00a0<\/p>\n This can also apply to\u00a0the total number of cumulative hits for\u00a0a given month. For example, if in\u00a0November\u00a0you only\u00a0retain\u00a010 million hits out of 20 million and\u00a0in\u00a0December\u00a0only\u00a010 million hits out of 100 million, the 20 million hits retained\u00a0are clearly not\u00a0representative of the total of 110 million.\u00a0It\u2019s also not possible to draw an average number of hits.\u00a0<\/p>\n \u00a0<\/p>\n Now imagine your history displays 14 million hits and 360,000 visits.\u00a0You\u00a0will only be\u00a0able to\u00a0collect\u00a070% of the data to respect\u00a0your sampling\u00a0quota.\u00a0This can\u00a0have a notable\u00a0effect\u00a0with\u00a0seasonal variations.\u00a0For\u00a0example,\u00a0if the traffic for the month of December is twice that of any other month, then a quota of 70% is too large. This figure will\u00a0thus\u00a0be reduced to 35%, meaning that data will not be collected once the 35% limit has been reached.\u00a0On the other hand,\u00a0if\u00a0February is a weak month (half of a normal month) then there is no point in sampling since the real value is less than the quota.\u00a0<\/p>\n \u00a0<\/p>\n Your analytics solution should be able to collect and measure every single interaction a user has with your digital platforms, at any moment, all the time. And during periods of heavy traffic requiring strategic analysis (such as during sales or major events), it\u2019s even more critical that your solution can capture all data, without missing a beat.\u00a0<\/p>\n Let\u2019s say you\u2019re running an important promotion and your campaign includes TV spots to drive traffic to your website. In the minutes following your ad broadcast, your site receives a huge surge in visits, but your analytics solution\u2019s collection server can\u2019t handle the traffic volume and\u00a0inevitably\u00a0fails. Not only are you missing\u00a0a\u00a0big\u00a0piece\u00a0of\u00a0data,\u00a0but\u00a0that\u00a0piece\u00a0just\u00a0happens\u00a0to\u00a0be\u00a0a\u00a0crucial\u00a0one\u00a0as\u00a0well,\u00a0because\u00a0it\u00a0represents a\u00a0goldmine\u00a0of\u00a0information\u00a0indicating\u00a0whether\u00a0your\u00a0TV\u00a0ad\u00a0is\u00a0driving\u00a0the\u00a0intended\u00a0results\u00a0and\u00a0how\u00a0well\u00a0it\u2019s generating ROI. You now have an incomplete and,\u00a0therefore,\u00a0inaccurate view of your campaign performance because of sampled data.\u00a0<\/p>\n Your data must be complete and rich enough to answer very specific questions from all different departments of your company, such as:\u00a0\u00a0<\/p>\n \u00a0 \u00a0<\/p>\n Using small, sampled data sets\u00a0can significantly undermine\u00a0decision-making\u00a0within your\u00a0organisation.\u00a0Although\u00a0sampled data can highlight general trends, the smaller your sample, the less representative it is of the truth. This is particularly the case when\u00a0carrying out\u00a0granular analysis on small, sampled data sets.\u00a0<\/p>\n In order for your data-driven decisions to be truly accurate, they must be based on data that is complete, comprehensive and sufficiently rich. Your analytics tool must therefore collect all necessary data, and also provide the right processing and enrichments that will enable you to translate this data into action. When data is missing or corrupt, you risk\u00a0making strategic decisions\u00a0based on skewed information that doesn\u2019t fully reflect reality.\u00a0\u00a0<\/p>\n AT Internet\u00a0doesn\u2019t use\u00a0any\u00a0sampled data\u00a0whatsoever,\u00a0which allows you to act in the confidence that your decisions are based on complete,\u00a0reliable and accurate\u00a0information.\u00a0<\/p>\n \u00a0<\/p>\n Data sampling is\u00a0widespread\u00a0in\u00a0analytics\u00a0and is\u00a0standard practice for\u00a0several of the major players.\u00a0It\u2019s important to be aware of the potential dangers of sampling and\u00a0negative\u00a0financial impact it can have on your organisation.\u00a0 \u00a0 What is data sampling in analytics?\u00a0 Data sampling is the\u00a0practice\u00a0of\u00a0analysing\u00a0a subset of your traffic data,\u00a0which\u00a0is used to estimate the overall results.\u00a0Instead of\u00a0gathering\u00a0all the data,\u00a0you only get […]<\/p>\n","protected":false},"author":1,"featured_media":33030,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12033],"tags":[11603,248,7322,17165],"dealstore":[],"offerexpiration":[],"class_list":["post-33029","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-analytics","tag-data","tag-quality","tag-quantity","tag-sampling"],"yoast_head":"\nWhat is data sampling in analytics?<\/strong>\u00a0<\/h2>\n
How does data sampling work?<\/strong>\u00a0<\/h2>\n
\n
\n
What are the limits of data sampling?<\/strong>\u00a0<\/h2>\n
1. Representative samples\u00a0<\/strong><\/h3>\n
\u00a02. Limited sample\u00a0quotas<\/strong>\u00a0<\/h3>\n
Example 1:\u00a0Data collection\u00a0cut-off\u00a0once you have reached your\u00a0sample\u00a0quota<\/strong>\u00a0<\/h4>\n
\n
\u00a0<\/li>\n<\/ul>\nExample 2:\u00a0Using\u00a0a\u00a0percentage\u00a0of sampled data<\/strong>\u00a0<\/h4>\n
The importance of a comprehensive data set<\/strong>\u00a0<\/h2>\n
\n
If your data lacks a certain layer of information, such as geolocation data or information about the device used, you\u2019re missing out on a valuable piece of the picture.\u00a0<\/p>\nWhich solutions can you use to avoid data sampling?<\/strong>\u00a0<\/h2>\n
The five criteria of comprehensive data<\/strong>\u00a0<\/h3>\n
\n
\n
\n
\n
\n