When you pull data from Google Trends at deep levels of granularity — for example, results split by one day — the results Google returns can vary widely.
This happens because Google Trends surfaces its results based on random samples of Google's actual search volume. This helps data move more quickly, but it also means that the random sampling from one query to the next can differ.
Additionally, Supermetrics caches this data. This can mean that you see different results in Google Trends and in your Supermetrics reporting, as the Supermetrics numbers will reflect the data in the API rather than in Google Trends' user interface.
Because it uses random sampling, there's no way to reduce discrepancies in Google Trends data when it's split at a very granular level.
To help reduce these inconsistencies, we recommend running queries with longer date ranges that split data into periods of a week or more.
Data instability issues
Google Trends doesn't always show reliable data, and can show zero values for some date ranges. Running the same query again the next day may sometimes produce different results with non-zero values for the same date ranges. This issue is due to a limitation of the Google Trends data source.