Data Explorer enables you to read and export any data collected (hits) or any data computed (metrics) by AB Tasty. It can also be used to create payload for Data Explorer API (note that advanced filters combination is only available through the API for now).
Main capabilities
-
Export metrics for one or multiple campaigns
-
Export metrics with advanced breakdown
-
Export more specific metrics thanks to metric filters and population filters
-
Export raw hits received by AB Tasty
-
Export all time NPS hits (answers) for one or multiple campaigns
-
Switch from one account to any other you have access to
-
and much more…
Introduction to Data Explorer
Data Explorer provides you with 2 tabs: "Explore metrics” and “Explore hits”, they respectively allow you to export :
-
Metrics: Data computed by AB Tasty like the one you can access in the reportings (ex: conversion rate, revenue, actions count, NPS average…)
-
Hits: events triggered by your visitors and collected by AB Tasty, each hit corresponds to a unique action triggered by a unique visitor at a unique timestamp (ex: visitor X did view page Y at Z date, visitor X did click on Y at Z date…)
Each tab allows you to easily export either metrics or hits.
You can also export “Hits” via the “Metrics” tab by selecting “Hits count” as a metric and “Hit id” as a dimension. This solution corresponds to the Data Explorer API technical behavior.
Exporting metrics
-
Select the “Explore Metrics” tab
-
Pick the metrics you want in your result columns (ex: visitors count, transaction revenue total, conversion rate…)
-
You can filter metrics via the “funnel” icon to create more precise metrics that fit your needs
-
You can add additional metrics to your export via the “Add another metric” button
Make sure to always filter metrics with coherent filters. In most cases, you can only filter a metric by filters related to the metric type or general filters.
👉 Filtering a transaction metric by transaction property will work
👉 Filtering a transaction metric by a pageview property will return empty result
👉 Filtering a pageview metric by pageview property will work
👉 Filtering “NPS average” metric by “Page URL” will return empty results, as NPS answers are not related to a page URL
👉 Filtering “NPS average” metric by “Country (Iso code) will work as country is a general filter -
- Pick the dimensions that you want to breakdown your result metrics by (ex: country, browser, transaction shipping method…)
You can add as many dimensions as you want in this field
By adding a dimension, all data that is not related to this dimension will be excluded from the end results. Make sure not to use too many dimensions too early, as it can return empty results unexpectedly.
👉 Exploring the “Visitors count” metric grouped by the “Browser“ dimension will return all visitors count grouped by browser as all visitors have a browser.
👉 Exploring the “Visitors count” metric grouped by the “Transaction Currency” dimension will only return visitors count based on users who did a purchase as the “Transaction Currency” dimension is empty for visitors who did not do a purchase. -
Filter the data (learn more in “Filtering data” section) and optionally set result options
-
Check your quota usage (learn more in “Quota usage optimization“ section)
-
Run the export query.
Exporting hits
-
Select the “Explore hits” tab
-
Pick the hit properties you want for each hit
-
You can add as many hit properties as you want in this field
-
If you use incompatible hit properties, all hits will still be returned. Example: By selecting both “Transaction Id” and “Action tracker name (ea)” hit properties, no hit will both have a “Transaction Id” and a “Action tracker name (ea)” as these properties are not related to the same kind of hits. In that situation, transaction hits will be returned with an empty “Action tracker name (ea)” and action tracker hits will be returned with an empty “Transaction Id” property.
-
-
Filter the data (learn more in “Filtering data” section) and optionally set result options
-
Check your quota usage (learn more in “Quota usage optimization“ section)
-
Run the export query.
Filtering data
All filters available in the Data Explorer are hit-based filters. Users, sessions and campaigns don’t have properties in themselves, you can only filter users, sessions and campaigns based on related hits.
Period filters
You can filter data via pre-defined periods or on a custom period. The start and end dates are always included. It doesn’t impact user filtering or sessions filtering.
Campaign filters
When filtering on a specific campaign, all hits triggered in the same session as this specific campaign will be taken into account for metrics computing, not only hits explicitly related to this campaign.
Why such a behavior? For example, pageviews and action trackers available for the whole account are not explicitly related to a single campaign. By filtering campaign based on campaign activation during the session as we do, all hits related to the campaign can be explored.
For personalization campaigns only, you should also include campaign variation ids as some hits are directly attached to these variations and not the parent campaign. Learn more here.
User filters
When filtering users, all users who did trigger a hit matching your filters in their whole lifetime will be taken into account for metrics computing.
Session filters
When filtering sessions, all sessions who did trigger a hit matching your filters will be taken into account for metrics computing.
Hit filters
When filtering hits, all hits matching your filters will be taken into account for metrics computing.
Result options
Via this section, you can:
-
Sort results: display order can be picked based on one of your selected metrics.
-
Change result rows limit: lower limit may speed up query time. All hits are always taken into account, this option only reduces resulting rows.
-
Exclude QA, bots and corrupted data: this option is activated by default. It excludes hits tagged as QA hits, bots activity or corrupted hits (ex: hits with multiple allocation can’t be processed properly by our data pipeline) to only return safe trustworthy data.
-
Potential discrepancy with AB Tasty reporting results: AB Tasty reportings exclude Bots and corrupted data but didn’t exclude QA activity. It may explain some data discrepancies between reportings and Data Explorer results.
-
What does “corrupted” hit mean? It corresponds to a very specific edge case of misconfigured hits, for example we categorize hits that are assigned to multiple variations of the same test for a unique visitor as corrupted data. These scenarios should not happen and are just an additional shield to ensure AB Tasty data trustworthiness.
-
QA hits are the hits that have been generated in the QA Assistant context.
Quota usage optimization
Each account has a dedicated usage quota available each month. Quota tracking helps you build optimized requests and get results as fast as possible. Quotas are also a way for us to ensure Data Explorer’s reliability and efficiency. If you need additional quota, reach your CSM or KAM for more information.
Quota is only consumed when the query is launched.
What impacts quota usage?
-
Period filter: the most important parameter to reduce your quota usage. We advise you to test your query with a small period (ex: “yesterday”) first before picking the full period once your query is fully ready.
-
Metrics, dimensions and hit properties amount: more data = more quota used.
-
Data Filters amount: more filters = more quota used.
-
The following parameters do not impact quota usage at all: metric filters, results sorting, results limit.
Quota optimization best practices
-
Run test queries on short periods
-
Do not add too many dimensions too early
-
Keep an eye on query data cost to understand which options impact quota usage the most.
Known edge cases
Data discrepancy between reporting and Data Explorer
AB Tasty reporting and Data Explorer are using the exact same data sources. Depending on your settings, some discrepancies can be observed but are only due to mismatching configuration. The most recurrent ones are:
-
Data Explorer always return the most recent data updated live, AB Tasty Reporting updates its data on a regular basis but not as often as Data Explorer. If a campaign is running, Data Explorer will probably return more data than the Reporting as its data is more recent.
-
Data Explorer filters QA hits by default but AB Tasty reporting doesn’t. It may cause small differences with a few hits more displayed on your reporting than with the Data Explorer.
Filtering on a personalization campaign
For personalization campaigns only, you should also include campaign variation ids as some hits are directly attached to these variations and not the parent campaign.
To find the related variation ids, you can either go through the visual editor or the reporting.
-
On the visual editor, click on the “…” next to the variation name to open variation menu. Variation id is accessible at the bottom of this menu.
-
On the reporting, go to any goal and hover over the name of the variation. Variation id will appear and can be copied in a single click.