Adobe Analytics reports allow us to analyse user behaviour without leaving the platform. However, some projects require a greater level of detail or a way to combine that information with data held in other systems.
Data Feed lets us export hit-level data for use outside Adobe Analytics. In this article, we’ll look at what it’s useful for, how to set it up and what to keep in mind when interpreting the files it produces.
Contents
What is Data Feed in Adobe Analytics?
Analytics Data Feed or Data Feed is an Adobe Analytics feature that exports data from a report suite at hit level. Each row represents a call to Adobe’s data collection servers, which might correspond to a page view or an interaction, for example.
This changes how we work with the information. Rather than receiving a report with visits or conversions grouped by a dimension, we get individual records that we can use to build our own analyses.
That level of detail is the main difference between Data Feed and Data Warehouse. Data Warehouse exports aggregated reports, whereas Data Feed provides individual hit records. If we only need a report with specific dimensions and metrics, Data Warehouse may be enough.
What is Data Feed used for?
One of its most common uses is combining browsing data with information from other systems. We can link it to orders, CRM data or sales completed offline, provided we have a shared identifier that connects the two sources.
For example, a business that collects enquiries online and closes sales over the phone can use this export to connect earlier interactions with the eventual sale. This allows the analysis to continue beyond the form submission.
It’s also useful for analysing sequences of behaviour, building custom attribution models or investigating how particular variables were collected and processed. Working with individual records gives us the flexibility to answer questions that may be difficult to address in Workspace.
That flexibility comes with some work to prepare and maintain the data. Alongside storage, we need a process to load it and an environment where we can query it. It’s worth knowing what we want to analyse before enabling an export that, depending on traffic and the selected columns, can generate a substantial volume of data.
How to set up a Data Feed
To create a Data Feed, we need an account with Analytics administrator permissions or access to the feature through a product profile. We also need a supported storage destination, such as Amazon S3, Google Cloud Storage or Azure.
The settings are under Admin > Data feeds > Create data feed. We give the feed a name and select the source report suite. The export must use an underlying report suite. We can’t select a virtual report suite directly.
Next, we choose the columns to include. We can add them individually or use a template. Selecting the columns we need makes the data easier to work with later and avoids exporting fields we won’t use.
In the scheduling settings, we define the start date and frequency, usually hourly or daily. We can also set an end date or keep the feed running continuously. Deliveries require processing, so they shouldn’t be treated as real-time data.
We then configure the destination account and location, along with the compression and packaging options. Including a manifest allows us to check which files belong to each delivery.
Once the settings are saved, we can review individual runs in the Jobs tab. Before automating the loading process, it’s worth checking that the first files arrive correctly and contain the expected columns.
What the exported files contain
Each delivery includes compressed files containing the hit data. With a single-file export, we’ll find a file called hit_data.tsv, with values separated by tabs. If we choose the multiple-file option, the data is split across several files.
Column headers are delivered separately in column_headers.tsv. We may also receive lookup files that translate codes into readable values, such as a browser name. The manifest, if included, contains information we can use to verify the delivery.
The available columns include page data, dates, identifiers, props, eVars and events. Their contents depend on the implementation and the fields selected. A single row can contain several events, so a hit shouldn’t be confused with a single custom event.
An important distinction is the difference between the original columns and those with a post_ prefix. For example, evar1 contains the value sent for that variable on the hit, while post_evar1 reflects the value after processing, including persistence.
If we capture a campaign in an eVar when someone enters the site, subsequent hits may not send it again. Its value can still appear in post_evar1 for as long as it persists under the variable’s configuration. The right column depends on whether we want to examine what was sent or the resulting value after processing.
Why the data can differ from Analysis Workspace
Exporting the data doesn’t mean we can reproduce any report simply by counting rows. Workspace applies calculation rules and exclusions that we need to account for in our queries, alongside any segments and filters in the report we’re comparing against.
For example, Data Feed can include hits that are excluded from standard reporting. The exclude_hit column identifies them: when comparing with Workspace, we need to keep records where exclude_hit = 0. Depending on the implementation, other criteria may also need to be checked, such as background hits from mobile apps.
The choice of columns matters too. Using an eVar’s original value to analyse conversions can produce a different breakdown from using its persisted value. Comparing the two without accounting for that distinction can make a difference in logic look like an error.
Finally, each metric requires its own calculation. The number of rows isn’t the same as the number of visits or page views. Counting visits requires us to combine the relevant identifiers, while calculating purchases also means accounting for duplicate purchases that should be excluded. Before comparing results, we should check that we’re using the same date range, the same filters and an equivalent definition of the metric.
In conclusion, Data Feed gives us access to detailed Adobe Analytics data and a way to connect it with other sources. Its usefulness depends as much on what happens after the export as on the export itself: choosing the right columns, understanding their contents and applying consistent calculation rules are what turn those files into useful analyses.

