Before you Begin
- There are no default restrictions on viewing the Recommendation Dashboard — anyone with access to your Uberflip Hub can do this.
- If needed, access to the Recommendation Dashboard can be granted or removed for groups or individual users using the Dashboards > View Recommendation Dashboard permission.
- The Recommendation Dashboard will only display data if Content Recommendations, Site Engager, or both are set up and active on your Hub.
About the Recommendation Dashboard
The Content Recommendations and Site Engager features allow you to offer truly personalized content recommendations to your visitors, helping to surface your most effective Hub content for each person. If you're using these features, the Recommendation Dashboard is where you can see exactly how the content recommendations they generate are performing:
The Recommendation Dashboard gives you access to metrics around the recommendations generated by Content Recommendations and Site Engager. At a high level, this gives you a way to gauge the overall effectiveness of these features at engaging and converting your visitors. Plus, by digging into the more granular data, you can gain valuable insights — some of the questions that you can answer with the Recommendation Dashboard include:
- How well are specific Items performing when recommended to visitors?
- Which Items are the most effective at driving conversions?
- Which specific setups (recommendation rules) work best?
- How does the context in which it is recommended (i.e. on different web properties) affect the way your content performs with visitors?
With this information, you can make data-driven decisions about your content experiences and recommendation setups, helping to improve their performance over time.
Open the Dashboard
To open the Recommendation Dashboard:
- Log in to Uberflip.
- Navigate to the Hub you want to view.
- Click on Dashboard in the sidebar menu on the left to expand that menu section.
- In the menu, click on Recommendation Dashboard:
- The Recommendation Dashboard will appear on the right.
The Recommendation Dashboard consists of three main sections: Date Range & Filter Controls, Overview Metrics, and Item-Level & Rules-Level Metrics.
Configure the Dashboard
Set Date Range
You can set the date range for which metrics will be shown either by specifying a custom range of dates, or by choosing a preset date range. The date range you set will affect all metrics shown in the Dashboard.
To specify a custom range of dates, use the date picker:
Or, choose from any of the following preset date ranges, which are relative to the current date:
- Yesterday: The previous calendar day
- Last 7 Days: The previous 7 calendar days, ending on the day before the current day
- Last Week: The previous week, running Sunday to Saturday
- Last Month: The previous full calendar month
Choosing a preset date range will also update the date picker with the corresponding start and end dates.
Notes on Date Range
- The maximum period that can be selected is 31 days
- Data is updated nightly: the most recent available data is always the previous day's data
- Records are available back to January 22, 2019
Using the filter menu bar, you can filter the recommendation metrics in four primary ways:
In addition, there is also a conditional Keyword filter that appears when you make certain selections on the primary filters:
Filter Type: Location
Filters by the location in which recommendations are displayed, based on the feature used to display them (e.g. Content Recommendations or Site Engager).
- All Locations (Default): Recommendations served on the current Hub and all web properties on which Site Engager for this Hub is deployed
- This Hub Only: Only recommendations served on the Hub you're currently in (when selected, activates the Keyword filter to allow filtering by Stream name)
- Site Engager: Only recommendations served across all web properties where you are using Site Engager
Filter Type: Domain
Filters by the domain on which recommendations were displayed. This is specific to Site Engager recommendations, and can't be used when the Location filter is set to This Hub Only.
- All Domains (Default): All domains on which Site Engager served a recommendation, combined
- Any individual domain on which Site Engager served recommendations (when selected, activates the Keyword filter to allow filtering by URL string)
Important: Unexpected Domains
The list of domains shown under the Domain filter may include some unexpected or unfamiliar entries. These "unexpected domains" can appear if you did not restrict the URL(s) where Site Engager can appear while setting up a Site Engager rule, and Site Engager fired in the context of an unexpected domain.
Situations in which this could happen include:
- Any time your site appears on another domain within an iframe, such as if it has been embedded. Some web performance testing tools rely on this.
- When someone accesses your site through a proxy, in which case you will see the proxy's domain.
To prevent Site Engager from firing on unexpected domains, choose either the Page URL Includes or Page URL is Exactly options (under the Placement tab) when setting up a Site Engager Rule.
Filter Type: Visual
Filters by the type of interface used to serve the recommendations, across both Content Recommendations and Site Engager.
- All Visuals (Default): Recommendations across all of the visual interfaces that both Content Recommendations and Site Engager can use
- Hub - Panel: Only recommendations served in the Content Recommendations Panel (sidebar) interface
- Hub - Carousel: Only recommendations served in the Content Recommendations Carousel ("More Content") interface
- Hub - Next/Previous: Only recommendations served in the Content Recommendations Next/Previous Item Flyout interface
- Site Engager - Promo Tile: Only recommendations served in the Site Engager Promo Tile interface
- Site Engager - Exit Intent: Only recommendations served in the Site Engager Exit Intent interface
- Site Engager - Next: Only recommendations served in the Site Engager Next interface (after clicking on a Promo Tile or Exit Intent recommendation)
- Site Engager - Previous: Only recommendations served in the Site Engager Previous interface (after clicking on a Promo Tile or Exit Intent recommendation)
Filter Type: Type
Filter by the method used to generate the recommendations that were served, across both Content Recommendations and Site Engager.
- All Types (Default): Recommendations generated by both Uberflip AI and from single Streams
- AI Engine: Only recommendations generated by Uberflip AI
- Single Stream: Only recommendations pulled from single Streams
Filter Type: Keyword
Filter by searching for a full or partial keyword match. Available only when the following filters have been selected:
- Domain > Specific Domain: Only domains matching a full or partial URL keyword
- Location > This Hub Only: Only Streams matching a full or partial Stream Name keyword
When available, the Keyword filter appears on the right side of the main filter menu bar:
Overview Metrics Explained
Three key metrics are highlighted in the center of the Recommendation Dashboard, along with the relationship between them. These overview metrics are always visible, and respect the currently selected date range and filters.
The metrics shown are:
- How many times visitors saw recommendations for Items from the selected Hub.
- See the note below to learn how impressions are defined and counted.
- How many times visitors clicked on Item recommendations that they saw.
- Form CTA Conversions
- How many times visitors submitted a Form CTA after clicking on an Item recommendation.
The overview metrics are useful for seeing how your recommendations are performing at a high level. In general:
- Impressions tells you how many of your recommendations are actually being seen by visitors
- Clicks indicates how relevant your visitors found those recommendations based on how much they actually engaged with them
- Form CTA conversions gives you an idea of how effective the recommendations are at converting visitors into prospects
As you can see, this is a funnel: impressions lead to clicks, which in turn lead to conversions. This is represented by the arrows with percentages above them that appear between each of the tiles:
These figures represent the conversion rate between the metrics, i.e. the proportion of the metric on the left that converted into the one on the right. In the screenshot above, for instance, you can see that out of 335 visitors who clicked through on a recommendation, 17 submitted a Form CTA — a conversion rate of 5.07%. Conversion rates visualize the relationship between the metrics, and are useful for seeing at a glance how "efficient" your recommendations are at turning your visitors into prospects.
What counts as an impression?
An impression is counted each time a recommendation is seen by a visitor. This means that the recommendation must be actually visible to the visitor to be counted: only Items that are fully rendered onscreen in the visitor's browser count as an impression.
For example, in the screenshot below, the top Item would count as an impression, but the bottom Item would not, because it is cut off (not fully visible):
We also have built-in error correction that will record an impression if an Item registers a click-through without a corresponding impression. For example, if a visitor clicked on the bottom Item in the screenshot above, this would increase the click count but not the impression count, so an impression is automatically recorded in this situation to prevent a discrepancy.
Item-Level and Rules-Level Metrics Explained
You can also view these same metrics (impressions, clicks, and Form CTA conversions) broken down by individual Items and recommendation rules. These more granular metrics are shown in the tables below the Overview Metrics section under the Items and Rules tabs:
This table breaks down impressions, clicks and Form CTA conversions (along with conversion rates between them) by individual Items that were recommended to visitors. All metrics shown reflect the currently selected date range and filters.
By default, the table is sorted in descending order by impressions (i.e. the Item with the most impressions at the top). You can sort the table by any of the available metrics by clicking on its column header, including the conversion rate columns between each of the metrics. The column currently used for sorting is highlighted and has an arrow beside it to indicate the sort order (click a column header once for descending, and click it a second time for ascending):
You can get some useful insights by sorting the table in various ways:
- Items with a lot of impressions are recommended frequently, but this doesn't necessarily mean that they're effective — look at their conversion rates to the other metrics to identify how you might improve them. Items that are recommended a lot tend to be the ones that are the best fit for your visitors' interests, so if you are seeing an Item of this type with relatively few clicks, this can be a signal that you need to improve its title and/or thumbnail.
- Items with a lot of clicks, especially relative to impressions, are grabbing the attention of your visitors. These Items are worth analyzing to see what makes them so appealing, as you may be able to apply what you learn to your other content as well.
- Items with a lot of Form CTA conversions relative to clicks are your high-performers. To get the most out of them, you want to get these Items in front of as many visitors as possible, so look at impressions and clicks to see where you can improve.
This table breaks down impressions, clicks and Form CTA conversions (along with conversion rates between them) by the specific recommendation rules that you have set up. All metrics shown reflect the currently selected date range and filters.
By default, the table is sorted in descending order by impressions (i.e. the rule that generated the most impressions at the top). As with the Item-level metrics table, you can click on a column header to sort the table by that metric, and the sorting column is highlighted with an arrow indicating sort order. This includes the conversion rate columns between the metrics, but does not include the Location column, which can't be sorted (use filters instead).
The Rules-level metrics are helpful for understanding the effectiveness of your different recommendation rules:
- If a rule is generating a lot of impressions, don't assume that it must be working: this might simply be because it is widely deployed, and doesn't mean it's actually effective. Look for rules that have a lot of impressions but a relatively low number of clicks (i.e. a low conversion rate from impressions to clicks). These rules are showing Items a lot, but the low clickthrough rate can indicate that these Items may not be appealing or relevant to most visitors who see them, so the rule may need to be tweaked.
- If some of your rules are narrowly targeted on purpose, you would not expect them to generate many impressions. However, because of their narrow focus, you could generally expect them to have a higher click-through rate than the average, widely-scoped rule. For this type of rule, look for instances with a low clickthrough rate to identify opportunities for improvement. You can also look at similar rules which are more effective to see what you can learn from them.
- Make sure to also look for rules that are generating few or no impressions. If they aren't being seen, this could be due to a setup problem.
Export Recommendation Metrics
If you want to slice-and-dice these metrics further, you can export the raw data and import it into an external data analysis tool or data warehouse. The exported data is available in CSV format, which most tools of this kind support as an import option.
As the Recommendation Dashboard doesn't support date ranges longer than 1 month, you can use the CSV export instead: simply export the previous month's data each month, which will allow you to view data over longer time periods in your external data analysis tool.
To export data, click on the Export CSV button on the right side of the filter menu bar:
This will generate and automatically start downloading a CSV file with a file name in the format Recommendations - Account XXXXXX - Hub XXX - YYYY-MM-DD to YYYY-MM-DD.csv:
The export feature will apply the selected date range and all selected filters at the time the Export CSV button is clicked to the exported file.