Ideal Frequency Report


Frequency management on its own is important for controlling costs, but it's only when we layer in conversion that we can say what those costs yielded for a client.

Our frequency conversion analysis starts out like our typical frequency reports — the frequency is typically on the x-axis — and adds key metrics on top.

We offer this report at the portfolio-level, by campaign, by site, and by campaign and site together.

Available KPIs (CTR, CVR, ROI, Profit, eCPA) are enabled and users can input custom value for CPM and Revenue per Conversion based on which the KPIs are calculated.

Here's what the report looks like:


Above, you can see the frequency on the x-axis. The number of impressions that fell into a frequency bucket is represented by the blue bars, with corresponding values on the left y-axis. Meanwhile, the conversion per impression (one of metrics enabled) at a given frequency is found by the orange curve, with corresponding values on the right y-axis. In looking at the campaign above, a marketer can see that the conversion rate spikes at a frequency of 6.

We can also input parameters for CPM and Revenue Per Conversion. These values affect some of the other metrics we report on by frequency, such as profit, ROI, and eCPA:


Again, the x-axis shows the frequency, and the y-axis shows the key metric of interest at that frequency. The dots are data points, and the lines are trend lines.

Note that the frequency bucket with the highest conversion rate may not be the same bucket that is identified to maximize profits. There will always be trade-offs when deciding efficiency vs. volume. Capping frequency at say, 1 ad per user, may achieve the lowest cost per conversion, but you may end up getting a very low number of conversion. Optimal frequency levels are often much higher than frequency levels that maximize conversion rates.

The question needs to become: "At what frequency do I still meet my profitability targets on my conversion?"

Key Metrics







Converters x Revenue per Conversion

Media Cost

CPM x (Impressions/1,000)



Conversions per Impression


CVR per User



(Converters x Revenue per Conversion) – Media Cost


Profit /  Media Cost


Media Cost / Converters

On Conversion Signals:
Events vs. Segments

There are two types of frequency conversion reports: frequency conversion based on events and frequency conversion based on a segment.

For frequency conversion based on events, we only count the earliest conversion event, throwing out all subsequent conversion events. Impressions that "count" toward that single conversion event are only impressions that precede that single earliest conversion event. All other conversion events AND impressions are not counted after the initial conversion event.

For frequency conversion based on a segment, our ability to discern the conversion time is less refined. Because segments tend to have users from the last 30 or 90 days, it can be difficult to do an analysis far in the past. We do not have access to a timestamp – rather, a kuid's membership in a segment & that segment's definition is all we have at our disposal in terms of conversion-detection.

To illustrate with an example, suppose a Northern Trails Outfitter customer sees 3 boot ads in the morning (3 impressions) and buys the boots at lunchtime (converts), and, while still at work, sees one ad for socks (1 impression) and then buys the socks that night after he gets back from mountaineering class. How will his impressions and conversions be counted?

  • Events: He would be counted as one user (a converted user) in the 3-impressions bucket, because we throw out both the impression & conversion event that happen after the initial conversion event.
  • Segment: He would be counted as one user (a converted user) in the 4-impressions bucket, presuming that the second conversion did not fall outside of the 30 or 90 day segment window.

Best Practice Recommendations

  • Since segments do not record the timestamp, usage of events as the frequency conversion type are strongly recommended.
  • It is preferable to define at least 30 days or whole month(s) as the date range where possible, as this will allow for fluctuations on the delivery within the time period.


How do I get a report like this?

  • This feature will be autoprovisioned
  • You can find this on-demand report in the Manage Report center
  • This feature is included in Salesforce Audience Studio MCA and Marketer Pro Editions, and is also available as an add-on to Marketer Basic  
Have more questions? Submit a request


Please sign in to leave a comment.