KPIs for Media and Publishing

Open rate

  • Definition: In the context of personalization, we use the open rate to measure the newsletter's effectiveness, where the subject line is personalized. The open rate is calculated by dividing the total number of emails sent with a personalized subject line in the campaign by the number of emails opened.
  • Note: The open rate can is used in email newsletters. It indicates the added value of personalization in case the subject line is personalized compared to the non-personalized version of the subject line.
  • Measurement unit: Percentage, % (#email opened/#email sent with personalized subject line)

CTR (Click-through rate)

  • Definition: The number of clicks on the recommended products divided by the number of recommendations displayed.
  • Note: CTR is a primary metric that indicates the relevancy of recommendations and suggests the user experience level.
  • Measurement unit: Percentage, % (#clicks/#recommendations displayed)

Number of clicks

  • Definition: The number of times the users click on the items recommended by the recommendation engine.
  • Note: Clicks play the role of primary feedback from users that they find a recommended product is interesting. This feedback is so-called implicit user feedback.
  • Measurement unit: Natural number (count)

Search position clicked

  • Definition: The position of the item in the search result list on which the user clicks
  • Note: It’s used as a raw metric in qualifying search personalization accuracy. A lower number means more accurate prediction. 
  • Measurement unit: Natural number

Successful search rate

  • Definition: The ratio of successful personalized searches. It’s calculated as the number of search results clicked divided by the total number of personalized search results
  • Note: This metric provides a hint on the search personalization accuracy. The higher rate means better search personalization. 
  • Measurement unit: Percentage, % (#personalized search result lists with a click/#personalized search result lists)

GMV through recommendations

  • GMV stands for Gross Merchandising Value or Gross Merchandising Volume
  • Definition: The value of conversions 'through-recommendations'. A conversion is considered to be generated “through-recommendations” when the recommended product is purchased within a specific period and as a result of clicking on a recommendation.
  • Note: GMV through recommendations is a useful KPI to compare two or more personalization engines' performance in terms of revenue attributable to the solution. It's typically used for comparing product recommendations.
  • Measurement unit: Currency

GMV through recommendations per 1000 recommendations

  • Definition: Value of conversions 'through-recommendations' divided by thousand recommendations displayed, used to calculate return on service costs
  • Note: Useful KPI to compare the performance of two or more recommendation engines in terms of average revenue attributable to the solutions; typically used for comparing product recommendations
  • Measurement unit: Currency (total income from recommendations / 1000 recommendations)

Number of recommendations

  • Definition: The number of recommendation requests provided by the recommendation engine. More simply put: the number of times the recommendation engine generates a recommendation for the users.
  • Note: The number of recommendations is not a performance indicator, but it's used to measure the personalization solution's traffic.
  • Measurement unit: Natural number

Total revenue

  • Definition: Value of the total revenue of Yusp's customer
  • Note: Considering the total revenue of our client is useful for determining the personalization engine's added value. For example, we can compare the total revenue of a non-personalized website and a personalized website.
  • Measurement unit: Currency

Conversion into views (Long-term CTR)

  • Definition: The number of recommendations resulting in content views - where at least 10% of the content's length is viewed - divided by the number of recommendation requests.
  • Note: If the goal is to drive content consumption, this metric is important implicit feedback to the algorithms about the quality of recommendations. The 10% limit of the content length secures that we exclude clicks on clickbait type content and unintentional clicks.
  • Measurement unit: Percentage, % (#click with minimum 10% content view/#recommendation requests)

Time spent on site

  • Definition: The total time users spend on the website.
  • Note: Time spent on site is an important KPI measuring the site's overall performance and the level of user engagement.
  • Measurement unit: Time - seconds

Conversion into sales (revenue from pay-per-view)

  • Definition: The number of recommendations resulting in content purchases divided by the number of total recommendation requests (i.e., the user purchased an item because she clicked on it in a recommendation widget within a specific period).
  • Note: Like in eCommerce, this KPI is used to qualify the personalization engine's impact on a transactional type streaming business model, typical of telecommunications companies.
  • Measurement unit: Percentage, %

Number of article pages viewed from recommendations

  • Definition: This absolute measurement counts the number of article pages opened by clicking on a recommendation.
  • Note: It's used in the News and Publishing industry to qualify the impact of recommendations on user engagement.
  • Measurement unit: Natural number (count)

Normalized number of article pages viewed from recommendations

  • Definition: Similarly to the 'number of article pages viewed from recommendation' KPI, this absolute measurement counts the number of article pages opened by clicking on a recommendation, but normalized by the number of users receiving the recommendations and calculated within a specific period.
  • Note: This metric is used to compare two or more personalization engines and their relative impact on user engagement in the News and Publishing sector. The time frame is usually the duration of an A/B-test.
  • Measurement unit: Real number (#article views from recommendations/#users)

Conversion rate

  • Definition: The number of purchases within a specific time after a recommended product is clicked divided by the number of recommendations displayed.
  • Note: The conversion rate indicates the efficiency of the recommendation engine.
  • Measurement unit: Percentage, % (#purchases after click/#recommendations displayed)