Metrics and Strategy: Selecting the Best Metrics for the Job

0

Presentation Transcript

  • 1.Metrics and Strategy: Selecting the Best Metrics for the Job STMIK Pontianak Prepared by: Amar P. Natasuwarna
  • 2.2 References
  • 3.3 Project Requirements Who is my audience? Am I doing better than my competitors? What are people saying about my brand? When is the best time to publish? Which content does my audience enjoy the most? Which social network is the best for my brand? How can I have a better performance on social media?
  • 4.Focus on Being Actionable The goal of metrics is that they can be actionable, or as actionable as possible. If a metric does not drive us to take action, either it is because we are high above our goals (and might need new goals), because we are not using the best metric, or because we are not using enough metrics. 4
  • 5.Dealing with Straightforward Questions Some questions or project requirements are straightforward and lead directly to the metrics needed to answer them. When is the best time to publish? That is one example. Immediately, we think of a metric that deals with performance per time of the day and per day of the week 5
  • 6.6 Posts/interactions comparison
  • 7.7 Content Ranking Tables
  • 8.8 Interactions by Content Type
  • 9.9 Types of Interaction
  • 10.10 Interactions Distribution by Post Type
  • 11.11 Model A model in this case can be any combination of metrics or dashboards that we use. A model is likely to be used again and again in different projects. This model can start very simple in some cases, just as one custom metric, and then become as complex as we need it, becoming a collection of metrics grouped into one dashboard or report, displaying all the different parts of a bigger story
  • 12.12 Dealing with Complex or Subjective Questions The following are some examples: Am I doing better than my competitors? What does my audience enjoy the most about my content? What is it that gets them engaged? What do I need to do to grow my audience and boost my engagement? A good point to start the approach into complex or subjective tasks is to clarify the meaning behind each question, objective or goal we are given. Step One: A Simple Example Step Two: Building a Story
  • 13.13 Step One: A Simple Example Let’s look at other questions that we can make when the brief includes “Am I doing better than my competitors?” Do you have a list of your competitors? “Better” in which aspect? Overall interactions? Conversions? Audience growth? Over which period of time? By product line? By theme? By questions answered? The use of a two-way channel? By sentiment, despite head-to-head interaction comparison?
  • 14.14 Step Two: Building a Story Find a good order of metrics to make up a storyline. Look at the details of building dashboards and reports. A story can also be applied to the way we present the data, so that people understand the events they see. A simple example of this is thinking of posts generating interactions. Look first into posts, understanding what the posts are talking about, frequency, and timing Then jump into interactions to start correlating types of interactions with types of posts, and the timing of the campaign. Finally, go back and conclude the story with insights from the results.
  • 15.15 Some insights come to mind after exploring linearity, following a line of thought such as “if this, then that,” which means that once we find out about one event, we look for a correlation of that event with other events in a timeline. The sense of linearity can make the analysis process much easier to perform. Building a linear line of analysis, however, requires a setup process; hence, time is a primary resource in our work. Our objective is, in essence, to look for the fastest way to reach an insight. An insight, in this case, can be the final answer to one of our complex questions. In the very end, however, linearity plays a definitive role in giving us the final answers we are looking for. Insight and Linearity
  • 16.16 Ideal Metrics Ideal metrics are found with experimentation, trial and error, and looking at different charts until we find the best ones. At some point, we will have enough experience to choose metrics very quickly, even without having to open a tool to do so. Estimated metrics: avoiding bad decisions It is important to keep causality in mind when working with estimates. Causality is vital to analytics in general.
  • 17.17 What Are Estimated Metrics Exactly? In short, these metrics take into account potential facts. There are two main categories of estimated metrics: Given by the social networks. Calculated by third-party technology.
  • 18.18 Given By the Social Networks With such metrics, social networks are indicating to us how far our content has gone within the network. The aim of such metrics is to help us have an extra reference toward our true interactions. It helps us have an idea of the dimension of the network, and of the distribution of our content within the network. Sometimes these metrics can help understand the effectiveness of good copy. Audience insights are also often estimated, even by the social networks themselves.
  • 19.19 Calculated by Third-Party Technology Potential reach and potential Impressions are included here. Different from true reach and true impressions, these metrics are calculated by third party technology. This is where we can be as careful as possible before including such metrics in our dashboards and reports. The first step is to have a clear view of what exactly is the formula behind such metrics. A common approach for potential reach, for example, is to calculate the amount of followers. It indicates our content counted as being potentially reached by our content. The aim of such metrics is only to give us an extra reference; by no means do these metrics aim to be our top key performance indicators.
  • 20.20 Making Good Use of Estimated Metrics Once we understand what is behind an estimated metric, being it a formula, hidden data sources or a machine learning process, we can then safely place such metrics into our strategy and make very good use of them. The following two points are common applications of estimated metrics: Patterns and correlation between estimate and true performance Supporting metrics: from substitutes for missing data, to predictive and prescriptive analytics A very common use, for example, is to look for any correlation between reach and impressions to interactions and conversions. When we have higher reach, do we also have higher interactions?
  • 21.21 Impressions vs. Interactions
  • 22.22 Figure 11-23 shows how interesting it is to relate an estimated metric to actual interactions. There is usually a correlation between the spikes of both metrics. All expectations, sometimes we even have more interactions at a point when we have a comparatively lower amount of impressions. So more impressions does not mean we have more interactions in this example. Impressions vs. Interactions
  • 23.23 The essence of metrics is to have a revealing view of the facts. Metrics then become part of tactics, where we apply them toward strategic goals and learn how to repeat such tactics when successful. The tactical use of metrics gives us experience, and initially much of it is done with experimentation. While in time we will not need as much experimentation to reach our goals, it is very likely that we enjoy being open-minded and continue to experiment with metrics independent of being master analysts. Metrics and Tactics
  • 24.Thank you Special thanks to all the people who made and released these awesome resources for free: Presentation template by SlidesCarnival Photographs by Unsplash Illustrations by Undraw.co