SAP

Author

Jérôme Ivain

Jérôme Ivain

Jérôme is a Product Manager for SAP Analytics Cloud's Smart Assist feature portfolio. He holds an Engineer Diploma in Data warehousing and a certification in Data Sciences from Polytechnique School in France. Jérôme has 10 years of experience in the software industry in a variety of roles, from Software Developer to Product Management. When not at work, Jérôme enjoys playing drums and sports like rock climbing and triathlon.

Keep in touch

Subscribe for the latest news, updates, tips and more delivered right to your inbox.

Subscribe to updates

Category

Learning

Connect with us

Smart Discovery is one of SAP Analytics Cloud’s Augmented Analytics smart features. These smart features augment the analytics process to get you the information you need in less time and with far less manual work. They use machine learning technology to automatically analyze your data, surfacing hidden patterns and complex relationships that can lead to actionable insights.

As a business analyst and story creator, Smart Discovery helps you to understand business drivers behind core KPIs such as revenue, churn, and productivity.

Whether you want a quick way to start your analysis or want to dive deeper into an existing story, Smart Discovery will auto-generate pages of analysis to help you discover insights and answer your business questions.

With Smart Discovery you can:

  • Identify actionable insights
  • Expose the key influencers driving business-critical KPIs
  • Analyze outliers to identify impactful decisions
  • Predict future outcomes with interactive “what-if” simulations

What’s New with Smart Discovery?

In SAP Analytics Cloud version 2018.16 we began releasing an update to the existing Smart Discovery functionality. If you don’t have the new version yet, don’t worry. All subscribers will have access to the new functionality soon.

Download the updated Smart Discovery documentation

In this new version you will benefit from:

  • A more intuitive workflow: use Smart Discovery to build a new story or add Smart Discovery to an existing story
  • More charts and descriptive content are included on pre-built story pages
  • Explanations for key drivers, unexpected values, and simulated impact on your business
  • Integration with SAP Digital Boardroom
  • User experience enhancements
  • The ability to format and edit the auto-generated charts and calculations

How to use Smart Discovery to Build a Story Automatically

Now, you can use Smart Discovery to build a comprehensive story for you from scratch.

To start your story with a Smart Discovery, follow these steps:

  1. From the home page, select Create a new story
    Create Story
  2. Choose Run a Smart Discovery
  3. Select your datasource 
  4. Choose the measure or dimension you want to learn more about
  5. Set advanced options (described in next section) and click Run

How to use Smart Discovery in an Existing Story

Smart Discovery will also run within an existing SAP Analytics Cloud story.

To run a Smart Discovery in an existing story, follow these steps:

  1. From your open story, click on the three dots under the More section in the toolbar and select Smart Discovery from the menu
  2. Choose the measure or dimension you want to learn more about
  3. Set advanced options (described in next section) and click Run

Smart Discovery Advanced Options

Once you have selected the desired target measure or dimension to analyze, additional discovery components can be specified under advanced options. The same advanced options are available whether you’re creating a new story with Smart Discovery or run it in an existing story.

Advanced Options

Version
If you’re working with multiple versions of data (ie. actual, budget, forecast) you’ll want to select the version of the data for Smart Discovery to run on.

Record name
To help specify the context of your analysis, you can set a name for the data records you are considering in the discovery. This will enhance the discovery and make it easier to consume by colleagues who are less familiar with the raw data.

In the content on Smart Discovery pages, Analytics Cloud uses the record name indicated in advanced options or defaults to “record(s)” if one has not been specified.

Record name

Included Columns
Considering the information in the underlying data model, you may want to specify which measures and dimensions the Machine Learning algorithm of Smart Discovery should consider.

Filters
By adding filters, you’ll reduce the number of records to be analyzed. Filters allow you to choose specific measure and dimension members to include or exclude from the analysis.

The side panel helps you visualize how the filter selection will impact the overall Discovery capacity. You can also choose to allow viewers to use filters while they’re consuming your Smart Discovery.

Smar Discovery Filters

What does a Smart Discovery Look Like?

Smart Discovery automatically generates a rich story composed of four pages: Overview, Key Influencers, Unexpected values, and Simulation. These pages contain sophisticated information that is ready to be consumed. The augmented analytics process means that your story is prepared in just a few minutes, saving your hours of work.

Overview Page

The overview page shares meaningful insights that are automatically generated by SAP Analytics Cloud’s machine learning algorithms.

On the overview page you’ll find information on:

  • Target measure value and context (data model, applied filter and hierarchy) displayed as a key point with a comparison between actual and forecast versions, as well as the Min, Max and aggregated value (Total)
  • Distribution of the target measure for all records, represented by a histogram
  • Trends over time with a predictive forecast (provided that you have a time dimension in your metadata)
  • Other explanatory charts enriched with Dynamic Text Tokens

Smart Discovery Overview

Key Influencers Page

The concept of key influencers is important. It helps to explain how machine learning features, including Smart Discovery, work to augment the analytics process.

Key influencers are the dimensions that best explain (or influence) the value of the target measure in the context of the model.

For example, imagine a dataset where “Age” is a key influencer of “Revenue”. This would mean that the variation of age highly correlates with the variation of revenue; be it a positive (i.e. the older the population, the higher revenue) or negative (i.e. the older the population, the lower revenue).

Sometimes, certain key influencers can be removed from the analysis if the correlation is obvious from a business perspective. The real value lies in the key influencers information that may not have otherwise been observed without the machine learning power of the Smart Discovery.

On the Key Influencer page, Smart Discovery presents the top 10 key influencers of the target measure as a ranked bar chart. The top 3 influencers are highlighted.

The quality of the discovery is scored out of five. This is because when the Smart Discovery runs, machine learning algorithms create many possible predictive models. Then, the model that has the most explanatory power for the data provided is selected for the Smart Discovery. The model is scored based on how well it works when applied to new data. A quality score of 5/5 indicates high-quality insights that can be trusted.

Smart Discovery Key Influencers

Below, you can see how the top three key influencers contribute to the value of the target measure. In this example, the first chart compares Total Expected Revenue (the selected target) for each of the Customer Segment values. The second chart explains the distribution of the target over the whole dataset per Customer Segment.

Smart Discovery Key Influencers chart

Unexpected values

One of the predictive models created by the Smart Discovery explores any trends in the expected/predicted values for certain attributes, following a calculated regression line. If an actual value diverges from the regression line it is categorized as “unexpected”.

Smart Discovery Unexpected Values

Simulation

The Simulation page is also based on a calculated regression model. It shows a combination of contributors (influencers) and their impact on your selected target.

The simulation acts as a “what-if” analysis, helping you test the impact of various inputs (dimension/measures). Modify inputs independently or all together. Then, observe if the change positively, negatively or neutrally contributes to the target. You’ll dynamically see the delta in percentage between the previous and current value of the total target.

The included waterfall chart helps to visualize the whole configuration. Base Value is a “retro calculation” made for assessing whether the combination of contributing values brings an increase or decrease to the target.

 

What’s Next?

Once you’ve finished analyzing the results of your Smart Discovery you can save it as a new story or part of your existing story.

Next, you’ll want to add comments to highlight the most interesting insights that you’ve discovered and then share the story with other users in your organization. With augmented analytics capabilities like SAP Analytics Cloud’s Smart Discovery, you’ll be making better decisions with high-quality insights in no time.

SAP Analytics Cloud earns a top ranking from BARC

See how SAP Analytics Cloud performed in the world’s largest survey of Business Intelligence software users.