SmartSigns is Gainsight’s in-product ‘data-science’. You can point to one or more underlying metrics in the Gainsight Usage object to auto derive a score with the help of the SmartSign feature.

SmartSigns are ideal if you want to automate the creation of scores based on product usage but you are not sure what the right thresholds are, such as ‘how many page-views would be considered ‘Green’?’. It is a great way to incorporate transparent, proven, and mathematical methods into the health score.

SmartSigns autorank customers on a forced distribution curve for the chosen underlying metrics. These scores are then updated on a scheduled basis like rules, but the key difference is that the appropriate scores are automatically determined. Following are a few general guidelines to use SmartSigns:

  • When to use: You have identified the important usage measures for your scorecard, but you are not sure how to determine what thresholds should be ‘Red’, ‘Green’ or ‘Yellow’. You have more than one usage metric and want to generate a single score for those metrics. SmartSigns works only on "ACCOUNTLEVEL" data.
  • Example use case: You have multiple product features or product lines and want to generate scores for each. You can create a SmartSign for each product feature using the corresponding product measures. Emails Sent, Total Leads and Session Count comprise the Engagement Score, and the SmartScore determines how these measures translate to the Health Score.
  • How it should be updated: Scheduled for update in accordance with your data frequency (that is, if you have weekly data, you need to schedule updates every week).

Create a New SmartSign

Perform the following steps to create a new SmartSign:

  1. Navigate to Administration > Account Scorecards.
Creating a New SmartSign:
  1. Enter Measure Name.
  2. From the Type drop-down list, select SmartSign.
  3. Enter Validity Period in Weeks.
  4. Click to continue.
  1. Click Edit as shown in the following image.

The Configure Usage SmartSign window is displayed as shown in the following image.

  1. Click +ADD NEW to pick the metrics you want your SmartSign to be based on. You can pick one or more metrics.

You can also specify if you want to normalize the score using a metric, like # of users. This ensures that you compare apples to apples when you score your customers. If you have multiple metrics and select the Apply Normalizer to all check box, the same normalizer will be used for all metrics.

More is good: It is assumed that this is a positive action and this check box is selected by default.

After you provide all the required information on the Configure Usage SmartSign window, you can:

  • SAVE: Will be saved but will not run
  • SAVE & RUN: Will run immediately just once
  • SAVE & SCHEDULE: Will run based on the schedule you set

After one of these options is chosen, you will be automatically navigated to Scorecard> Measure and Group Configuration.

The new SmartSign measure is displayed as another card. You can hover over the SmartSign card and click on the following icons (refer to the following image).

  1. View Results
  2. Run Again
  3. Edit
  4. View Execution History
  5. Delete

Email Notification

When the SmartSign is finished running, you will receive an email notification, confirming that the RUN was successful, and the results are ready to preview.

Load to Scorecard Measure

When you click Load, a screen appears where you can click Ok.

Load to Scorecard Measure

Populated SmartSigns are available on the Gainsight Home Scorecard Dashboard or in the Customer 360 Scorecard section.

MDA Reporting

If needed, you can also view the results from the SmartSigns in the Report Builder. Gainsight creates an MDA Subject Area named SmartSign Name Scores. In this case, since SmartSign is called Product Adoption SmartSign, the corresponding MDA table is Product Adoption SmartSign Scores”. You can easily build a report like the one below if needed.

How are SmartSign Values Computed?

  • For each metric, Gainsight fetches the data for 14 weeks, if available. Churn and Inactive accounts are filtered out. Only Active accounts will be scored and may have fewer weeks.
  • Compute a time weighted average of each metric, where recent weeks have a higher weight. (The weights are linearly decreasing from most recent, to the oldest, and the norm of the weights = 1). This means more recent usage counts for more when computing the scores.
  • Normalize each time weighted average metric by its selected normalizer. This ensures that the apple to apple comparison is done.
  • Apple to apple comparison example: If Logins are used as the metric to compute SmartSigns. Large customer could possibly have more logins than small customers and hence, the large customers will get higher scores without normalization. If the time weighted average logins are normalized by number of users, then the metric represents average metric per user. This leads to an apple to apple comparison.
  • Rank customers for each metric from best to worst (for some metrics higher is better, and for others lower is better). Unless the More is good check box is deselected when setting up SmartSign, it is assumed that higher values of a metric is better.
  • Use a customer sorting algorithm that considers the individual ranks for each metric to rank all customers. The rationale is that for each two customers, if customer A has a higher rank than B in more metrics, then customer A will be ranked higher.
  • Gainsight translates the rank into a score from 0 to 100 (no bell curve anymore). The score is just the percentile rank. If a customer gets a score of 20, that means the same customer is better than 20% of the overall customer set for those metrics.

Note: When you create a SmartSign, if the selected metrics or columns have 0s/blanks/nulls in more than 50% of the rows, an error will be generated: Your SmartSign <name> could not be generated because All Score Metric dropped after suitability check. This is a functionality of the SmartSign feature that has been included because scoring WILL NOT work properly in a case where all the metrics/columns chosen are sparse. Edits can still be made as necessary to change the weight, or to change the measure that is being referred to.