[Legacy] Set weights and logic

After mapping data sources to your job architecture, the next step is to define how those data sources should be aggregated and prioritized when calculating benchmarks for each job. This is done by setting weights and logic.

How weights and logic work

Weights and logic help you control how mapped survey data is used to generate composite benchmarks for a job:

• Weights determine the relative importance of each mapped data source and how that data will contribute to a composite benchmark. For example, you might give 70% weight to one survey provider and 30% to another.
• Logic defines how data sources apply to specific jobs, pay zones, and pay types (including fallbacks in the case where data is missing). For example, you might use one survey provider for your base pay for Engineering jobs, but augment it with another survey provider for your GTM jobs. Or target 60th percentile for your Tier 1 pay zone, but 50th for Tier 2.

Together, weights and logic give you flexibility to tailor your market pricing methodology to your specific compensation philosophy and the strengths of different data sources for each job.

Prerequisites

• Map data sources to jobs

Set data source logic

Define how data sources apply to specific jobs, pay zones, and pay types. You will create groups of jobs, pay zones, and pay types, and then define how data sources should be applied to each group.

1. Go to Market Pricing > Benchmarking > Edit job mapping > Survey mapping

   

2. Under Data Sources, click either Add group to define logic for a new group of jobs or Edit to modify an existing group

   

3. Each group is comprised of:

   • Primary data source(s) - which data sources should be applied first

   • Fallback data source(s) - if data is missing from the primary, which data sources should be used next

   • Job architecture filter - which jobs, pay zones, and pay types should these data sources apply to

     

4. Select your primary data source from the dropdown and add more primary data source(s) as needed

   • To include several percentiles from the same data source, add a separate data source for each percentile

5. Set your data source weights

6. Add fallback data sources as needed

7. Select which jobs you want to apply your data source selection to:

   • Family, function, ladder: You can multi-select from either your family, function, or ladder from your job architecture (example: Design, Product Management, Software Engineer)
   • Track: You can multi select your tracks from your job architecture (example: IC, Manager)
   • Pay zones: You can multi select from your pay zones (example: US-1, EU-1, etc.)
   • Pay types: You can multi select from your pay types (e.g. Base Pay, Total Pay, Variable Pay, Total Equity, New Hire Equity)
   • Note: You must make at least one selection from each of these categories.
   • Note: Each distinct job (family + level + track + pay zone) and pay type can only belong to one group.

8. Once completed, click Save to save that group

:::info Note: benchmarks will only be generated for jobs, pay zones, and pay types that have been defined in a group.

:::

:::info Note: changes to groups will not be applied to benchmarks until benchmarks are regenerated.

:::

Set fallback data sources

You can define up to three fallbacks for a group. This means that if none of the primary data sources return data, the system falls back to the first fallback group and if none of the data sources in the first fallback group return data, the system falls back to the second fallback group and so on.

To configure fallback data sources:

1. Go to Market Pricing > Benchmarking > Edit job mapping > Survey mapping

2. Click Edit on any group you want to add a fallback data source to

3. Click Add fallback to create a new fallback data source

   

When generating benchmarks, the system will look for data in priority order:

1. Primary data source matches with weight > 0
2. First fallback matches with weight > 0
3. Second fallback matches with weight > 0
4. Third fallback matches with weight > 0
5. Any mappings with weight = 0

Only one data source with data is needed from the highest priority group.

Types of weighting

There are two approaches to weighting data sources:

1. (Most common) Manual weighting - Manually assign specific weights to each data source. This gives you full control but requires more effort to maintain.

2. Market weighting - Weight data source benchmarks by number of employees or number of companies represented in the data set. This enhances the accuracy and relevance of compensation benchmarks by considering the relative sample sizes of different data sources, but is heavily dependent on data coverage.

   

How manual weighting works

To give you fine-tuned control over your data source weighting strategy, Pave provides several knobs to control weights:

• Job family weighting - sets weight across data sources for entire job families. Set in job family mappings.
• Job level weighting - sets weight across data sources for entire job levels. Set in job level mappings.
• Pay zone weighting - sets weight across data sources for entire pay zones. Set in pay zone mappings.
• Overrides weighting - sets weight across data sources for individual jobs. Set in overrides mappings.
• Source weighting - sets weight across data sources for entire groups of jobs. Set in data source logic.

Using a default weight of (1) for all settings will result in an evenly weighted composite benchmark. Often, you may want to use more sophisticated weighting methodology.

How zero weights behave

Setting a weight to zero (0) for any mapping means that data source will be ignored entirely when generating a composite benchmark, even if it's the only survey with data for a particular job. However, the data point will still be viewable within the benchmarking experience as a reference.

How market weighting works

With Market Weighting, you can weight data sources by number of employees or number of companies (the sample size) represented. This enhances the accuracy and relevance of compensation benchmarks by considering the relative sample sizes of different data sources.

• Market Weighting by Employee: Weighting will be calculated based on the number of employees that make up a given data slice (the # of incumbents that make up the sample size from the survey)
• Market Weighting by Company: Weighting will be calculated based on the number of companies that make up a given data slice

:::info Note: If using a survey data source, there must be an employee or company count column for market weighting to work.

:::