CPI Systems Survey
Created by Collin Brown and Steve Martin on behalf of the Task Team on Scanner Data
Presentation created with RevealJS
Presentation Overview
- What this survey is about
- Why we are running the survey
- Explanation of key concepts
- Survey walkthrough
- What we hope to learn
What this survey is about
The interaction between
-
domain-aligned teams
-
centralized teams
-
software systems
-
input data
-
GSBPM processing steps, and
-
output data
and how it affects CPI production systems.
What this survey is about
- Teams own systems.
- Teams and systems interact with eachother.
- Systems transform data across the GSBPM process.
What this survey is about
For CPI Production systems around the world, we aim to
- characterize team interactions and system organizations,
- identify the tools and practices currently in use, and
- infer how these factors relate to key business outcomes.
Why we are running the survey
- Alternative Data Source (ADS) usage in CPI systems demands significant technical expertise across multiple skill domains.
- Developing and maintaining CPI systems that use ADS can cut across organizational silos.
- Lifetime of CPI systems and practices can span decades.
Why we are running the survey
As far as we know, we are the first to:
- characterize the state of team and system organization,
- identify the tools and practices being used, and
- understand how these characteristics relate to business outcomes (including ADS adoption).
Explanation of Key Concepts
In this section, we cover the key terms and concepts used throughout the survey.
System
A system is any indivisible1 software component that takes data input(s) and produces data output(s).
1: I.e., the component runs entirely or not at all.
| Process | Explanation | |
|---|---|---|
| Ingestion | Getting acquired data into a machine-readable state where it is ready for further processing. | |
| Processing | Preparing the data to produce elementary indexes. | |
| Elementary Indexes | Calculate elementary indexes using the processed data. | |
| Aggregation | Aggregate elementary price indexes into higher-level indexes. | |
| Finalization | Store price indexes for subsequent use and dissemination. |
Flow of Work
The direction of processing steps from raw data to final outputs.
Team
A group of individuals who maintains one or more systems.
Note: One system can cut across steps.
Note: More than one system can exist within a single step.
Domain-Aligned Team
A group of individuals who are functionally embedded within the CPI team. These teams tend to have significant CPI domain knowledge.
This is in contrast to a...
Centralized Team
A group of individuals who are centralized elsewhere in the organization. These teams tend to NOT have significant domain knowledge.
Putting it all together
Ingestion
- A (domain-aligned) group of economists in the CPI team maintain a single system for one data source.
- A (central) corporate IT team maintains a single system that manages the ingestion of the remaining data that calculates the CPI.
Processing and Elementary Indexes
- The top (domain-aligned) team maintains one system that handles both processing and elementary index calculation1.
- Each of the remaining (domain-aligned) teams maintain two systems that separately handle processing and elementary index calculation.
Note: For the top system, it is not straightforward to make significant changes to the processing component without also affecting the elementary index component.
Aggregate and Finalize
- A second (central) corporate IT system implements both aggregation and finalization for every upstream elementary index.
Concluding Thoughts
- Goal is to have a lingua franca for all CPI employees to express team organization and system architecture concepts.
- Our goal is not to create an exact replica of your CPI system using these diagrams; rather, we are trying to roughly characterize the germane aspects of team and system organization.
Survey Walkthrough
In this section, we go through each page of the survey.
Page 1: System Boundaries
The point of this page is to understand if and when systems cut across steps.
We are also interested in how many systems are used for each GSBPM step.
Page 1: The two extremes
Page 1: The two extremes
Page 2: Team Boundaries
In this page, we're trying to learn which kinds of teams are involved in each step.
| Team Type | Description | |
|---|---|---|
| Corporate IT | Information Technology (IT) professionals who are part of an organization-wide central group. | |
| Domain-embedded IT | IT professionals who are functionally embedded in the CPI team (i.e., part of the CPI team, not the corporate IT department). | |
| Domain-embedded Analysts | Professionals without a formal IT background (e.g., economists, statisticians) who are functionally embedded in the CPI team. | |
| Elsewhere Analysts | Professionals without a formal IT background who are not functionally embedded in the price statistics team. | |
| External Consultants/Contractors | Professionals outside of the organization to whom system development/maintenance work is contracted. |
Page 3: Joint Team-System Organization
- For a representative group of systems used in your CPI production, we want to understand the joint team-system organization.
- It is often not feasible to ask respondents to characterize the joint team-system topology for every system invovled with CPI production.
- No strict definition - the idea is to think of the group of systems that participate in the typical flow of work.
Example: Representative Group
Representative Group Takeaways
- The important part of this question is being able to accurately describe which teams maintain which groups of systems.
- The point of specifying a representative group in terms of system groups is that there may be a group of distinct systems that all behave the same way. For the purposes of this question, we can consider these distinct systems as part of the same grouping.
Page 4: Tools
In this page, we simply ask about the various software tools used in your CPI production systems.
Page 5: System Age and Update Frequency
- In this page, we are trying to learn how old the majority of systems are, and approximately how frequently the majority of systems are updated.
- We are only looking for coarse-grained answers here. The answers are presented as intervals so that you need not know precicsely how old a system is or exactly how frequently it is updated.
Page 6: Number of Individuals to Implement Changes
- In this page, we are trying to learn how many individuals need to be involved with small and large changes to systems in each step.
- Again, in this question we are asking about the typical system in your organization.
Page 7: Obstacles to ADS Adoption
- In this page, we are trying to understand what the implementation gap in adoption of ADS in the production of the CPI.
- We also ask respondents to rank various common challenges to see if there are any challenges that are commonly faced across organizations.
Page 8: How long changes take
- In this page, we are trying to understand how long it takes to introduce various changes to systems in each step.
- Similar to previous questions, we are asking for this response about the typical system in each step.
Page 9: General Challenges Facing CPI Teams
- In this page, we are trying to understand the primary challenges facing CPI teams across organizations, regardless of whether ADS are used.
- Our hope here is to better understand if there are any common challenges facing organizations.
Pages 10 and 11: General Feedback and Task Team Feedback
- Page 10 provides a plain text field where you can write any additional information about your CPI systems (we will personally read this field for each respondent).
- Page 11 is for users to provide feedback about the Task Team on Scanner Data.
What we hope to learn
- Provide descriptive statistics to characterize the state of CPI systems around the world in 2025.
- Discover "archetypes" for organizations that have certain characteristics in common.
- Correlate archetypes with business outcomes of interest.
- Where possible (if anywhere), make inferences about changes that can lead to improved busienss outcomes.