| Next revision | Previous revision |
| software:metrics [2024/06/05 16:45] – created smartin015 | software:metrics [2024/06/05 16:47] (current) – [Creating New Metrics] smartin015 |
|---|
| There's some prior work on understanding the ebb and flow of membership: | There's some prior work on understanding the ebb and flow of membership: |
| |
| Various scripts in the Scripts & Colabs shared drive folder are hacked-together attempts to pull data from Neon, Airtable, and Sheets. | * Various scripts in the [[https://drive.google.com/drive/u/3/folders/1HP1lX4PmWm_EefC0_tdp9oN_Bu-s-9s1|Scripts & Colabs shared drive folder]] are hacked-together attempts to pull data from Neon, Airtable, and Sheets. |
| | * The [[https://colab.research.google.com/drive/1LlrLj3vkUepvqL6VKYMVqegMED6AZ_KJ?authuser=3|Class Registration Data Extraction]] colab pulls event and registration data from Neon and dumps it to CSV for later analysis. This can be useful for detecting occurrence/frequency of particular classes, occupancy and earnings etc. It doesn't include any walk-ins (e.g. techs backfilling not-quite-full classes) which are charged through Square. |
| The Class Registration Data Extraction colab pulls event and registration data from Neon and dumps it to CSV for later analysis. This can be useful for detecting occurrence/frequency of particular classes, occupancy and earnings etc. It doesn't include any walk-ins (e.g. techs backfilling not-quite-full classes) which are charged through Square. | * The [[https://colab.research.google.com/drive/1UkpGV1on_rSV0aCdptDOi9HLY3PGYf4g?authuser=3|Identify paying, active members for a given period]] colab crawls through all accounts on Neon CRM, accumulates membership information, and uses the data to compute a histogram of new, leaving, returning, and sustained membership on a month-to-month basis. It's not super optimized and takes several minutes to run. |
| The Identify paying, active members for a given period colab crawls through all accounts on Neon CRM, accumulates membership information, and uses the data to compute a histogram of new, leaving, returning, and sustained membership on a month-to-month basis. It's not super optimized and takes several minutes to run. | * Please be careful when executing scripts here, as some of them actually mutate data when you run them. |
| Please be careful when executing scripts here, as some of them actually mutate data when you run them. | |
| | |
| Materialization of these scripts from previous runs: | Materialization of these scripts from previous runs: |
| |
| * Membership analysis sheet using the output of the paying/active members colab above, plus a couple of rules and charts to track how we're doing on acquisition | * [[https://docs.google.com/spreadsheets/d/17vqm_xuvpfTXa_o6ES3xk-eFWNgULQoFLdtoERLUtdk/edit#gid=0|Membership analysis]] sheet using the output of the paying/active members colab above, plus a couple of rules and charts to track how we're doing on acquisition |
| |
| ===== Creating New Metrics ===== | ===== Creating New Metrics ===== |
| |
| I recommend thinking carefully about: | Some questions to answer before creating a new metric or dashboard: |
| |
| * What questions do we always need to answer, on a regular enough basis that it's worth paying (time, attention, effort) to maintain a dashboard/pipeline for? What is this frequency/basis? Is the audience just the M&P committee, the whole board, or broader? | - What questions do we always need to answer, on a regular enough basis that it's worth paying (time, attention, effort) to maintain a dashboard/pipeline for? What is this frequency/basis? Is the audience just the M&P committee, the whole board, or broader? |
| * For each of these questions, can it wait a few minutes, or do we need an answer at any moment? (operational metrics vs higher-level reporting) | - For each of these questions, can it wait a few minutes, or do we need an answer at any moment? (operational metrics vs higher-level reporting) |
| * What would be an ideal "data view" or schema that would support deeper, ad-hoc analysis to answer various questions that emerge from the questions in #1? | - What would be an ideal "data view" or schema that would support deeper, ad-hoc analysis to answer various questions that emerge from the questions in #1? |
| |
