SIS SIG - Meeting 19 - 2025.10.21

1. Opaque governance: extensions defined without our input, unpredictability

…leads to spec instability

• Strong implementation partner that understands SIS vendor pain points, open channel to share feedback across partners/agencies

• First draft review by vendors (versus current process is to give a near final or final draft). Should be before SEA does significant work

• Plan with data available (see flipped model)

2.Spec instability: scope projects before specs are finalized

• See above

• Avoid accelerating the timeline by reducing the draft/review/analysis period. Stick to the timeline and it will reduce rework, complications in the future

• Flipped model: turn on core Ed-Fi data, reconstruct calculations from the existing data and identify gaps in the data. Collect all data via core profile, then build from existing data. +2

3. Legacy systems persist: CSV pipelines are not retired, creating dual systems

• Recommend to states to collect granular data and drive logic based on that granular data

• In some cases (funding, membership, accountability), the state will not address it from granular data if they are not sure, they can collect the calculation and risk guessing wrong. Would rather support a calculation from a legacy system than an extended version in Ed-Fi. "If you don't KNOW what granular data you need, don't guess, just collect the calculated version"

4.Parallel year model: maintain old and new systems simultaneously

5.Underestimated LEA training needs, little or no planning for change management

• States that understand the time it takes and it cannot be rushed, it goes better. States that try to shortcut, rush, or underfund it, puts too much burden. They need to understand that it is a mindset shift for districts.

• Spec instability also makes this worse. Train on one process and change it the following year

• Starting with easier chunks to collect first help with retraining. For example, starting with simpler collections, like demographics that are more 1:1.

5.Supporting multiple API versions without tooling or funding

“Plan first, see data later” leads to unrealistic requirements

Dual reporting and version misalignment

Validation responsibility: expected to validate ODS data without SEA support