Based on the knowledge sources, Elementary already primarily operates at the end of a dbt run by utilizing

on-run-end

hooks. These hooks are designed to extract data from dbt's

results

and

graph

objects and upload them to the Elementary schema after your models or tests have finished executing [[Elementary dbt package on-run-end hooks](https://docs.elementary-data.com/data-tests/dbt/on-run-end_hooks#page-title)]. However, if you are referring to the Elementary models themselves (the tables where results are stored), there are specific recommendations for when and how they should run: ### 1. Running Elementary Models You do not need to run Elementary models (e.g.,

dbt run -s elementary

) every time you run your production pipeline. These models only need to be built: * During the initial installation. * After a minor version upgrade (e.g., upgrading from 0.7.5 to 0.8.0) [[Slack forum user question](https://elementary-community.slack.com/archives/C02CTC89LAX/p1741272625948369?thread_ts=1741272625.948369&cid=C02CTC89LAX); [Slack forum user question](https://elementary-community.slack.com/archives/C02CTC89LAX/p1689861891276609?thread_ts=1689861891.276609&cid=C02CTC89LAX)]. During standard runs, the

on-run-end

hooks will automatically populate these existing tables with the latest data [[Elementary dbt package on-run-end hooks](https://docs.elementary-data.com/data-tests/dbt/on-run-end_hooks#page-title)]. ### 2. Excluding Elementary from Daily Runs If you notice Elementary models are running at the beginning or middle of your standard

dbt run

or

dbt build

, you can explicitly exclude them to ensure they don't consume resources unnecessarily:

bash dbt run --exclude package:elementary

[[Slack forum user question](https://elementary-community.slack.com/archives/C02CTC89LAX/p1747229110843109?thread_ts=1747229110.843109&cid=C02CTC89LAX)]. ### 3. Running the EDR CLI For production observability, it is recommended to run the Elementary CLI (

edr monitor

or

edr send-report

) at the very end of your pipeline, after your dbt runs and tests are complete, to ensure the alerts and reports contain the most up-to-date data [[Slack forum user question](https://elementary-community.slack.com/archives/C02CTC89LAX/p1684518982383579?thread_ts=1684518982.383579&cid=C02CTC89LAX); [Slack forum user question](https://elementary-community.slack.com/archives/C02CTC89LAX/p1709134617550489?thread_ts=1709134617.550489&cid=C02CTC89LAX)]. ### Summary of Order A typical efficient production flow looks like this: 1.

dbt run --exclude package:elementary

(Runs your models) 2.

dbt test --exclude package:elementary

(Runs your tests) 3. Automatically: Elementary

on-run-end

hooks update the metadata and result tables. 4.

edr monitor

(Sends alerts based on the data just collected).