Nintex DocGen Admin Checklist: Monthly & Quarterly Routine

Estimated effortXS hours S a day or two M about a week L weeks XL a workstreamIndicative, relative sizes for a typical mid-size org — your org’s evidence decides. Dates and estimates belong to your implementation partner.

The DocGen health loop: scan, grade, decide, fix, verify by re-scan — The loop the routine runs: scan, grade, decide, fix — verified only when the rule goes silent.

Every month — about an hour

The Monthly Hour S

Re-scan the estate and compare the report card with last month’s: any area that changed standing, and why.
Clear new Requires Attention items first — the ones that can stop a document: template rows with no file, relationships reading objects that are not there, missing starting objects, incomplete packages.
Answer new decisions, or accept the recommendations after reading each one.
Approve the quick wins that are waiting; after the next scan, confirm each shows verified (its rule went silent).
Check that nothing named TEST, SAMPLE or COPY became active in production.

The Fix and validate queue: approved quick wins waiting, each verified by re-scan — The monthly hour’s workbench: the fix queue, with each applied item proving itself on the next scan (test estate).

Every quarter

Rationalize M

Rationalize: consolidate near-duplicates and per-variant packages; retire version clones and packages unchanged and unsent for a year.
Naming: one convention for packages, delivery options and buttons; one word for each kind of delivery.
E-mail: every e-mail option has a subject source, a body source and help text; frequently sent wording lives in a Salesforce e-mail template.
Storage: generated output saved as Files, not Attachments; templates stored as Files, not classic Attachments or Documents; no macro-enabled templates without a reason; no template file far above the org’s norm.
Launch: no JavaScript buttons; every button labelled and described; package and delivery not left fully open to the user when one was intended.
Automation: generation that should be automatic runs from a Flow, not a legacy outbound message.
License: current, renewal date known, seats assigned against seats used.
SOQL: each SOQL relationship still justified, on one line, and bounded.

On every change

Event-Driven XS

Before a Loop upgrade: read the upgrade and change-risk section of the latest assessment; scan a sandbox on the new version first.
Before a Salesforce release, or a schema change (renamed or deleted objects and fields): re-scan, then check relationships, lookups and starting objects.
When a package goes live: generate it once for a real record, then re-scan; it should land in Good Standing.
When a package is retired: set it inactive first and keep it for one cycle before any deletion.
When the admin changes: hand over the latest assessment, the decision register and the findings export.

Once a year

The Deep Pass M

Run the full assessment including the interview items configuration cannot answer: branding consistency, document standards, testing practice, ownership.
Review by hand the areas configuration can barely see: formula and transformation architecture, deployment and sandbox health, errors and observability.
Review Agentforce document-generation readiness, if it is on the roadmap.

Sources & further reading

References

1. Nintex · DocGen for Salesforce Documentation — The vendor’s own documentation for packages, templates, relationships and delivery. ↩︎

“Salesforce”, “Agentforce” and related marks are the property of Salesforce, Inc. Nintex and DocGen are marks of Nintex. Horus Eye is an independent product and is not affiliated with, endorsed, or sponsored by Salesforce, Inc. or Nintex.

Take it with you as a PDF

Printable download

What you’ll get

The standing routine — the monthly hour, the quarterly rationalization pass, the on-every-change checks and the yearly review — formatted as a printable PDF with tick boxes, ready to hand to whoever owns the estate next.

Everything in the PDF is on this page too — the download is for printing, sharing with the team, and the project binder.

Download the checklist

Protected AI — the Trust Layer

Every AI read in Horus Eye is metadata-first: built from an allowlist, sanitized by pattern, made through one audited gateway and checked by an output firewall on the way back — and AI is off by default, per org. Every request carries an audit id and an AI Data Scope manifest you can open.

See the Trust Layer

The DocGen admin’s routine