Sales Order Entry Performance Audit

1. Executive Summary

2. Methodology

This audit followed the Sonar Transaction Entry Performance Audit protocol, six phases, all read-only:

1. Phase 0 — Environment snapshot (account, role, features).
2. Phase 1 — Script inventory: SuiteQL against scriptdeployment ⋈ script for recordtype='SALESORDER'.
3. Phase 2 — Static code analysis: pattern scans (fileGrep) of each readable source file for known anti-patterns.
4. Phase 3 — Workflow analysis: all active released workflows, trigger types, and target record types.
5. Phase 4 — Custom field surface: customfield rows whose visibleontransactions includes SalesOrd.
6. Phase 5 — Execution-log heat: scriptnote volumes for the Phase-1 script ids.
7. Phase 6 — Bundle attribution: script.frombundle + API version per script.

Every claim is tagged: MEASURED backed by account data · OBSERVED seen in actual source code · HYPOTHESIS plausible, unconfirmed, with the confirming step stated.

3. Environment Snapshot (Phase 0)

4. Findings

Phase 1 — Script inventory MEASURED

13 active deployments (status RELEASED, isdeployed=T) on Sales Order. All 13 are deployed to all roles (allroles=T). Two further deployments exist but are switched off (ATB_UE_SalesOrder, Packship Non-Adv-Inventory UE) — correctly excluded from cost.

Phase 2 — Static code analysis

File 24196, lines 8–14:
    const pageInit = (scriptContext) => {
        disableOrEnableProcessShipmentOnIFFlag(scriptContext);   // ← query on EVERY open
    }
    const fieldChanged = (scriptContext) => {
        if(['shipmethod', 'subsidiary', 'location'].includes(scriptContext.fieldId))
            disableOrEnableProcessShipmentOnIFFlag(scriptContext); // ← query again

File 24196, lines 149–150, 243:
    const getShipCentralPreference = () => {
        var shipPrefQuery = query.create({ type: 'customrecord_packship_shippreference' });
        ...
        return shipPrefQuery.run().asMappedResults();   // ← uncached round-trip

Phase 3 — Workflows MEASURED

Phase 4 — Custom field surface MEASURED

MEDIUM 25 line-level fields multiply with every order line, and 15 of them are sourced — each sourcing evaluation stacks with the 7 client scripts' fieldChanged/postSourcing handlers on the same keystroke. Many are bundle-owned (Electronic Payments CUSTCOL_2663_*/CUSTCOL_9572_* file-format and entity-bank fields, STE tax columns, Packship pack-instructions). No mandatory custom fields — good. Remediation is form hygiene: hide unused bundle fields from the active SO entry form rather than deleting them.

Phase 5 — Execution-log heat MEASURED, limited

The scriptnote table (script reference column: scripttype) contains zero entries for the 13 SO-deployed scripts — log levels are set above DEBUG and/or entries have aged out. Log-volume heat ranking is therefore unavailable in this account; findings above rest on code inspection instead. This also means script logging is not itself a performance drag here.

Phase 6 — Bundle attribution MEASURED

5. Concerns & Risks

• Aggregate stacking. Even where each script is individually clean, seven client scripts execute on every fieldChanged-capable keystroke and six UEs on every open/save. The sum is the product users feel; per-script review alone understates it.
• Vendor-locked hot path. The CRITICAL finding sits in managed-bundle code (534418). It cannot be edited in-account; options are audience narrowing (immediate), vendor escalation (fix), or feature reconfiguration (if "Process Shipment on IF" isn't used on SO).
• Blind spots. Four hidden UE sources + one minified CS mean roughly a third of the synchronous stack was not code-reviewed. APM instrumentation is the way to close this without source access.
• SS1.0 longevity risk. Bundle 47193's 1.0-API scripts predate modern async patterns and are a standing performance and deprecation liability.

6. Prioritized Remediation Plan

1. Narrow deployment audiences (Trivial, admin config): change allroles to targeted role lists on Packship CS and any fulfillment-only deployments. Expected effect: immediate open/per-line improvement for sales roles. Do this first.
2. Escalate the uncached query to the Packship vendor (Moderate, vendor): request module-scope caching of getShipCentralPreference() results. Expected effect: removes 1–2 client-side query round-trips per open and per ship-field change for remaining users.
3. Instrument with NetSuite APM (Trivial, admin install): Page Time Summary will attribute wall-clock per script — including the four hidden UEs and the minified USR script — converting the HYPOTHESIS items to MEASURED.
4. Form hygiene pass (Trivial–Moderate, admin): hide unused bundle CUSTCOL fields (Electronic Payments bank/file-format columns, etc.) from the active SO entry form.
5. Review ATB SalesOrderGateway (Moderate, developer): confirm whether the custom controller makes per-line server calls; if so, batch or debounce them.
6. Plan SS1.0 retirement with the SCM bundle owner (Significant, vendor/dev): upgrade path off 1.0 APIs for bundle 47193 scripts.

7. Next Steps Beyond This Audit

• Timed browser baseline: with DevTools open (Network + Performance), record wall-clock for (a) SO form open, (b) adding one line, (c) save — three runs each, before and after each remediation step. This turns the plan into a measured burn-down.
• NetSuite APM SuiteApp: install and review Page Time Summary and SuiteScript Analysis for the SO record type — the authoritative per-script timing source, and the only visibility into hidden bundle code.
• Re-run this audit after remediation and diff the two scorecards as the close-out artifact.

8. Appendix — Reproducibility

All queries executed via SuiteQL under role id 3 on 2026-07-02. Tool calls: agentSession, accountFeaturesGet, runSql ×10, fileGrep ×9, fileGetLines ×3, evalJs (aggregation).

-- Phase 1: script inventory (15 rows; 13 active)
SELECT s.id, s.name, s.scripttype, s.scriptid, s.scriptfile, sd.id, sd.status,
       sd.isdeployed, sd.allroles, sd.eventtype
FROM scriptdeployment sd JOIN script s ON s.id = sd.script
WHERE sd.recordtype = 'SALESORDER'
ORDER BY sd.isdeployed DESC, s.scripttype, s.name;

-- Phase 3: workflows (10 rows; none target SalesOrder)
SELECT name, scriptid, recordtypes, inittriggertype, initoncreate,
       initonvieworupdate, initsavedsearchcondition, islogenabled, frombundle
FROM workflow WHERE isinactive = 'F' AND releasestatus = 'RELEASED';

-- Phase 4: custom field surface (grouped; aggregated client-side to
-- BODY=31, COLUMN=25 containing token 'SalesOrd'. Note: LIKE/INSTR on
-- customfield.visibleontransactions returns 0 in WHERE context in this
-- account — a SuiteQL quirk; grouping by the raw column then filtering
-- client-side is the reliable pattern.)
SELECT fieldtype, visibleontransactions, COUNT(*) cnt,
       SUM(CASE WHEN source IS NOT NULL THEN 1 ELSE 0 END) sourced,
       SUM(CASE WHEN ismandatory='T' THEN 1 ELSE 0 END) mandatory
FROM customfield WHERE fieldtype IN ('BODY','COLUMN')
GROUP BY fieldtype, visibleontransactions;

-- Phase 5: execution-log heat (0 rows — no log entries for these scripts;
-- note scriptnote's script-reference column is named 'scripttype')
SELECT scripttype, type, COUNT(*) FROM scriptnote
WHERE scripttype IN (1793,473,3365,2182,477,1201,2386,418,2192,801,436,1155,2389)
GROUP BY scripttype, type;

-- Phase 6: bundle attribution + API version (13 rows)
SELECT id, name, frombundle, apiversion FROM script
WHERE id IN (1793,473,3365,2182,477,1201,2386,418,2192,801,436,1155,2389);

Files inspected: 24196 (packship_cs_order.js — full anti-pattern review), 7568/7620 (STE dispatchers), 24576/90318 (LRC), 15869 (ATB CS), 5015 (SCM CPN CS), 24710 (USR CS — minified, unreviewable). Access denied: 5021, 5899, 5093, 24736.

Limitations: static analysis only — no wall-clock measurements; four UE sources and one minified CS unreviewed; scriptnote empty for target scripts; audit reflects account state as of 2026-07-02.