Short answer
Digital SOP and connected-worker platforms hold two things worth asking for: structured procedure content with steps, media and conditions, and execution records showing who completed which step, when, with what entries or photos. Execution records are closer to practice than a written SOP. They carry no motion or force data, and both the platform's terms and the customer's rights govern whether they can be licensed.
The buyer's problem
A written SOP says what should happen. Whether it happened is a different question, usually answered by paper forms that nobody digitized. Operators who moved procedures onto tablets or phones now hold structured step content plus a log of each time it was run. That pairing, procedure and execution, is rare in other sources and valuable for evaluating whether a plan matches practice.
The difficulty is access. Data sits in a vendor's cloud under a customer agreement. Export tools vary, and what the operator can lawfully pass on depends on contract terms. This page covers the asset, what to request and what to settle before spending effort. For paper equivalents, see completed checklists as execution records.
Two layers of data
Procedure content (the template): a title, revision, applicability, ordered steps, each with text, optional images or video clips, conditions or branching, required inputs and sign-off requirements.
Execution records (the instances): one record per run, containing user or role, start and end times, per-step completion times, entries such as measurements or pass/fail flags, comments, photos or scans captured at steps, exceptions raised and approvals.
The quality management context explains why both exist. ISO 9001:2026, the current edition of the quality management system requirements, covers documented information within support and operation requirements, and describes requirements that are generic and applicable to any organization, with certification optional (ISO). Many operators therefore maintain controlled procedures and records of their use whether or not they use a digital platform. In regulated pharmaceutical production, US rules require procedures to be followed and documented at the time of performance and any deviation to be recorded and justified (21 CFR 211.100). Digital platforms are one way to meet such obligations; the deviation side is covered in deviation reports.
What this data can support, and what it cannot
It can support:
- Procedure-versus-practice analysis. Step timings, skipped steps, repeated steps and exception rates show where the written method diverges; see procedural drift.
- Realistic step durations as priors for planning and for segmenting video.
- Success and failure labels. Per-step pass/fail entries and exceptions give outcome labels that SOP text lacks.
- Visual evidence. Photos captured at steps pair a procedure state with its visible result, discussed on photo evidence of step completion.
- Training and competency context, where platforms track sign-offs; see competency assessments and signoffs.
It cannot support:
- Physical control. A tap on "step done" records completion, not how the hands moved.
- Sensor streams, unless the platform integrates instruments, and then only the readings it logs.
- Unbiased behavior. Workers know they are being recorded; people tick boxes in batches after the fact, so timestamps can mislead.
- Complete work. Tasks outside the platform, informal fixes and verbal instructions are absent.
- Generalization across sites. Execution patterns reflect one operator's staffing, equipment and culture.
Illustrative export package
# Illustrative - synthetic identifiers and values
export: connected_worker_v0
procedure_templates:
count: "supplier-stated"
fields: [template_id, revision, steps[], step_media[], conditions[], required_entries[]]
media: {images: true, short_clips: true, licensed_with_template: "to confirm"}
executions:
count: "supplier-stated"
fields:
- execution_id: ex_0008812
template_id: tpl_0042
revision: "D"
role: operator_role_3 # pseudonymous role, not a name
site: site_017
started_at: "synthetic-iso-time"
steps:
- step_id: "5"
completed_at: "synthetic-iso-time"
entry: {type: numeric, value: 11.8, unit: "N*m"}
photo: img_0048121
result: pass
exceptions: [{step_id: "7", reason_code: "part_missing"}]
approver_role: supervisor_role_1
timing_note: "completion times may be batched; supplier to report sync behavior"
removed: [worker_names, free_text_names, device_ids]
Evaluation points
- Revision linkage. Each execution must name the template revision it ran, or step IDs lose meaning.
- Entry types. Count free-text, numeric and photo entries; free text is where personal information hides.
- Timestamp semantics. Is time recorded on tap, on sync, or by the device clock? This changes whether durations are usable.
- Exception coding. Controlled reason codes are far more useful than comments alone.
- Media retention. Platforms may purge photos after a period; ask what is still stored.
- Site and role diversity. One plant's executions teach one plant's habits.
- Sample before scale. Request a small redacted extract and test the schema before negotiating a bulk export.
Platform terms and customer rights
This asset is unusual because three parties are involved: the platform vendor, the customer who authored the procedures and recorded executions, and the workers who generated the records. The vendor's terms may state who owns customer data, whether aggregate or derived data may be used, and what export rights the customer has. The customer may owe confidentiality to its own clients. Workers may be identifiable through role, time and site even without names.
Diligence questions: Does the customer own the content and execution data under the platform agreement? Does the agreement permit export for licensing, or restrict transfer of data to third parties? Does the vendor claim any right in the data? Are worker notices and works-council or union agreements in place? This is not legal advice; resolve these with counsel. Removing worker names does not settle whether the operator may license the records.
Limitations and additional preparation
- Export fidelity. Bulk export tools may flatten branching, drop media or omit conditions. Request a test export.
- Schema variation. Each platform has its own model; mapping to a common schema such as the procedure step schema is preparation work.
- Free-text hygiene. Comments can include names, phone numbers and incident details.
- Selective use. Operators adopt platforms for certain procedures, usually audits, inspections and changeovers, so coverage is uneven.
- No behavior ground truth. Execution logs show recorded completion, not quality of performance.
- Pair with demonstrations. To connect steps to hand motion, use recordings aligned to the platform's step IDs; see SOP-to-video step alignment.
Structured authoring systems that create the content upstream are covered in structured authoring exports.
Questions to ask suppliers
- Which platform and version, and what export formats does it offer?
- Does your agreement with the platform vendor permit export and licensing of both templates and executions?
- How are template revisions linked to each execution?
- What do timestamps represent, and are completions ever batched?
- What entry types appear, and how is free text reviewed for personal data?
- Are photos or clips retained, and who owns them?
- How many sites, roles and procedures are covered, and over what period?
- Which fields have you removed, and can the removal be documented?
- Are there worker notices or labor agreements affecting use?
Reading execution records critically
Execution logs reward careful reading before any modeling. Look at the distribution of step durations: a spike at one or two seconds usually means a worker tapped through several steps at once after finishing the job, so those steps carry no timing information. Look at the share of runs with zero exceptions; a perfect record at scale suggests that problems are handled outside the platform. Compare the order steps were completed in with the template order, since platforms that allow free navigation reveal real sequencing habits, while those that enforce order hide them.
Revision history matters as well. When a template changes mid-period, executions of revision C and revision D are different tasks under one name; the SOP revision histories page explains why a delivery should keep both and label them. A buyer who wants step-level durations as priors should ask for the raw timestamps and the sync rules, then decide which runs to exclude, instead of accepting a pre-computed average.
Discuss your requirements
If you are looking for procedures paired with records of how they were run, describe the task domain, record types and permitted uses to SourceX. SourceX looks for US businesses that hold the material, checks the data and the supplier's licensing permissions, and coordinates the license and delivery. A request does not guarantee a match, and SourceX follows up about potential sources. Start at the humanoid training data guide or the procedures hub; the catalog page on workflow and task histories describes related operational data.
Discuss your humanoid training-data requirements with SourceX
Sources
- ISO 9001:2026 Quality management systems - Requirements — documented information and process requirements; certification optional.
- 21 CFR 211.100 Written procedures; deviations — procedures documented at performance; deviations recorded.