How It Works
Eight Stages from a Product Row to a Published Product
You give GearDock a spreadsheet and whatever manufacturer documents you hold. Each stage below takes a specific input, produces something you can see, and hands it to the next. You decide the identity questions, the approval, the release, and where content is allowed to go.
The Eight-Stage Workflow
Intake
Establish the product
- 01System work
Import Product Data
Bring in a spreadsheet, selected cloud file, PIM profile, or connected catalog.
- 02Human control
Establish Product Identity
Resolve each row with identifiers. Ambiguous or conflicting identity waits for a person.
Evidence
Make support visible
- 03System work
Connect Trusted Evidence
Link manuals and spec sheets to products so facts can cite exact passages.
- 04System work
Build Structured Product Facts
Surface missing, conflicting, weakly supported, and unavailable information.
Governance
Draft and decide
- 05System work
Generate Governed Content
Draft website-ready copy from supported facts, then run the Claim Firewall.
- 06Human control
Validate Claims and Resolve Issues
Run integrity and Claim Firewall checks, then resolve blocked gaps or conflicts before review.
Distribution
Release with control
- 07Human control
Review, Approve, and Release
Release approved content through a separate, recorded governance gate.
- 08System work
Distribute and Monitor
Send released content to an authorized destination and retain delivery evidence.
Step by Step
What Actually Happens at Each Stage
Every stage lists what it takes in and what it hands on, so you can tell exactly where your data is at any point and which steps need a person.
Import Product Data
You upload the spreadsheet you already keep and tell GearDock which column is which. Every row is checked on its own, so one malformed row cannot block several thousand good ones. Nothing reaches your catalog until you have looked at the batch.
Product import →- Takes In
- A spreadsheet export, a supplier list, or a single product record
- Produces
- A batch of candidate rows waiting for your review
- A result for each row, with the reason for any failure
Establish Product Identity
GearDock works out which real product each row refers to, starting with manufacturer and model or part number and widening from there. Each candidate comes with how confident the match is and why it was proposed. Everything after this step depends on getting it right.
You confirm the uncertain matches. Uncertainty tends to cluster, so one decision usually clears many rows at once.
Product matching →- Takes In
- Checked rows and whatever identifiers they carry
- Produces
- Confident matches, which carry on
- Uncertain matches, held with the candidates shown side by side
- Rows that match nothing, sent to an exceptions list
Connect Trusted Evidence
Brochures, spec sheets, and manuals go into your organization's document library. GearDock reads them so their contents can be searched and quoted instead of retyped. One manual often covers a whole family of products.
Document intelligence →- Takes In
- Source documents you upload
- The products they were matched to
- Produces
- Document contents you can search
- Links between each document and the products it covers
Build Structured Product Facts
GearDock compares what it knows about each product with what a product of that type normally carries, and reports the difference. For most teams this is the first honest picture they have had of their own catalog.
Content completion →- Takes In
- Product records and the documents linked to them
- Produces
- Supported values, each with the passage that backs it
- Weakly supported values worth a second look
- Conflicts, where two documents disagree
- Gaps, where no document you hold covers the value
Generate Governed Content
A title, a summary, a full description, and formatted specifications are written from those structured facts. Each factual statement keeps a link to the passage behind it. Anything no document supports is left out rather than filled in.
Claim Firewall →- Takes In
- Structured facts
- The evidence passages
- The gap report
- Produces
- A draft with its citations attached
- Claim Firewall findings on anything it could not tie back to evidence
Validate Claims and Resolve Issues
A person reads the draft with the evidence beside it and records approve, ask for a revision, or reject, along with the reason. Review is fast or slow depending almost entirely on whether the source is right there, which is why it is.
This step is always a person. GearDock does not approve its own drafts, and there is no setting that changes that.
Review and release →- Takes In
- The draft, its evidence, and any Claim Firewall findings
- Produces
- A recorded decision with a name, a time, and a reason
- An approved version, or a revision request explaining what is wrong
Review, Approve, and Release
Release marks one approved version as allowed to leave GearDock. It is a separate decision from approval, and it is the last point where changing your mind costs nothing outside GearDock.
Release is decided per item by a person, not flipped on for a whole batch.
Review and release →- Takes In
- Approved content
- Produces
- A released version that distribution is allowed to pick up
- A recorded release decision tied to that exact version
Distribute and Monitor
GearDock packages the released version as a file or sends it to a system you have connected. Before it sends anything it re-checks that the version is still released, it delivers into the destination's draft state rather than making it live, and it records the outcome.
Exports and integrations →- Takes In
- Released content
- A destination you have configured and authorized
- Produces
- A downloadable package, a draft record at the destination, or a file drop
- A permanent record of what was sent, where, by whom, and how it went
The Output
What a Finished Product Record Looks Like
Every value carries its evidence state alongside it. This is the screen a reviewer works from, and it is the reason review is fast enough to run across a whole catalog.
- A supported value names the document and the passage it came from.
- A gap is shown as a gap. The field stays empty.
- A conflict is put in front of you rather than settled quietly.
- The whole record is versioned, so what was approved stays visible.
| Attribute | Value | Evidence |
|---|---|---|
| Manufacturer | Acme Medical | Source-backedAM-4200 datasheet, p.1 |
| Model | AM-4200 | Source-backedAM-4200 datasheet, p.1 |
| Flow rate range | 0.1-999 mL/h | Source-backedAM-4200 datasheet, p.4 |
| Weight | 2.4 kg | Source-backedService manual rev. C, p.12 |
| Battery runtime | No source | GapNo source supports this |
| Power input | 100-240 V AC | Conflict2 sources disagree |
Illustrative example using fictional product data.
The One Rule That Shapes the Whole Sequence
Common Questions
What People Ask After Seeing This
How long does one product take?
The machine steps are quick. What sets the pace is human review and how many documents you hold. A product with a good spec sheet attached reviews in a couple of minutes. A product with no documents at all needs a decision about whether to publish it yet.
Do I need documents for every product before I start?
No, and almost nobody has them. Products with documents finish quickly and can be published first, while the gap report tells you which documents are worth collecting next. Partial coverage is the normal starting point, not a problem to solve first.
What happens to rows that do not match anything?
They go to an exceptions list instead of being processed anyway. A row that matches nothing is usually a data problem, such as a mangled identifier or an internal SKU sitting in the model column, and it is worth fixing in the source file.
Can I edit the generated content?
Yes. Editing creates a new version. Earlier versions stay in the product's history along with any decisions recorded against them, so what was approved is still visible later.
Does anything publish on its own?
No. Content needs human approval, then a separate release decision, and an export re-checks the release state at the moment it runs. Shopify delivery creates draft products that you make live yourself.
What if some products already have descriptions?
Bring them in with the import. GearDock reads the values you already hold rather than replacing them, then reports which of those values a document supports, which conflict with one, and which nothing you hold backs up.
Private Beta
Run This on Your Own Products
Bring a few hundred representative products and whatever documents you hold for them. Seeing the gap report on your own catalog tends to settle the question faster than any description of it.