Collection Page Merchandising Auditor
Pricing
from $11.50 / 1,000 collection page analyzeds
Collection Page Merchandising Auditor
Audit collection/category pages for product count, sort/filter signals, merchandising blocks, out-of-stock products, duplicate tiles, price ranges,
Pricing
from $11.50 / 1,000 collection page analyzeds
Rating
0.0
(0)
Developer
junipr
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
7 days ago
Last modified
Categories
Share
Audit public ecommerce collection pages at product-card level. The Actor inventories cards and reports duplicate tiles, promoted out-of-stock products, missing links/prices/images, filters, sort options, price range, and pagination evidence.
It is designed for ecommerce merchandisers, SEO teams, category managers, and agencies. Runs are deterministic for supplied HTML, bounded for explicit URL inputs, and produce schema-valid dataset objects plus downloadable report artifacts.
What you get
- Collection URL and name
- Product name, URL, card position, numeric price, and availability
- Total/visible product counts
- Filter count and sort options
- Out-of-stock count and duplicate group
- Numeric page price range
- Pagination status and merchandising issue labels
Every row includes source URL/type, page title, record type/name, issue code, severity, an optional bounded evidence snippet, a recommendation, canonical URL, HTTP status, and scan timestamp. Missing evidence remains null or an explicit diagnostic state; it is not replaced with invented values.
How it works
- Read supplied collection HTML or bounded explicit collection URLs.
- Optionally follow same-domain
rel=nextlinks withinmaxDepthandmaxCollectionPages. - Extract visible product cards, names, links, numeric prices, images, and availability.
- Calculate page-level product count, visible count, filter count, sort options, out-of-stock count, and price range.
- Emit one object per product card with concrete issue labels; empty collections receive an explicit diagnostic object.
The Actor prefers user-supplied HTML because it is deterministic and inexpensive. Live HTTP retrieval is bounded by page limits, timeouts, delays, and an optional domain allowlist. Loopback and private IPv4 targets are rejected. Broad link discovery is disabled.
Inputs
| Field | Purpose |
|---|---|
collectionUrls | Explicit public category/collection URLs. |
htmlInputs | Preferred captured collection HTML. |
followPagination | Enables only bounded same-domain rel=next traversal. |
maxCollectionPages / maxDepth | Page and pagination bounds. |
maxProductCards / maxProductsPerCollection | Card caps. |
extractFilters / detectDuplicateTiles | Optional audit controls. |
maxRules | Maximum issue labels evaluated per card. |
includeEvidence | Set false to omit evidence snippets from dataset rows. |
includeRawSnapshots | Opt in to bounded raw HTML records in the key-value store. |
timeoutMs / requestDelayMs | Bound each public request and pacing. |
maxChargeUsd | Stops before a paid event would exceed the run budget. |
includeReport | Controls paid report generation and key-value-store artifacts. |
startUrls remains available for task compatibility. Actor-specific URL fields are clearer for new integrations. sitemapUrls is accepted as workflow metadata but does not start a broad sitemap crawl.
Input example
For repeatable automation, store baseline or captured HTML outside the Actor and send only the bounded records needed for the current comparison. Avoid embedding credentials, session cookies, private customer data, or authenticated page captures.
Dataset contract
The dataset schema is defined in .actor/dataset_schema.json. Each paid output is pushed as an individual object. Arrays are never submitted as a single dataset item. This makes API pagination, CSV export, and downstream validation predictable.
Common diagnostic conventions:
*_ok: observable fields passed the implemented checks.*_requiredor*_no_*: required source evidence was absent.*_source_unavailable: the supplied/fetched source was blocked or unavailable.- Actor-specific issue codes identify the exact mismatch, change, merchandising issue, review condition, or variant defect.
- HTTP status
0means the public request failed before a response;nullmeans no request was attempted.
Evidence snippets are bounded and can be disabled. Raw snapshots are off by default and, when enabled, are capped before storage.
Reports
When includeReport is true and its charge is accepted, the Actor writes:
- a Markdown executive summary;
- a CSV containing the schema-aligned records;
- a JSON issue list;
- bounded evidence excerpts;
- run metadata with counts, issue codes, pricing mode, and consumed input fields.
The billing-status record reports accepted spend, configured maximum, and the number of rows pushed. If a charge is rejected, partially accepted, or reaches an Apify event limit, the Actor stops before the corresponding paid output.
Pay-per-event pricing
| Event | Price | When charged |
|---|---|---|
actor-start | $0.03500 | Once after initialization. |
collection-page-analyzed | $0.01150 | Once per distinct collection page. |
product-card-record-extracted | $0.00650 | For each extracted product card. |
merchandising-issue-detected | $0.01150 | For each distinct card issue result. |
executive-report-generated | $0.40000 | When report output is requested. |
Apify platform-usage pass-through is off. The configured event prices reserve the applicable platform and discount margin. maxChargeUsd is enforced before every event, including actor start and report output. A value of zero therefore produces no paid dataset output.
Billing sequence:
- Initialize the Actor.
- Accept the
actor-startevent. - Accept the distinct source/record event.
- Accept any actor-specific comparison, extraction, theme, issue, or change event.
- Push one validated dataset object.
- Accept report events before storing paid report artifacts.
- Write billing status and exit.
The Actor exits in a finally path. Dataset push failures are recorded as dataset-push-failed before the error is rethrown, which keeps failures observable and prevents a hanging run.
Limits and interpretation
- Duplicate detection uses normalized product URL, or product name when a URL is absent.
- An out-of-stock card is flagged as a merchandising observation, not as proof that it should be removed.
- Filter and sorting output reflects observable controls only; backend facet correctness is not asserted.
- The Actor does not add products to carts or trigger storefront actions.
Additional safeguards:
- Only
http:andhttps:URLs are eligible for fetching. - Localhost, loopback, link-local, and common private IPv4 ranges are rejected.
allowedDomainscan restrict retrieval to exact hosts and their subdomains.- Response HTML is truncated before analysis.
- Pagination, where supported, stays on the same hostname and within configured depth/page limits.
- No credentials or private API access are requested.
- Results describe observable source evidence at scan time; they are not legal, financial, inventory, or compliance guarantees.
Running on Apify
Use the input editor, API, task, webhook, schedule, or integration of your choice. Keep automated runs small at first, inspect the dataset and billing-status record, then increase caps deliberately.
A successful default run uses embedded public-page-shaped HTML, performs no network request, finishes quickly, and writes a non-empty dataset. For production monitoring, replace that input with your own public URLs or snapshots and preserve prior snapshots externally where comparisons require them.
Running from source
The smoke command clears package-local storage, executes the real entrypoint with examples/input.tiny.json, writes one object per dataset file, and creates every configured report artifact. The test suite also checks schema types, Store defaults, caps, SSRF guards, pricing events, charge limits, real entrypoint behavior, and actor-specific extraction cases.
Troubleshooting
The run returns a required-input diagnostic
Confirm that the actor-specific source field is populated and that the HTML record contains sourceUrl, html or currentHtml, and an appropriate HTTP status. Comparison Actors also need a usable baseline.
The source is unavailable
Check the URL, public accessibility, domain allowlist, timeout, and robots/site policies. Prefer a captured public HTML snapshot when stable automation matters.
The dataset stops before the configured cap
Inspect the billing-status record. maxChargeUsd, an Apify event limit, or a partial charge can stop output before a row is pushed.
Expected fields are null
The field was not observable in the supplied source or was disabled by input. Review the evidence snippet and source HTML. Do not treat null as a negative factual claim.
Interactive content is missing
Supply server-rendered HTML or an authorized captured snapshot. This package intentionally analyzes static, observable evidence and does not claim browser-only interactions it did not perform.
