USPTO Trademark Keyword Search

Research trademark wordmarks with structured filing dates, registration details, displayed owners, goods summaries and official record links. Start with one phrase and export a small set of matches as CSV or JSON.

Look up wordmark phrases in the public USPTO search interface and export structured matches for brand research. This is keyword discovery, not serial-number-only lookup, a legal opinion, trademark clearance, or an exhaustive register search.

Input

{"queries": ["USPTO"], "maxResults": 10}

The Actor accepts up to five literal phrases and 100 total unique serial numbers. Each phrase reads at most two normal result pages through the built-in Results on page export. It does not use the bulk export option, log in, solve challenges or use proxies. The default is one phrase and ten records. The runtime deadline is 90 seconds at 1 GB; a large input can hit that limit and fail before output is stored. Each phrase is limited to 100 characters; expert search operators are not supported.

Pricing

Launch price: $25 per 1,000 complete unique trademark records. This is $2.50 per 100 records, or $0.025 per record. A ten-record export costs $0.25. This is a margin-first price for a bounded browser export, not a claim to be the cheapest data source.

Only validated unique records stored in the dataset are eligible for the default dataset-item billing event. Diagnostics, incomplete cards, duplicate serial numbers and empty matches are not stored as paid rows. Source failures before persistence create no synthetic results. A persistence or summary failure is not a promise of an automatic refund for rows already stored; check the run's charge record in that case.

RUN_SUMMARY.billableRows counts eligible stored records, not collected customer revenue. Account billing and run charges are authoritative. Repeated runs returning the same record are separate exports, not a change-monitoring subscription.

Output

Each row contains the serial number, exported wordmark, live/dead status, optional registration-state badge, classes, displayed owner, goods/services summary, observation time, matched queries and official record URL. Official exports additionally supply filing date, filing basis, priority date, registration date/number/type and supplemental registration date where present. Missing values remain null. Export query, serial order and page ranges must match the visible results. Images and owner postal-address fields are discarded, not returned.

Duplicates merge by serial number while retaining query-level observations. An absent second status badge remains null; live does not mean registered. Certification and collective classes A, B and 200 are retained separately from Nice classes 001-045. Records with absent owner or goods summaries are excluded from the dataset and disclosed in the free summary. If only incomplete cards are available, the run fails.

Wordmarks and goods/services can be truncated. Explicit flags retain this limitation; goodsTextComplete is always false because the source export identifies its goods field as truncated. Dates and registration numbers come from the export, never inference. Source totals may exceed the bounded export. Empty results do not establish name availability.

Free RUN_SUMMARY records coverage, caps and failures. Diagnostics never enter the results dataset. A source, parser, authentication or throttle failure fails the run rather than writing synthetic results.

First Run And Export

In Apify, supply one phrase with maxResults=10, click Start, inspect the results and RUN_SUMMARY, then export useful rows as JSON or CSV. Follow recordUrl for manual verification at the source. In an agent/MCP workflow, pass the same bounded input and inspect the summary before treating returned matches as complete. Do not schedule bulk register harvesting through this Actor.

Open the Actor input page, keep one phrase and maxResults=10, review Pricing, then click Start in Apify Console. The five saved example Tasks are not publicly available yet. The input page is the working first-run route; it does not contain precomputed results.

Research Workflows

Collect initial wordmark matches for a brand-research worksheet, compare filing and registration evidence across phrases, or shortlist official records for professional review. Keep source links with exported evidence and preserve nullable fields. Query order matters when the overall record limit is reached.

Troubleshooting

Inspect RUN_SUMMARY for excluded incomplete records, query states, page coverage and source caps. A successful empty query does not prove a proposed mark is available. Reduce the number of phrases if a run reaches the deadline. Report reproducible problems through the Actor's Issues tab with the run link and non-sensitive input. Do not include credentials or confidential trademark plans in a public issue.

Responsible Use

Source: United States Patent and Trademark Office, https://tmsearch.uspto.gov/. Respect https://www.uspto.gov/terms-use-uspto-websites and source rate/access controls. The website interface is not a bulk-data service; the Actor stops on access denial and has no circumvention route. It does not return trademark images, reproduce the USPTO seal, enrich private contact details or imply government endorsement. Rights in marks remain with their owners. Seek qualified advice for consequential trademark decisions.

Explore other Neuton research Actors for separate data workflows. Successful technical tests establish bounded extraction behavior, not customer demand, legal clearance or guaranteed future availability.