---
title: "Patent Data for Competitive Intelligence: A Practical Guide"
description: "Which patent fields show where a competitor invests (assignee, priority and filing dates, family status), the 18-month publication lag, and a worked example."
url: https://perceptlayer.com/blog/patent-data-competitive-intelligence
---

[Blog](https://perceptlayer.com/blog) / [Web data by industry](https://perceptlayer.com/blog/category/web-data)

# Patent Data for Competitive Intelligence: A Practical Guide

3 Sept 2026 Perceptlayer 7 min read

Patent data shows where a competitor has committed money to an idea, years before a product ships, but it shows it late. The fields that matter are the assignee, the priority, filing and grant dates, the family's status in each country and the classification codes. The catch is timing: applications are generally published 18 months after their earliest priority date, so the newest work is always invisible. Research papers fill part of that gap.

## Key takeaways

-   Filter by `assignee` on `GET /v1/google-patents/search` and sort with `sort=new` to see a company's newest publications; `status` separates `GRANT` from `APPLICATION`.
-   The priority date is the closest public proxy for when the work was done; the grant date is a legal event, often years later.
-   In the recorded search, both patents were granted about 9.5 and 10 years after their priority dates, and one is kind code B1, meaning it had no earlier pre-grant publication.
-   `family_status` shows where protection is active by country: the second recorded patent's WO and EP family members are `NOT_ACTIVE`, a hint about where its owner chose not to pursue it.
-   Scholar results for a topic can show research directions before the patents publish; tracking 10 competitors monthly is about 20 credits.

## What can patent data tell you about a competitor?

Three things, with decreasing confidence. Volume over time says how much a company files in an area, which tracks R&D spending more honestly than press releases. Topic, through titles, snippets and classification codes, says which problems it's working on. Geography, through family status, says which markets it thinks are worth protecting, since filing and keeping a patent in each country costs money.

What it can't tell you is what the company will build. Companies patent defensively, patent ideas they abandon, and keep important work as trade secrets. Treat a cluster of filings as evidence of investment, not of a product plan. And remember the lag: anything filed in the last year and a half is mostly invisible.

## Which fields matter, and why?

The ones that place an invention in time, ownership and geography. The [Google Patents search endpoint](https://perceptlayer.com/apis/google-patents/search) returns them for each result. The table maps each field in the recorded response to what it's good for in competitive analysis.

| Field | What it is | What it tells you |
| --- | --- | --- |
| `assignee` | The owner of the patent | Whose investment it is; normalise names, since one company files under several |
| `dates.priority` | Earliest claimed filing date | Closest proxy for when the work was done |
| `dates.filing` | Filing date of this application | Later than priority for continuations; shows a company extending a family |
| `dates.grant`, `dates.publication` | When it was granted and published | When the public could first see it |
| `publication_number` | Number plus kind code, e.g. `US12438891B1` | `B1`/ `B2` granted, `A1` an application |
| `family_status` | Status per country or office | Where protection is active or has lapsed |
| `title`, `snippet` | Title and a claim excerpt | Topic, for clustering |

Classification codes are the other key field for topic analysis. The Cooperative Patent Classification is the scheme the USPTO and the EPO developed jointly and both use ( [CPC](https://www.cooperativepatentclassification.org/about), 2026), and codes cluster filings far better than keywords. They aren't in the recorded search response; the [Details endpoint](https://perceptlayer.com/apis/google-patents/details) returns a single patent's full record, so check there. The inventor field is also returned; it names people, so this guide doesn't use it.

## Why do the newest filings stay hidden?

Because of the publication rule. In the US, applications are published promptly after 18 months from the earliest filing date they claim, with exceptions such as applicants who certify they won't file abroad ( [USPTO MPEP 1120](https://www.uspto.gov/web/offices/pac/mpep/s1120.html), 2026). Most other major offices publish on the same 18-month schedule. Until then, an application is confidential, so no data source, paid or free, can show it.

That sets a floor on how current patent intelligence can be. A search today shows work whose priority date is at least about 18 months old, plus older applications granted recently. A company that starts filing heavily in a new area this quarter won't appear in patent data until around 2028. Plan analysis around that: patent data is for direction over years, not for this quarter's moves.

Some patents never appear as applications at all. The USPTO's kind codes mark a granted patent with no previously published pre-grant publication as `B1`, and one with an earlier publication as `B2` ( [USPTO, kind codes](https://www.uspto.gov/learning-and-resources/support-centers/electronic-business-center/kind-codes-included-uspto-patent), 2026). A `B1` was invisible until the day it was granted.

## What does the recorded data show? A worked example

The recorded response for `q=machine learning` holds two patents. Here's an excerpt, trimmed with `…`, with inventor names cut:

```json
{
  "success": true,
  "data": {
    "total_results": 122797,
    "total_pages": 99,
    "patents": [
      {
        "patent_id": "patent/US12438891B1/en",
        "title": "Anomaly detection based on ensemble machine learning model",
        "publication_number": "US12438891B1",
        "dates": { "priority": "2015-08-31", "filing": "2022-02-18", "grant": "2025-10-07", "publication": "2025-10-07" },
        "assignee": "Splunk Inc.",
        "family_status": [{ "country": "US", "status": "ACTIVE" }],
        …
      },
      {
        "patent_id": "patent/US12518214B2/en",
        "title": "Distributed machine learning systems including generation of synthetic data",
        "publication_number": "US12518214B2",
        "dates": { "priority": "2016-07-18", "filing": "2023-04-21", "grant": "2026-01-06", "publication": "2026-01-06" },
        "assignee": "Nantomics, Llc",
        "family_status": [{ "country": "WO", "status": "NOT_ACTIVE" }, { "country": "EP", "status": "NOT_ACTIVE" }],
        …
      }
    ],
    …
  }
}
```

Work out the gaps from the dates. They show how far a grant can trail the work it protects:

| Patent | Priority to filing | Filing to grant | Priority to grant |
| --- | --- | --- | --- |
| US12438891B1 | 6.5 years | 3.6 years | 10.1 years |
| US12518214B2 | 6.8 years | 2.7 years | 9.5 years |

Both were filed more than six years after their priority dates, which is consistent with a continuation: a later application in an existing family, claiming the original date. Both were granted in 2025 and 2026 on work first filed in 2015 and 2016. If you counted these as "machine learning patents in 2025", you'd be measuring decade-old work. Count by priority year instead.

The `B1` on the first means its owner received a patent with no earlier pre-grant publication of this application. The second's WO and EP entries are `NOT_ACTIVE`, so its owner isn't, or is no longer, pursuing protection through those routes, while it holds a US grant. Two results aren't a trend, but they show which questions each field answers.

## How do research papers fill the gap?

Papers can appear well before related patents publish, and they name institutions. The [Google Scholar API](https://perceptlayer.com/apis/google-scholar) search takes `q`, a year range with `as_ylo` and `as_yhi`, and `scisbd` for date sorting. The recorded response for `q=machine learning&as_ylo=2020` reports 38,300 results, with each result's `publication_info.summary`, a source and year, and `inline_links.cited_by.total`, 292 for its second result.

Citation counts favour older papers, so compare papers of similar age. Scholar results don't carry an affiliation field in the recording, so to follow one company, search its name together with a topic, or follow known researchers with the [Author endpoint](https://perceptlayer.com/apis/google-scholar/author), keeping in mind that author profiles are personal data. Our [Google Scholar API guide](https://perceptlayer.com/blog/google-scholar-api) covers the parameters.

## How do you set up monthly tracking, and what does it cost?

Run one patent search per competitor per month, filtered by `assignee` and sorted newest first, plus one Scholar search for its main topic. Store results keyed by `publication_number` and count by priority year. These requests haven't been run for this article; the `assignee` value must match how the company appears in patent records, so check it in the playground first:

```bash
curl "https://api.perceptlayer.com/v1/google-patents/search?q=anomaly+detection&assignee=Splunk+Inc.&sort=new" \
  -H "X-API-Key: $API_KEY"

curl "https://api.perceptlayer.com/v1/google-scholar/search?q=anomaly+detection+splunk&as_ylo=2025&scisbd=1" \
  -H "X-API-Key: $API_KEY"
```

Every call costs 1 credit on success; failed calls cost nothing. The recorded patent search reports 99 pages for 122,797 results, so a broad query can't be read in full; filter by assignee, `status`, `before` and `after` to keep result sets to a few pages. Our [Google Patents API guide](https://perceptlayer.com/blog/google-patents-api-search-details) covers paging and when the official bulk datasets are a better fit.

| Job | Calls a month | Credits a month |
| --- | --- | --- |
| 10 competitors, 1 patent search + 1 Scholar search | 20 | 20 |
| The same, plus details for 30 new patents | 50 | 50 |
| 50 competitors, 2 pages of patents + 1 Scholar search | 150 | 150 |

All three fit within the 1,000 free credits every verified account gets each month. See [pricing](https://perceptlayer.com/pricing) for what credits cost and the [credits docs](https://docs.perceptlayer.com/credits) for billing. To try it, [get a free API key](https://perceptlayer.com/signup) and search a competitor's name as assignee in the playground on the [Google Patents API page](https://perceptlayer.com/apis/google-patents). Data is collected from publicly available sources and returned as found at the time of the call.

## APIs in this post

-   [Google Patents **1 credit per call** · 2 endpoints](https://perceptlayer.com/apis/google-patents)
-   [Google Scholar **1 credit per call** · 5 endpoints](https://perceptlayer.com/apis/google-scholar)

## Frequently asked questions

### How is patent data used for competitive intelligence?

Analysts count a competitor's filings by priority year and topic to see where it's investing, read family status to see which countries it protects, and watch new grants and applications for direction changes. It's evidence of R&D spending, not of product plans, and it runs at least 18 months behind because of publication rules.

### Why can't I see a competitor's latest patent applications?

Because they aren't public yet. Applications are generally published 18 months after their earliest priority date, and some US applicants opt out of publication if they don't file abroad. Until publication, no source can show them. Granted patents with kind code `B1` had no earlier publication at all.

### Which date should I use to count patents by year?

The priority date, the earliest filing date a patent claims. It's closest to when the work was done. Grant dates can be years later: the two recorded patents were granted 9.5 and 10.1 years after their priority dates, after continuation filings in 2022 and 2023.

### How much does tracking competitors' patents cost?

Each Google Patents or Google Scholar call costs 1 credit per successful call, and failed calls cost nothing. Ten competitors with one patent search and one Scholar search a month is 20 credits, within the 1,000 free monthly credits. Opening a patent's full record with the details endpoint costs 1 more credit.

### Are inventor names personal data?

Yes. Patent records name inventors, and Scholar results name authors, so handle them as personal data under the law that applies to you. For company-level analysis you rarely need them: assignee, dates, family status and classification codes answer most competitive questions without processing individuals' names.

patents competitive intelligence google patents google scholar r&d

## More from the blog

1.  [Web data by industry](https://perceptlayer.com/blog/category/web-data) 21 Oct 2026 Perceptlayer 8 min read
    
    ## [Job Posting Data for Labour-Market Research](https://perceptlayer.com/blog/job-posting-data-labour-market-research)
    
    Which Indeed posting fields support trend analysis, where postings mislead versus hires, and a weekly collection plan that costs 3,000 credits a month.
    
2.  [API guides](https://perceptlayer.com/blog/category/api-guides) 13 Oct 2026 Perceptlayer 6 min read
    
    ## [Google Patents API: Search Patents and Get Details](https://perceptlayer.com/blog/google-patents-api-search-details)
    
    Search Google Patents by keyword, inventor or assignee with one GET call, then fetch full details by patent\_id. Both endpoints cost 1 credit per call.
    
3.  [Web data by industry](https://perceptlayer.com/blog/category/web-data) 6 Oct 2026 Perceptlayer 7 min read
    
    ## [Salary Data from Job Sites: Reading Ranges Without Bias](https://perceptlayer.com/blog/salary-data-job-sites-bias)
    
    Glassdoor's salaries endpoint returns p10 to p90 total pay, a currency and a national count. How to read those ranges and the biases of self-reported pay.