What Patent Examiners Look for in a Provisional Application

Key Definitions

35 U.S.C. §112(a)

The statutory requirement that a patent specification contain a written description of the invention, enable a person skilled in the art to make and use it, and disclose the best mode of carrying out the invention. Provisionals must satisfy the written description and enablement requirements (but not the best mode requirement for priority purposes).

Enablement

The requirement that your description is detailed enough for a PHOSITA (person having ordinary skill in the art) to make and use the invention without undue experimentation.

Written Description

The requirement that your specification demonstrates you were in possession of the claimed invention at the time of filing. This is distinct from enablement: enablement asks “can someone build this from what you wrote?” while written description asks “did you actually have this invention when you filed?”

PHOSITA

Person Having Ordinary Skill In The Art. The hypothetical reader your specification must be written for. Not a complete beginner, but not a world-class expert either. A competent practitioner in your technology field.

New Matter

Any material added to an application after its filing date that was not part of the original disclosure. New matter is prohibited in both provisional and non-provisional applications.

Priority Date Benefit

The right of a non-provisional application to claim the filing date of an earlier provisional for claims that the provisional adequately supports under §112(a).

Introduction

There is a common misconception that because provisional patent applications are not examined, they do not need to meet any particular standard. This is wrong, and it leads to provisionals that fail at the one job they are supposed to do: support the claims in your non-provisional application.

Here is the reality. A provisional application is never reviewed by an examiner during its 12-month pendency. No one at the USPTO reads it, evaluates it, or issues Office Actions on it. But the moment you file a non-provisional application claiming the benefit of that provisional, the examiner reviewing your non-provisional will look at the provisional's disclosure to determine whether it supports the claims you are now asserting.

If the provisional's disclosure does not adequately describe the claimed invention, those claims do not receive the provisional's priority date. They get only the non-provisional's filing date. And if any prior art emerged between the two filing dates, your claims may be invalid because of it.

This is not a theoretical risk. In New Railhead Manufacturing v. Vermeer Manufacturing (Fed. Cir. 2002), the Federal Circuit held that a provisional application failed to adequately describe key claim limitations, stripping the patent of its provisional priority date and rendering it invalid due to a commercial sale that occurred between the provisional and non-provisional filing dates.

This guide explains what the §112(a) requirements actually mean for your provisional, how examiners evaluate priority claims during non-provisional examination, and how to structure your provisional so that every claim you care about is fully supported.

The Legal Standard: §112(a) Applied to Provisionals

Under 35 U.S.C. §119(e), a non-provisional application can claim the benefit of a provisional's filing date only if the provisional discloses the claimed invention in the manner provided by §112(a), except for the best mode requirement. This means two specific requirements must be met.

The written description requirement

The written description must demonstrate that the inventor was in possession of the claimed invention at the time the provisional was filed. This does not mean you need to use the exact same words as your claims. The disclosure can support claims using different terminology, as long as a person skilled in the art would recognize from the disclosure that you had the invention.

What triggers a written description failure? Typically, it happens when the non-provisional claims are broader than what the provisional actually described. If your provisional describes a specific implementation (one algorithm, one architecture, one data structure) and your non-provisional claims cover a genus (any algorithm, any architecture, any data structure), the examiner may find that the provisional did not provide written description support for the broader claims.

For software inventors, this is a particularly common trap. A provisional that describes a specific neural network architecture does not necessarily provide written description support for a claim covering “a machine learning model” generically. The provisional needs to either describe multiple model types or use language that a PHOSITA would understand as covering the broader category.

The enablement requirement

The specification must enable a person skilled in the art to make and use the invention without undue experimentation. For software inventions, this means your description must be detailed enough that a competent developer in your field could build a working implementation from what you wrote.

Enablement failures in software provisionals usually stem from describing what a system does without describing how it does it. “The system uses machine learning to predict user behavior” is not enabled. “The system applies a gradient-boosted decision tree model to a feature vector comprising user session duration, page visit frequency, and interaction recency scores, generating a churn probability score for each user” is enabled (assuming a PHOSITA in the ML field could implement that from the description).

What is NOT required in a provisional

Formal patent claims are not required. An oath or declaration is not required. An information disclosure statement is not required. Best mode disclosure is not required for priority purposes (though it is required in the non-provisional itself). However, the absence of a claims requirement does not mean claims are unimportant. Including claims in your provisional helps ensure that the disclosure actually supports the boundaries you intend to claim later. (See our guide on how to write patent claims for claim drafting fundamentals.)

How Examiners Evaluate Provisional Support During Non-Provisional Examination

When an examiner reviews your non-provisional application, they follow a specific process for evaluating whether your claims are entitled to the provisional's filing date.

The claim-by-claim analysis

The examiner evaluates priority on a claim-by-claim basis, not on the application as a whole. Some claims in your non-provisional may be entitled to the provisional's priority date while others are not. A claim is entitled to the provisional's filing date only if the provisional provides adequate §112(a) support for that specific claim. This means that if your non-provisional has 20 claims and 15 of them are fully supported by the provisional, those 15 get the provisional's priority date. The remaining 5, if they cover subject matter not adequately disclosed in the provisional, get only the non-provisional's filing date.

What examiners specifically look for

Element-by-element support. The examiner checks whether every limitation in a claim has corresponding disclosure in the provisional. If Claim 1 recites “a processor configured to apply a weighted priority score to each item in a queue based on a real-time latency measurement,” the provisional must describe that specific operation. Not in identical words, but in substance.

Scope alignment. If the claim is broader than the provisional's disclosure, the examiner evaluates whether the provisional's disclosure would convey to a PHOSITA that the inventor had possession of the broader invention. Describing one species (a specific neural network) may or may not support a genus claim (any machine learning model), depending on the context and the field.

Consistent terminology. While exact word matching is not required, inconsistent terminology between the provisional and the non-provisional can create ambiguity. If your provisional calls a component a “pattern recognition engine” and your claims call it a “behavior analysis module,” the examiner must determine whether a PHOSITA would understand these as the same thing.

Drawing support. Drawings filed with the provisional are part of the disclosure. If a claim limitation is shown in a drawing but not described in the text, the drawing may still provide support. Conversely, drawings that were not included in the provisional cannot be added later without constituting new matter.

What happens when support is insufficient

If the examiner determines that a claim is not supported by the provisional, the claim's effective filing date becomes the non-provisional's filing date. The examiner then evaluates novelty and non-obviousness against prior art available as of that later date. If prior art exists between the provisional and non-provisional filing dates that anticipates or renders the claim obvious, the claim is rejected. The examiner may also issue a §112(a) written description rejection if the non-provisional's own specification does not adequately support the claims, independent of the priority question.

Structuring Your Provisional for Maximum Support

Given how examiners evaluate provisional support, here is how to structure your provisional to give every claim the best chance of receiving the provisional's priority date.

Write the detailed description as if it were a non-provisional

The single most effective strategy is to treat your provisional's detailed description with the same rigor you would apply to a non-provisional specification. Describe the system architecture, the data flows, the algorithms, the processing steps, the data structures, and the component interactions in enough detail that a PHOSITA could build it.

For software inventions, this means describing:

The system architecture. Identify each major component (client, server, database, API gateway, processing engine, ML model) and describe how they connect and communicate. Use reference numerals if possible. Specify protocols, data formats, and communication patterns.

The processing steps. Walk through what the system does step by step, from input to output. For each step, describe the input data, the operation performed, and the output data. If a step involves an algorithm, describe how the algorithm works, not just what result it produces.

The data structures. Define the specific data objects your system creates, transforms, and stores. Describe their fields, their relationships, and how they change as they move through the system.

Alternative embodiments. This is where most provisionals fall short. Describe variations: different algorithms that could perform the same function, different architectures that could host the same logic, different data structures that could represent the same information. Each alternative you describe in the provisional expands the scope of claims your non-provisional can support.

The Patent Geyser Module 5 workflow generates provisional specifications with this structure built in. The system produces a title, background, summary, detailed description (including architecture, data structures, operational flow, and alternative embodiments), ramifications and scope, abstract, and claims. (See our guide on how to draft a provisional patent application for the complete process.)

Include claims even though they are not required

Drafting claims at the provisional stage serves as a structural check on your disclosure. If you write Claim 1 and then look at your specification and cannot find support for one of its limitations, you know exactly where the gap is, and you can fix it before filing. Claims in the provisional also create a record of what you intended to protect at the time of filing.

Include all drawings you may need later

Under 35 U.S.C. §113, a drawing necessary to understand the invention cannot be introduced after the filing date because of the prohibition against new matter. If your non-provisional needs a system architecture diagram, a flowchart, a data flow diagram, or a UI wireframe to support a claim, that drawing must be in the provisional. For software inventions, include at minimum: a system architecture diagram showing all major components and their connections, flowcharts for each novel process or method, data flow diagrams showing how information moves through the system, and any other visual that illustrates an aspect of the invention you may want to claim.

Define your terms

If your invention uses specialized terminology, define it explicitly in the provisional. “As used herein, ‘event stream’ refers to a temporally ordered sequence of user interaction records transmitted in real time from a client application to a server via a persistent WebSocket connection.” This definition prevents ambiguity during examination and ensures the examiner interprets your terms the way you intend.

Do not characterize anything as “routine” or “conventional”

During examination, if the examiner finds language in your specification describing an element as “routine,” “conventional,” or “well-understood,” that language can be used against you in §101 eligibility analysis (as evidence that the element does not constitute an inventive concept) and in §103 obviousness analysis (as an admission that the element was known). Describe what your invention does and how it works without editorializing about whether any particular component is novel or routine. Let the examiner draw those conclusions independently.

The Five Most Common Provisional Failures

These are the patterns that most frequently result in non-provisional claims losing their provisional priority date.

1. The pitch deck provisional. An inventor files a two-page description that reads like a business plan: the problem, the market, the solution described at a high level, and a competitive analysis. This provides virtually no §112(a) support for any technical claim. The solution: write a technical specification, not a pitch deck.

2. The single-embodiment provisional. The provisional describes one specific implementation in detail but includes no alternatives. The non-provisional then claims a broader genus. The examiner finds that the provisional only supports the specific species, not the genus. The solution: include alternative embodiments for every key component and method.

3. The “what it does” provisional. The provisional describes the invention's functionality and user-facing behavior without explaining the underlying technical mechanism. For software, this means describing the UI without describing the backend. The examiner finds insufficient enablement. The solution: describe how the system works, not just what it does.

4. The code dump provisional. An inventor attaches source code as the entire provisional disclosure. While source code can contribute to enablement, it does not substitute for a written description. Examiners and courts evaluate written description based on what a PHOSITA would understand from the specification, and raw code without explanatory text may not convey the inventive concept. The solution: write a specification that explains the invention in technical prose, supported by code excerpts where they add clarity.

5. The evolving invention provisional. The inventor files a provisional early, continues developing the product, and files a non-provisional 11 months later with claims covering features that did not exist when the provisional was filed. Those features get the non-provisional's filing date, not the provisional's. The solution: file additional provisionals as the invention evolves, and claim priority to all of them in the non-provisional.

How This Connects to the Non-Provisional Filing

The provisional and non-provisional are not separate documents. They are two stages of a single patent strategy. The provisional creates the foundation; the non-provisional builds on it.

During the 12-month provisional window, you should be doing several things that directly improve the strength of your non-provisional:

Deepening the technical description. Use the time to expand your provisional's disclosure, fill gaps, and add detail. This expanded content goes into the non-provisional specification.

Conducting prior art research. Understanding the competitive landscape helps you draft non-provisional claims that are both novel and properly supported by the provisional.

Refining claims. The claims you file in the non-provisional should be shaped by both the provisional's disclosure and the prior art landscape. Each claim must be supportable by the provisional to receive the earlier filing date.

Having a practitioner review. A registered patent practitioner can identify support gaps between your provisional and your intended claims before you file the non-provisional. This is far less expensive than discovering the gap during examination. If you need help finding a practitioner, the PatentFit Directory scores practitioners based on their actual USPTO filing history across 112 CPC technology areas.

For a complete comparison of provisionals and non-provisionals, see our guide on provisional vs. non-provisional patent applications.

Conclusion

A provisional patent application is not examined, but it is not a throwaway document. It is the foundation on which your entire patent rests. Every claim in your non-provisional that receives the provisional's priority date does so because the provisional provided adequate written description and enablement under §112(a). Every claim that does not loses up to 12 months of priority protection.

The standard is clear: a person skilled in your technology field should be able to understand and make the invention from your provisional's disclosure. For software, SaaS, and blockchain inventors, this means describing the system architecture, the processing steps, the algorithms, the data structures, and the alternative embodiments in enough technical detail that a competent developer could build it.

Treat your provisional with the same care you would give a non-provisional specification. Include drawings. Include claims. Define your terms. Describe alternatives. And before you file, have a registered patent practitioner review it. The effort you invest in the provisional pays off at every subsequent stage of the patent process.

Frequently Asked Questions

Examiner Standards

Do patent examiners actually read provisional applications?

Not during the provisional's 12-month pendency. Provisionals are not examined on their merits. However, when you file a non-provisional application claiming the benefit of the provisional's filing date, the examiner reviewing the non-provisional will evaluate whether the provisional provides adequate §112(a) support for each claim. At that point, the provisional's disclosure is scrutinized claim by claim.

What happens if my provisional does not support all of my non-provisional claims?

Claims that are not adequately supported by the provisional receive the non-provisional's filing date instead of the provisional's. If prior art exists between the two filing dates that anticipates or renders those claims obvious, the claims may be rejected. Claims that are supported by the provisional keep the earlier filing date.

Does the provisional need to use the same terminology as the non-provisional claims?

No. The disclosure does not need to use identical words. The standard is whether a PHOSITA would recognize from the provisional's disclosure that the inventor had possession of the invention as claimed. However, consistent terminology reduces ambiguity and makes the support argument cleaner during examination.

Practical Drafting

How detailed does my provisional need to be for a software invention?

Detailed enough that a competent developer in your field could understand and build the system from your description. This means describing the system architecture (components and connections), the processing steps (inputs, operations, outputs), the data structures (fields, relationships, transformations), and the algorithms (how they work, not just what results they produce). If you find yourself writing “the system performs X” without explaining how it performs X, the description is not detailed enough.

Should I include source code in my provisional?

Source code can supplement your disclosure but should not replace a written specification. Raw code without explanatory prose may not convey the inventive concept to the examiner. The best approach is to write a complete specification in technical prose and include code excerpts where they clarify specific algorithms or data structures. The specification must stand on its own as a readable description of the invention.