Editorial Guidelines

last_updated: 2026-05-03

This page documents how VeganStove decides what to publish, how recipes are tested, where data comes from, and what we will and will not do. If anything on the site does not meet the standard below, write to us. We update what is wrong.

Last updated: 2026-05-03. Maintained by: Sage Bendly, editor and lead test cook.

1. Mission

Plant-based recipes that work. Each one is cooked at least three times before publishing. The method is the headline. Sources are cited. Mistakes are annotated. Versions are stamped.

VeganStove is not a lifestyle blog, not a wellness brand, and not a publication that ships recipes from a press release. It is a working test kitchen with notes.

2. Who writes here

One byline: Sage Bendly. Six years inside Pacific Northwest food media before this site existed. All recipe development, testing, photography, and writing is hers unless a guest contributor is named at the top of the post (which has not happened yet, but the policy is in place).

If a recipe came from a reader, a cookbook, or a chef, the source is named at the top of the post. Adapted recipes credit the original author, the publication, and the year. We do not republish a third party’s recipe text. We test, rewrite in our voice, document the changes, and credit.

3. Topic selection

How a recipe gets into the queue:

Reader demand. If three or more readers ask for the same dish in email, it moves to the top of the list.
Search demand. Keyword research for the niche (vegan recipes, USA market). High-volume topics with low competitor quality get priority.
Seasonal fit. Lighter air-fryer mains in summer, hearty pastas in winter, brunch projects when daylight is short.
Pillar coverage. Each of the six categories (bowls, breakfast, burgers, pasta, air-fryer, desserts) is filled out methodically. Holes in coverage become drafts.

What we say no to:

“Top 10 vegan products” roundups assembled from PR pitches.
Recipes that exist only to capture a trend with zero personal interest.
Cuisine from cultures we do not understand well enough to represent without distortion. We refer readers to creators inside that culture instead.

4. The three-test protocol

Nothing publishes without three real test cooks. The protocol:

Test 1: baseline. Cook the recipe as drafted. Time everything. Photograph the failure points. Note what felt off.
Test 2: revise. Apply the lessons from test 1 (often: too much sauce, not enough rest time, wrong pan size). Cook again with changes documented.
Test 3: lock. Cook the revised recipe under the conditions a reader would have (no special tools beyond what is in the equipment list). If this run produces the result we want, the recipe is locked. If not, we go back to test 2.

Variations on the same base (different vegetable, different sauce) are not separate tests. They are extensions. Real different conditions (different brand of beans, different stove, different day) count.

For tools and gear posts, we use the product for at least three weeks of normal cooking before forming an opinion. For ingredient sourcing posts (e.g. “best canned chickpeas”), we test at least three brands side by side, on the same day, blind where possible.

5. Source citation

Every claim that goes beyond “I cooked this” cites a source.

Cooking-time reference values for the calculator come from the consensus of three named industry sources (USA Dry Pea & Lentil Council, Rancho Gordo, Bean Institute). The full method is on the cooking time methodology page.
Nutrition figures come from USDA FoodData Central. We do not invent macros.
Food-safety claims (cooling times, storage windows, internal temperatures) come from USDA, FDA, or peer-reviewed food-science sources. Cited inline.
Historical or cultural claims about a dish’s origin require at least two independent sources. We will not pass off a marketing legend as history.

Where a claim is our own opinion or experience (“I find this works better than that”), it is flagged as such. Personal opinion is not dressed up as authority.

6. Photography

All photographs of finished recipes are taken in our kitchen, of the food we actually cooked. No stock photos of the finished dish. No AI-generated images of food.

Step-by-step photos may be cropped, color-corrected, or tonally adjusted. They are not composited. The bowl in the hero shot is the bowl that just came off the stove.

For ingredient images and editorial-style photographs (lifestyle, hands-in-the-frame, props), we may use AI assistance for visualization mockups while planning a shot, but the final published image is photographic unless the post explicitly says otherwise.

7. AI and writing

This is the question every recipe site is asked, so here is the explicit position.

What AI does on VeganStove:

Spell-check and grammar suggestions.
Drafting outlines that the editor then rewrites in her own voice.
Brainstorming alternative section orders or headings.
Checking for accidental repetition across articles.

What AI does not do on VeganStove:

Decide what to cook.
Invent timing data.
Write the experience of standing over a hot pan.
Generate images of finished food that we then publish as if we cooked it.
Write the “what I learned on test 2” notes.

The editorial voice is one human’s voice. Where a passage was AI-drafted and then heavily rewritten, the rewritten version is what ships. AI is a spell-checker, not a co-author.

8. Versioning and corrections

Every recipe carries a version number, starting at v1.0.

v1.0: the recipe at original publish.
v1.1, v1.2, …: small corrections (typo, swapped step order, clarified measurement). The change is logged at the bottom of the recipe with the date.
v2.0, v3.0, …: a meaningful change (different ratio, different cooking time, different recommended pan). The previous version is preserved in the version log.

When a reader catches an error, we fix it within 48 hours of confirming the error, bump the version, log the change, and credit the reader unless they ask not to be named.

If a recipe is so flawed that it should not have shipped, we say so in the version log and consider unpublishing. We do not silently delete posts to hide bad work.

9. Comments and reader contributions

Comments are moderated. We aim to publish them within 24 hours on weekdays.

Substantive corrections are addressed in the comment thread and, if right, applied to the recipe with a version bump.
Substitution questions get answered if the substitution is something we have actually tested. If not, we say “I have not tested that, my best guess is…” and label the guess.
Comments are removed if they are spam, hostile, off-topic, or attempts to insert affiliate links. The standards are spelled out in the terms of use.

10. Sponsored content and affiliate links

Standing rules:

We only link to tools and ingredients we have personally used.
Commission does not influence whether something gets a positive review.
Sponsored posts are labeled at the top of the article.
Brands cannot pay to be added to a “best of” roundup.

The full position is on the affiliate disclosure page.

11. Conflicts of interest

If the test cook has a personal or financial relationship with a brand, ingredient, or tool that appears in a post, that relationship is disclosed in the post itself, not buried in a sitewide page.

The current relationships:

Amazon Associates affiliate (US).
Newsletter platform: a paid Mailchimp (Intuit Inc.) account.
Comment spam filter: Akismet (Automattic Inc.).
Web hosting: a paid commercial host.

None of the above currently has editorial influence over what gets published.

12. What we get wrong

Probably more than we have caught yet. Some failure modes we know about and try to mitigate:

Sea-level bias. The test kitchen is at near-sea-level altitude in Portland. Cooks at altitude (Denver, Mexico City) will see longer bean times. We try to flag this in recipes where altitude matters.
Single-stove bias. One range, one oven, one air fryer. Your equipment may behave differently. Where we know a model behaves differently, we say so.
Single palate bias. “Salt to taste” reflects one person’s salt tolerance. We give a starting amount, not a final amount.

13. How to challenge a piece of content

Send a message through the contact page with the URL, the specific claim or step you think is wrong, and (if you can) the source you would point to instead. Substantive corrections are reviewed within 48 hours on weekdays. Persistent, well-sourced corrections shape the recipe.

14. Updates to these guidelines

We update this page when we change how we work. The “last updated” date at the top reflects the latest revision. Substantive changes are noted in the change log below.

Change log format. Entries are listed in reverse-chronological order (most recent first), each with an ISO date, a short summary of what changed, and the author of the change.

Change log

2026-05-03 · Sage Bendly · First public version of these editorial guidelines.