Editorial Policy

Last updated: September 30, 2026

The one-sentence version

Every workflow, verdict, and failure mode we publish was observed by us, running real automations, with our own accounts and usually our own money.

How we test

When we compare tools, we don't read documentation and paraphrase it. We rebuild the same workflow in each tool, with the same trigger and the same end result, and we run them side by side for as long as the comparison needs to be honest — typically two to four weeks. We note:

• How long setup actually took us, including the dead ends;
• What broke, at what traffic volume, and what the error looked like;
• What each configuration costs at 3–4 realistic volume levels (e.g. 500, 2,000, 10,000 runs/month), because pricing pages hide more than they reveal;
• Which failure modes are recoverable in the UI versus which ones require you to rebuild from scratch.

What we pay for

We buy our own subscriptions whenever a tool has no workable free tier. Where a vendor offers us press access or a temporary upgrade, we disclose that in the article. We don't accept "free review access" in exchange for coverage — if a vendor's terms say a free account obligates us to publish, we decline the account.

Independence

BuiltAG earns money from display advertising and affiliate commissions (see theaffiliate disclosure). Neither revenue stream gets a vote in our verdicts. Concretely: our most frequently recommended option for cost-conscious readers — self-hosted n8n — pays us nothing. If a vendor asks us to soften a negative finding, the answer is no, and we keep the email.

How we use AI

We use AI writing tools the way a good operator uses a power drill: for the boring, repetitive parts, under supervision. Drafts, outlines, and rewording lumps of text — fine. Facts, verdicts, prices, and anything we claim to have tested — always human-verified against what we actually ran. If we can't verify a claim against a real workflow we built, the claim doesn't publish. This is both a quality policy and a self-interested one: search engines increasingly demote unverified AI content, and readers smell it instantly.

Corrections and updates

We get things wrong sometimes. When we do:

• Factual errors are fixed inline as soon as confirmed, with an update note at the top or bottom of the article saying what changed and when. We don't silently edit verdicts.
• Pricing and feature changes — the fastest-rotating facts in this niche — are re-verified quarterly for our most-read comparison articles, and the article's "last verified" date is updated when we do.
• Verdict changes happen when the tools change or when we learn something new. When a recommendation reverses, the article says so explicitly ("we used to recommend X for this; we no longer do, and here's why") rather than quietly rewriting history.

What we don't publish

• Sponsored posts, paid placements, or "advertorials" not labeled as such;
• Reviews of tools we haven't run;
• Pricing claims without a link you can check yourself;
• Content created by outside writers we haven't worked with directly.

Found a problem?

Email support@builtag.com with the article link and what you think is wrong. Corrections get the fastest replies in our inbox.