What Does a Site Crawl Reveal About Your Brand Voice?
You fed an AI writer your "friendly but authoritative" tone notes, and it still returned robotic mush. Then came the tone drift, and then the prompt-engineering homework.
That frustration is fair. The voice you keep trying to describe already exists in the pages you have published, and learning what a site crawl reveals about brand voice is the fastest way to recover it.
Picture the setup. You drop a domain into a setup wizard, the crawl finds the XML sitemap, inspects published pages, and detects the CMS. That is how an AI SEO content engine starts, and it beats any blank-page workshop.
This guide moves from crawl to signals to rules to verification. Watch for one twist along the way: one of the strongest voice signals isn't on any page at all.
Key Takeaway: A crawl turns published pages into evidence, including vocabulary, rhythm, point of view, CTA style, topic focus, keyword footprint, and linking habits. Each signal then becomes a written rule.
What Does a Site Crawl Reveal About Your Brand Voice?
A site crawl reads your published pages through the sitemap and surfaces the voice you actually ship. It exposes recurring terms, sentence rhythm, point of view, heading and CTA style, topic focus, keyword footprint, and anchor phrasing.
A site crawl is a process that follows your XML sitemap to inspect each published page it lists. Brand voice is the consistent personality your writing shows across many pages, not one polished homepage.
Think of it like reading someone's sent emails instead of asking how they think they write. The self-description flatters, but the outbox is accurate.
Popular brand-voice videos teach you to build a voice from scratch with adjectives and archetypes, and they never audit the content you already have. That gap matters because the workshop voice is a wish, while the crawled voice is the record of how your team writes under deadline. A crawl compares many real choices, such as which words repeat, how long sentences run, whether you write as "we" or "you," and how you phrase calls to action, so the resulting rules describe habits writers can reproduce.
Before a crawl gives a useful read, you need three things:
- A reachable XML sitemap with a sitemap-listed inventory of live pages.
- Enough published pages to show patterns rather than one-off quirks.
- An exclusion plan for pages that would distort the read.
Once extracted, voice rules can live in an editable Markdown context file, so you adjust a rule without rewriting a prompt.
How to Turn a Site Crawl Into Brand Voice Rules Step by Step
Crawl the sitemap, exclude distorting pages, read structural signals, read prose signals, and convert each signal into a written rule. The output is a rule set any writer or AI model can follow and you can verify.
Prerequisites: a live XML sitemap, a crawler or setup wizard that reads it, and a text editor for a Markdown rules file. Budget roughly two hours for a first pass.
- Crawl the sitemap (about 10 minutes). Paste your domain into the setup wizard and let it locate the XML sitemap. Success looks like a full list of live URLs and a detected CMS. But there's a catch.
- Exclude pages that corrupt extraction (about 20 minutes). Remove these before reading anything:
- Legal pages such as privacy policies and terms
- Legacy pages written in an abandoned voice
- Guest posts by outside authors
- Boilerplate tool pages with templated copy
- Read the structural signals (about 30 minutes). Map the sitemap hierarchy to see niche focus and topic clusters. Scan titles and slugs for the keyword footprint, and review anchor text for how the brand names its concepts.
- Read the prose signals (about 40 minutes). Note the lexicon, including words the site never uses. That absence is the signal hiding off every page. Then record rhythm and reading level, point of view, and heading and CTA conventions.
- Convert each signal into a rule (about 20 minutes). Fill in the table below. Success looks like one testable rule per row.
Why does exclusion matter so much? BloGoose's own sitemap mixes SEO guides with dozens of utility tools, such as calculators and converters. Those tool pages use short, templated, instructional copy that looks nothing like the long-form guides. If a crawl reads them alongside the articles, the extracted voice skews terse and generic, and rules built from that sample would push every new article toward boilerplate.
| Signal | Evidence | Rule |
|---|---|---|
| Lexicon | Mechanics named, fog absent | DO "Explain exact mechanics like crawling sitemaps, configuring Google Indexing API service accounts." DON'T "Use inflated startup jargon or marketing fog." |
| Rhythm | Short, direct paragraphs | Keep paragraphs to 1-3 sentences. |
| Point of view | Second person throughout | Address the reader as "you." |
| Heading and CTA style | Action-led headings and CTAs | Open headings and CTAs with a verb. |
| Niche focus | Guides cluster around SEO | Write only inside established clusters. |
| Keyword footprint | Titles follow "[Actionable Title] · BloGoose" | Place primary terms near the front of titles. |
| Anchor phrasing | Anchors repeat exact concept names | Link with the page's own concept name. |
If manual extraction sounds like the multi-tab stack you want to escape, BloGoose runs this crawl-to-rules workflow in its setup wizard. It stores the rules in editable context files and then drafts against them.
Rules on paper are only half the job, so next comes checking whether they actually worked.
How to Check the Voice Rules Actually Worked
Your voice rules worked if lexicon, point of view, and banned words stay the same across every page type while tone shifts to fit each page. Voice is the stable layer, and tone is the layer that flexes.
| Page type | Voice (should hold steady) | Tone (should flex) | Best for |
|---|---|---|---|
| Blog posts | Same lexicon, POV, banned words | Conversational, opinionated | Content leads reviewing cluster drafts |
| Pricing pages | Same lexicon, POV, banned words | Plainspoken, direct | Founders checking conversion copy |
| Docs | Same lexicon, POV, banned words | Instructive, step-based | Solo operators writing help content |
Four checks for every new draft
- Banned terms absent: No word from your exclusion list appears.
- POV consistent: "We" and "you" are used the same way as on your live pages.
- CTA phrasing matches: Calls to action follow the house pattern.
- Title follows the pattern: Headline structure matches your top pages.
A manual SEO content audit still has real strengths. An experienced editor catches awkward rhythm and off-brand jokes that a rule checklist can miss.
What a crawl can't tell you
A crawl captures what you shipped, not what you intend. It can't hear your sales calls, read your social posts, or judge whether a page was any good. Humans still edit the rules.
Site Crawl and Brand Voice FAQ
These answers cover the questions people raise most after their first crawl: timing, small sites, and long-term consistency.
How often should I re-crawl my site for brand voice?
Re-crawl when your published inventory or positioning changes meaningfully. A product launch, a rebrand, or a batch of new pages can each shift your patterns. No fixed interval fits every site. Tie re-crawls to real changes so your rules always reflect what you actually publish.
Does a site crawl work on a small site with few pages?
Yes, but fewer pages produce weaker patterns. A site with a handful of pages gives the crawl less evidence for lexicon and POV. Treat the output as a draft, and rely more heavily on human-edited rules to fill gaps the crawl can't confirm.
Can AI keep a brand voice consistent over time?
Yes, when the rules live outside the chat window. BloGoose stores them as Markdown files such as style-guide.md, target-keywords.md, and internal-links-map.md, and reads them before every draft. That same structure supports AEO content optimization, so each article stays both on-voice and citation-ready.
Make Your Crawl the Source of Truth for Every Draft
Your brand voice is already written down in the pages you've shipped. A crawl turns that evidence into rules that every draft can follow.
The workflow works only if you treat crawl output as evidence. Exclude noise pages, write rules instead of adjectives, check voice against tone across page types, and re-run the crawl when your site changes.
BloGoose builds that workflow for you. One crawl generates eight editable Markdown context files, including style-guide.md, target-keywords.md, and internal-links-map.md. Every draft reads them first. You stay in control because changing a rule means editing a file, not rewriting a prompt.
Next step: Run a crawl on your domain with BloGoose, then edit the generated voice rules until they sound like you. To compare plans, see BloGoose pricing.
Once your rules live in files, every article you publish strengthens the same voice.