The Claude Skills I Use to Run My Own Tutoring Company's SEO (No Agency)

What tools does a tutoring-business owner actually need to run their own SEO?

At minimum: Google Search Console (free, non-negotiable), Bing Webmaster Tools (free, and it’s what feeds Microsoft Copilot’s answers — most owners skip this entirely), a free Domain Rating checker, and a way to see what’s actually ranking for your target keywords. Everything past that — automated technical audits, structured data generation, backlink scoring — is where I’ve built a repeatable system on top of those same data sources, because doing it by hand every month doesn’t scale past a handful of pages.

Why I don’t pay an agency

Most tutoring companies either pay a generic SEO agency a monthly retainer for work that’s barely tailored to a local service business, or do nothing and hope the Google Business Profile carries them. I went a different way: I run my own SEO for both EquateIt and ClassQuill using a set of tools I built into Claude Code, wired to a handful of real, paid data sources. This isn’t a claim that everyone should do this exact setup — it’s a genuine look at what’s possible if you’re willing to build it, and honestly, at what you can skip and still get most of the value.

This is the SEO/marketing half of the stack. The delivery-side half — how I run actual tutoring sessions with the same kind of setup — is a separate piece I’m still writing up.

The stack, piece by piece

1. A technical SEO audit that actually checks the code, not just the page

Most technical SEO advice tells you to “check your meta tags.” Mine reads the actual repo — checking for missing meta descriptions, duplicate titles, broken internal links, redirect chains, and whether pages are even indexable, before a single page goes live. For a site you’re actively building (not just auditing after the fact), catching this pre-launch is worth more than any post-launch crawl tool.

2. Structured data that’s actually grounded in your real content

Schema markup (the structured data that helps Google and AI answers understand what a page actually is — a review, an FAQ, a local business) is one of the most skipped technical SEO tasks, because it’s fiddly to write by hand and easy to get subtly wrong. I generate it directly from the real page content and config, so it doesn’t drift from what the page actually says.

3. Real keyword and search-volume data — not guesses

This is the one that changes the most. For years, “keyword research” for a small business meant guessing or trusting whatever a blog post said sounded plausible. I pull real search volume, difficulty, and intent data through OpenSEO (backed by DataForSEO) before writing anything — and it’s routinely humbling: most hyper-specific “tutoring business” long-tail phrases return near-zero real search volume. Knowing that before you write 2,000 words chasing a phrase nobody searches is worth more than the content itself.

Worth being clear on the difference between this and Search Console, since the two get confused: Search Console shows what’s already true about your own site — the actual queries real people typed to find you. OpenSEO/DataForSEO shows the market-wide estimate for any keyword, whether you rank for it yet or not, and it’s not US-only — volume is pulled per country, so an AU-specific business gets AU-specific numbers, not an American default.

The real surprise wasn’t a single keyword, it was realising what stage ClassQuill was actually at. At DR-5 with a handful of referring domains and close to zero organic clicks, it’s squarely a new site — no authority to compete for the obvious head terms, whatever the volume number said. Once that was clear, the useful keywords weren’t the big ones, they were the undertargeted, bottom-funnel ones a bigger site hasn’t bothered defending: a rival ranks in the low teens for its own product’s “login” query off nothing more than a pricing page, which is real, high-intent traffic sitting there because nobody defended it. The other genuine whitespace was country-specific — AU compliance terms (WWCC renewal cost, GST thresholds for tutoring income) that US and UK competitors structurally can’t rank for, because the facts themselves are Australian. That reframed the whole approach: chase what’s undertargeted and geographically specific before chasing volume.

4. A backlink opportunity checklist — not just a directory list

I built a whole separate discipline around this — Domain Rating, real traffic, follow status, cost, and a plain pay/skip/negotiate verdict, run against every association, directory or guest-post pitch before I spend anything. That’s a big enough topic to get its own piece: how I decide whether a backlink is worth paying for →.

5. Google Search Console — pulling real performance data, not screenshots

Rather than logging into Search Console and eyeballing a chart once a month, I pull the actual query, page and position data programmatically — which makes it possible to actually spot “striking distance” keywords (ranking position 11–20, one push away from page one) instead of scrolling a dashboard and hoping something jumps out.

6. Competitor backlink and ranking checks

Before pitching a guest post or a directory, I check who else — specifically, which of my direct competitors — is already listed there. If three competitors are all on the same association’s member directory and I’m not, that’s not a nice-to-have, it’s a gap actively costing me trust-by-association.

7. Bing Webmaster Tools — the one everyone skips

Almost every small business I’ve looked at optimises for Google and completely ignores Bing. That’s a mistake for two separate reasons: Bing Webmaster Tools gives you a second, free source of real backlink and indexing data most competitors never bother claiming — and Bing is the index behind Microsoft Copilot’s answers, which is a genuine and growing source of “AI answer” visibility that Google-only optimisation misses entirely.

What you can actually replicate without any of this

This isn’t a “you need my exact setup” post. If you’re not going to build any of the above, do these things by hand and you’ll capture most of the value. If you were starting from genuinely zero, this is the actual order I’d do it in:

• [ ] Google Business Profile first. This is the single most important one for a local service business, ahead of anything else on this list — and don’t just set it up, post to it so it visibly shows you’re active.
• [ ] Google Search Console — free, and the second-most important tool on this list. Set it up if you haven’t.
• [ ] Link your socials — Instagram, YouTube and TikTok connected and active, not dormant accounts sitting there for the sake of it.
• [ ] Bing Webmaster Tools — free, five minutes, and almost nobody in tutoring does it.
• [ ] A free Domain Rating checker (Ahrefs has one) — before you pay for any directory or association listing, check its real number.
• [ ] Google’s own “site:” search — type site:yourdomain.com into Google to see what’s actually indexed. If pages are missing, that’s your first fix, before anything fancier.

Honestly, it took about a month to get into a real working flow — wiring up the standard base (backlinks, Search Console, Bing, the usual directory listings like Crunchbase and SourceForge) alongside actually building the content-with-SEO habit, running site audits, and learning to spot a genuinely good backlink from a waste of money. This isn’t a weekend project, and being upfront about that matters more than pretending it was instant.

The honest limitation

None of this replaces judgement. A tool can tell you a keyword has volume, a directory has a real Domain Rating, or a page is missing a meta description — it can’t tell you whether your actual pitch to a guest-post editor is any good, or whether the story you’re telling about your tutoring company is genuinely differentiated. That part’s still entirely on you, every time.

I haven’t had the tooling itself flag something that turned out to be wrong so much as I’ve had to keep re-learning which tool answers which question — Search Console versus OpenSEO being the clearest example (see above). The judgement calls the tooling doesn’t make for you are less about catching a lying number and more about knowing which real number you’re actually looking at.

For what it’s worth, the outcome has been real: one month of actually running this properly took a site from around 500 impressions and 2 clicks a month to roughly 30,000 impressions and 120 clicks. That’s not a guarantee anyone else gets the same curve, but it’s the honest number behind why I keep doing this myself instead of handing it to an agency.