Most teams pay for more SEO tools than they actually use, and several of them measure the wrong things. An audit takes 30 minutes. You keep the tools that show verified clicks, conversions, and citations in AI answer engines. You cancel the ones that only produce polished visibility scores.
An SEO tool is software that collects, analyzes, or acts on data about how your website appears in search engines. Today, that means both Google's link list and AI answers in ChatGPT, Gemini, and Perplexity.
The issue is concrete. A traditional tool can report a high visibility index while your brand is completely absent from AI summaries. That is SEO theater. Here is how you close the curtain.
Key takeaways
- Google Search Console is free and provides up to 16 months of click and impression data straight from Google, so it should always stay in your stack.
- A tool that only displays a proprietary visibility score, without showing which queries or AI answers that figure is based on, is a prime candidate for cancellation.
- Screaming Frog SEO Spider crawls up to 500 URLs for free, which is plenty for technical audits on many smaller sites.
- AI citations are tracked by running the same prompts in ChatGPT, Gemini, and Perplexity and logging whether your brand is mentioned or linked, not through an estimated score.
- Each market needs its own keywords and its own prompts, because a translated Swedish keyword list misses how German or Dutch buyers actually phrase their queries.
What do you get from a 30-minute audit of your SEO tools?
After 30 minutes, you will have a three-column list: keep, replace, cancel. Every tool is evaluated against four questions—roughly seven minutes per question, plus a two-minute inventory. The benchmark is simple: does this data lead to an action regarding content, technical SEO, or budget? If not, it is theater.
Start with the inventory. Open your credit card statement or expense report for the last three months. List every subscription, including add-ons and plugins. Flag anything no one on the team has logged into during the past 60 days. That is often your first win.
| Step | Time | The question you ask | Keep if | Cancel if |
|---|---|---|---|---|
| 0. Inventory | 2 min | What are we paying for? | Someone uses it every month | No login for 60 days |
| 1. SEO vs GEO | 7 min | Do we see Google and AI responses separately? | Two separate, traceable numbers | A blended index |
| 2. Multilingual | 7 min | Is each market written from scratch? | Keywords and copy per language | Machine-translated copy |
| 3. Brand voice | 7 min | Do drafts sound like us before editing? | Voice is ingested before drafting | The editor rewrites everything |
| 4. AI citations | 7 min | Can we drill down to prompt and source? | Log per prompt and engine | Theoretical visibility score |
You do not need to overhaul everything. Most stacks have a solid foundation: Google Search Console stores up to 16 months of search data, a crawler handles technical issues, and a CMS plugin manages basics. On WordPress, Yoast SEO and Rank Math are the standard choices for titles, meta descriptions, XML sitemaps, and structured data. The audit focuses on everything layered on top of that base. If you want to compare options before buying something new, read more on choosing the right tools.
Step 1: Does the tool distinguish between SEO and GEO?
A tool passes Step 1 only if it reports Google rankings and AI answers as two distinct metrics. SEO is about where you rank in a list of links. GEO, Generative Engine Optimization, is about whether an AI answer engine cites you in a generated response. These are different outcomes, and they require different measurement.
Run this five-minute test:
- Open Google Search Console and pick five queries where you rank in the top three.
- Rephrase each query into a question a customer would ask, such as “which SEO tool fits a small e-commerce brand?”.
- Enter the question into ChatGPT, Gemini, and Perplexity.
- Note whether you are mentioned, linked, or missing entirely.
- Compare the result with what your tool claims about your visibility.
If you rank in the top three on Google but do not appear in AI responses, yet your tool shows all green, you have found the theater. The ranking is real. The conclusion your tool draws from it is not.
Do not throw out your rank tracker right away. Position data is still useful for SEO. Just stop treating its visibility score as a proxy for AI presence. If the tool cannot show which prompts and engines a score is derived from, it belongs in the replace column.
Step 2: Does the tool write native content or just translate?
Translation modules that push Swedish text through machine translation fail Step 2. Search behavior differs across markets, not just vocabulary. A German buyer asks different questions, compares different competitors, and expects different examples than a Swedish buyer. A translated page simply answers Swedish questions in German.
Test it like this. Take one published page per language and verify three things:
- The keywords: were they researched specifically for that market, or was the Swedish list translated?
- The examples: are local currencies, regulations, and competitors mentioned, or does it still reference kronor and Swedish market conditions?
- The phrasing: does a native speaker read the text as if it was written in their language, or as a clumsy translation?
Two 'nos' out of three is enough reason to replace the module. Clunky translation damages your brand more than it saves in copywriting costs, and AI answer engines struggle to cite content that fails to match how queries are phrased locally.
Keep tools that handle keyword research per market and generate content natively in each language. If you run WordPress across several language versions, the technical setup is part of the same check, including hreflang and language-specific sitemaps. We covered those checkpoints for WordPress for multiple markets.
Step 3: Does the tool learn your brand voice before publishing?
A publishing workflow passes Step 3 if your brand voice is ingested before drafts are generated, rather than corrected by an editor after the fact. Post-editing is a hidden cost. It does not appear on the software invoice, but it shows up directly on your editor's calendar.
Your test takes seven minutes:
- Take the last three pieces produced by the tool.
- Remove the logo and company name.
- Ask yourself: would an existing customer recognize your voice?
- Measure how much the editor had to rewrite before publishing.
If an editor rewrites more than half of every piece, the tool is not saving time. It is simply shifting the workload.
A frequent trap involves content optimization tools like Surfer SEO and Clearscope. They compare your entities and linguistic patterns against the pages already ranking. That is useful for topical coverage. But when everyone optimizes against the same top ten, everyone ends up sounding identical. Use them to uncover content gaps, not as a substitute for your brand's voice.
Keep what learns your tone from existing assets and applies it consistently in every piece, across every language. Replace tools that churn out generic AI copy you have to clean up afterward.
Step 4: How do you track AI citations so the numbers are verifiable?
Verifiable tracking means every metric can be inspected down to four facts: which prompt, which engine, which date, and which source was cited. A theoretical visibility index gives you none of those. It hands you a trend line for a board meeting, but nothing you can act on.
Here is how to set up honest tracking:
- Write 20–30 prompts per market. Base them on questions customers actually ask sales reps and customer support.
- Run each prompt through ChatGPT, Gemini, and Perplexity. Perplexity cites sources transparently, making it easy to see who gets referenced.
- Log three items per answer: whether your brand is mentioned, whether one of your pages is linked, and which competitors appear.
- Repeat on the same day each week. AI responses fluctuate between runs, so a single snapshot tells you very little.
- Tie citations back to specific pages. Which URLs are cited? Do they also register as traffic in Search Console and your analytics?
Step five separates real tracking from vanity metrics. A citation that never yields a page view, a click, or an inquiry is interesting. A citation that drives a conversion is a business metric.
Running this manually across three markets quickly produces a large volume of responses to log. That is when automation makes sense—provided the tool preserves traceability down to each individual prompt.
What mistakes keep teams stuck building reports instead of driving growth?
The most common error is adopting a tool's proprietary score as the team's primary KPI. When you do that, you optimize for the tool instead of the customer. Clicks, inquiries, sales, and citations are verifiable. An index computed behind closed doors by a vendor is not.
Other recurring patterns:
- Duplicate data sources. Running two keyword databases that report different search volumes for the same term creates endless debates over who is right. Choose one. Industry benchmarks place Ahrefs starting around $129 per month and SE Ranking around $55. Check current pricing before purchasing.
- Ignoring Search Console. It gets dismissed as too basic, even though it is the only source providing first-party Google click and impression data.
- Measuring AI only once. A single run in ChatGPT becomes a sweeping conclusion. Answers vary, so track them systematically over time.
- Translating keyword lists. The same keywords are rolled out across languages, and no one understands why traffic fails to materialize in Germany.
- Treating the report as the product. Teams spend Fridays building dashboards instead of shipping content.
For more pitfalls specific to AI search, see our overview of common AI SEO mistakes. If you want to calculate what a tool needs to generate to justify its cost, you can calculate the break-even point for your entire SEO investment.
Worked example: what the audit looks like for a company in three markets
Imagine a growth company selling B2B software in Sweden, Germany, and the Netherlands. This scenario is constructed to demonstrate the methodology, not drawn from a single live client. The stack contains seven tools: Search Console, Ahrefs, Screaming Frog, Yoast, a rank tracker with a proprietary visibility score, a translation plugin, and a dashboard platform.
Minutes 0–2: Inventory. The dashboard tool has not been opened in three months. It goes straight to the cancel list.
Minutes 2–9: SEO vs GEO. The rank tracker shows climbing visibility in Germany. The team runs five test questions through ChatGPT, Gemini, and Perplexity. The brand is absent from every German response. The rank tracker is kept for monitoring positions, but its visibility index is dropped from reporting.
Minutes 9–16: Multilingual. The German and Dutch pages were machine-translated from Swedish and still refer to Swedish examples. The translation plugin is marked for replacement.
Minutes 16–23: Voice. The editor has to rewrite the bulk of every piece. That confirms the decision made in Step 2.
Minutes 23–30: AI citations. The team sets up 25 prompts per market. 25 prompts across 3 markets and 3 engines equals 225 responses per audit run. Suppose the brand is cited in 30 of them, nearly all in Swedish. The problem is not Google. The problem is that the other markets lack native content that AI engines can quote.
The outcome: four tools kept, one canceled, one replaced, and one vanity score swapped out for a prompt log. Follow up with a complete SEO audit of your website once your stack is streamlined.
Frequently asked questions
What is the best free SEO tool?
Google Search Console, because it provides Google's own data on clicks, impressions, and indexation. Pair it with Bing Webmaster Tools, which is also free and shows how Bing indexes your site. Bing's index is also leveraged by multiple AI services for web search.
Do I need Ahrefs or Semrush?
You need a keyword and backlink database if you conduct competitor analysis and look for topics where rivals capture traffic you are missing. Smaller teams can often get by with a budget option like SE Ranking or Ubersuggest. Avoid running two large databases simultaneously without a clear operational reason.
Can standard SEO tools measure visibility in ChatGPT?
Most traditional tools track Google rankings and estimate everything else. To verify whether ChatGPT, Gemini, or Perplexity cites your brand, someone has to run real prompts and record the answers. Insist that your tools show the exact prompt and source behind every metric.
How often should I audit my tool stack?
Every 90 days works well for most teams. Schedule an extra review right before annual renewals—such as Screaming Frog's paid license at £199 per year—so you never renew out of habit.
Are Yoast SEO or Rank Math enough?
They take care of the on-site fundamentals on WordPress: titles, meta descriptions, sitemaps, and schema markup. They do not write native copy per market or monitor AI citations. Treat them as a foundational layer, not the entire stack.
Run the audit before your next renewal
Block 30 minutes on your calendar this week. Pull up your recent statements, walk through the four steps, and populate your three columns: keep, replace, cancel. Then set a calendar reminder for 90 days from now.
If you want to avoid doing this manually, that is precisely what we built. UPGeoSEO pairs SEO and GEO audits, tracks citations across ChatGPT, Gemini, and Perplexity, and generates native content from scratch for each target market in your brand voice. Articles publish directly to platforms like WordPress, with social assets generated automatically. With Autopilot, optimization runs every month. You review and approve. We publish. And you monitor performance using verifiable figures rather than theater.
