An SEO audit doesn’t start by installing Screaming Frog or hiring the trendy tool. It starts with an uncomfortable question: what do I need to prove, and to whom? If you’re going to convince a client that their website has an indexing problem, an 80-page report generated by a paid suite is less impressive than a Search Console screenshot with the right date and filter. And if you manage an ecommerce site with 200,000 URLs, doing it only with free tools will cost you more hours than the license costs.
This article isn’t a list of tools. It’s a walkthrough of the decisions that separate a useful audit from a PDF that nobody reads, with the focus on where it’s worth paying and where the free plan is enough (or even better).
Before choosing a tool: define the real scope
I’ve seen failed audits because they started with crawling. The correct order is the reverse: first the scope, then the evidence, then the tool.
- Site size: below 500 URLs, almost everything can be done for free. Above 5,000, free crawling becomes a practical limitation, not a budget question.
- Business type: a local business with 12 pages needs an audit focused on Google Business Profile and NAP, not on crawl budget.
- Who’s going to read the report: if it’s a developer, you need logs and status codes. If it’s management, you need three charts and a prioritized list.
A nuance almost nobody mentions: an SEO audit doesn’t diagnose content quality. You can have a flawless crawl, zero 404 errors, and still fail to rank because your pages don’t match search intent. Tools tell you the page exists and is crawlable; they don’t tell you whether it deserves to be there. That’s for you to decide by reading the SERP.
Phase 1: indexing and crawling, the ground where free wins more than it seems
Here the free plan isn’t a consolation prize. Google Search Console is the only source that tells you what Google has actually crawled and indexed. Any paid tool gives you an estimate; Search Console gives you the data from the search engine itself.
What to check in Search Console without paying anything
- Pages report (Indexing): separate “Crawled, currently not indexed” from “Excluded by noindex tag.” They’re different problems with different solutions, and many people lump them together.
- Crawl stats: check whether crawling is concentrated on URLs with parameters or on filter pages that add no value.
- URL inspection: use it to verify the real status of a specific page before and after a change. It’s the fastest validation tool there is.
Honest limitation: Search Console shows aggregated, delayed data. It doesn’t tell you which specific bot crawled which URL or how many times. For that you need server logs, and there the free alternatives are few and far between.
The mistake almost everyone makes with free crawling
Free versions of desktop crawlers usually limit the number of URLs per session. The temptation is to crawl only the homepage and trust that the rest is fine. Wrong. A specific case: a site with a nested category architecture where the free version only reached the first 500 URLs left out precisely the product pages, which were where the badly set canonicals were. The report came out clean and the problem was still alive.
If your free crawl doesn’t reach the deep pages, you have two sensible options: pay for one month of license (not a year) for the one-off crawl, or use the Search Console indexing report as your main source and the crawl only to validate specific hypotheses.
Phase 2: performance and Core Web Vitals, where free is the standard
There’s no real debate here. PageSpeed Insights, Lighthouse, and Search Console field data cover the essentials and are free. The difference between free and paid isn’t in the measurement, it’s in the history and URL segmentation.
Paid tools let you see the evolution of LCP by template over months and detect regressions. Free ones give you a snapshot of the moment. For a one-off audit, the snapshot is enough. For monitoring after a redesign, the history saves you arguments with the development team.
A detail from experience: lab data (Lighthouse) and field data (CrUX) often disagree. If in your report you say “the website is slow” based only on Lighthouse and the client has good field data on mobile, you lose credibility in the first meeting. Always cite the source and the type of data.
Phase 3: content and on-page, the point where free falls short
The content audit is the phase most dependent on paid tools, and also the one most often misunderstood. Detecting keyword cannibalization, duplicate titles at scale, or orphan pages requires cross-referencing data that free versions don’t let you export comfortably.
What you can do for free, rigorously
- Export the Search Console performance report and cross-reference queries by page to detect two URLs competing for the same intent.
- Check duplicate titles and meta descriptions with a limited crawl or with browser extensions for small samples.
- Check the Hn structure on key pages manually. On a 40-page site, this takes an afternoon and is more reliable than any automation.
Where free doesn’t reach
Detecting near-duplicate content at scale, analyzing semantic similarity between pages, and historical position tracking by cluster are features that exist in a very limited form on free plans or with short data windows. If your audit includes deciding which URLs to consolidate on an ecommerce site with product variants, you’ll need a paid tool or manual work that probably won’t be cost-effective.
Link data is expensive to produce, and it shows. Free alternatives give you a sample, not the full profile. Using that sample to claim a competitor has a toxic profile is a methodological error, not a finding.
For an internal audit, the minimum defensible approach is to review new inbound links in Search Console, identify suspicious domains, and manually check for obvious patterns (repeated exact-match anchors, spam directories). You don’t need a subscription for that. What you can’t do for free is compare link profiles of five competitors with authority metrics and temporal evolution.
Practical comparison: when to pay and when not to
| Need | Free alternative | When it’s worth paying |
|---|---|---|
| Indexing status | Search Console | Almost never for this purpose |
| Full site crawl | Limited by number of URLs | Large sites or recurring audits |
| Core Web Vitals | PageSpeed Insights + CrUX | When you need history by template |
| Cannibalization and clusters | Manual cross-referencing in spreadsheets | More than 300 URLs or monthly audits |
| Link profile | Partial sample | Competitive analysis or disavow |
| Position tracking | Search Console data (delayed) | When you need daily positions per keyword |
The conclusion of this table isn’t “pay.” It’s that spending should be concentrated on the phase that drives the most decisions in your context. A consultant who does one-off audits can live with the free plan plus one month of license when the project demands it. An agency with ten active clients needs a subscription, if only for the export time.
How to put together the report that actually gets read
The tool doesn’t matter. The report is organized by impact and effort, not by technical categories. A structure that works:
- Findings that block indexing (with Search Console evidence and date).
- Problems that affect performance (with the example URL and the field data).
- Content opportunities (with the query and the page that should answer it).
- Link risks (only if there’s evidence, not suspicions).
And a warning worth putting in writing: no audit guarantees ranking improvements. It detects problems and proposes fixes; the result depends on execution, competition, and factors you don’t control. Promising otherwise is selling smoke.
What no audit covers
There are three things that escape even paid tools, and it’s worth saying so in the report:
- The user’s real intent changes over time and no metric captures it completely.
- The impact of algorithm changes can’t be predicted with an audit; it can only be monitored.
- Editorial quality requires human judgment. A tool flags a text as “optimal” based on length and density; an editor detects that it adds nothing.
Acknowledging these limits doesn’t weaken the audit: it makes it credible. The client who understands what you’re measuring and what you’re not values the report more than the one who receives an endless list of “errors” without context.
Final checklist before delivering
- Does every finding have an example URL and a cited source?
- Have I separated what’s a real problem from what’s a best-practice recommendation?
- Have I indicated which tools are free and which are paid in each phase?
- Have I prioritized by impact, not by order of appearance in the tool?
- Have I avoided claiming results I can’t back up?
With this, the decision between free and paid alternatives stops being ideological and becomes operational: you pay where the time saved or the depth of the data changes a decision. For the rest, the free plan doesn’t just suffice—sometimes it’s the most reliable source.