What are the most common SEO mistakes?
Ranked by damage rather than frequency, because the mistakes that appear on every list are rarely the ones costing you the most.
Most lists of SEO mistakes are ordered by how easy the mistake is to describe. Missing alt text and duplicate meta descriptions appear near the top because they are tidy, checkable, and produce a satisfying number in an audit tool.
They are also close to irrelevant. This list is ordered by what the mistake actually costs, which produces a very different ranking — and puts four problems at the top that most audit tools will not flag at all.
Key takeaways
- The expensive mistakes are invisible in the tools that grade your pages out of 100.
- Unindexed pages make every other effort on those URLs worthless.
- Migrations without redirects are the fastest way to lose most of your traffic in a fortnight.
- Cannibalisation is self-inflicted and affects nearly every site older than two years.
- Several classic “mistakes” cost almost nothing and absorb attention that belongs elsewhere.
The four expensive ones
1. Assuming pages are indexed.
This is first because it invalidates everything else. A page that is not in the index cannot rank, cannot be cited, and cannot receive a visit, no matter how well written or how many links point at it.
Sites publish for months without checking, and the gap is often enormous — a content programme producing forty pages a quarter where half never enter the index. Every hour spent on those pages was spent on nothing. Google Search Console’s Pages report gives you the number in about ten seconds, and the most common reason for exclusion is “Crawled — currently not indexed”, which is a quality judgement rather than a technical fault and therefore needs a content answer rather than a technical one.
2. Changing URLs without redirects.
The fastest way to destroy organic traffic, and it usually happens during a redesign that everyone is pleased with. Every URL accumulates history — links, engagement, ranking signals. Changing the address without a permanent redirect discards all of it and starts the new URL at zero.
The damage appears within two to six weeks and is frequently misattributed to a coincidental algorithm update, which means the actual cause goes unfixed. If traffic fell sharply and a launch happened in the preceding month, check the redirects before considering any other explanation. Our guide to diagnosing traffic drops works through the distinction.
3. Multiple pages competing for the same query.
Cannibalisation is the most common self-inflicted problem on any site that has been publishing for a few years, and it accumulates silently. You write about a subject in 2024, again from a different angle in 2025, and a third time as part of a campaign. Now three URLs target one intent.
Search engines respond by choosing one, splitting signals across the others, and often rotating which they show. None establishes a stable position. The fix — merge into one strong page, redirect the rest — regularly produces larger gains than a quarter of new publishing, because it concentrates authority that was being divided rather than creating new authority.
Find it by searching your own domain for each important query and counting how many of your pages appear.
4. Targeting queries the site cannot win.
Strategic rather than technical, and the most expensive in opportunity terms. A year spent pursuing head commercial terms against incumbents with a decade of authority produces nothing, while the specific queries the site could have won stay uncovered.
The check takes two minutes per query: search it, look at who ranks, and ask honestly whether your site could plausibly displace them within the time you have. If the answer is no, the query goes on a list for later rather than into this quarter’s plan.
The moderately costly ones
5. Titles written for the algorithm rather than the click. Pages that rank at position six with a 0.8% click-through rate are earning a fraction of what the position is worth. Titles are the cheapest lever on any site with existing impressions, and most sites have never deliberately written one.
6. Burying the answer. A page that spends four hundred words establishing context before answering the question loses the visitor, the featured snippet, and any chance of being extracted by an AI assistant. The instinct to build up to a conclusion is correct for essays and wrong for search.
7. Orphan pages. Pages with no internal links pointing at them get crawled rarely and read as unimportant. This happens constantly with pages created outside the normal publishing flow — landing pages, resources, anything built for a campaign.
8. Ignoring pages at positions 8 to 20. Not a mistake of commission but of allocation. These pages have proven relevance and need a nudge; new pages need everything. Sites consistently spend on the second while the first sits untouched.
9. Publishing volume without editorial control. Sixty thin pages lower the average quality of the domain, create cannibalisation that takes a quarter to unwind, and spread crawl attention across URLs that do not deserve it. Generation made this cheap to do at scale, which made it a far more common failure than it was five years ago.
10. Blocking resources in robots.txt. Blocking CSS and JavaScript directories prevents rendering the page as a user sees it, which can cause misjudgement of layout, mobile usability and content placement. It is usually a legacy line nobody has read in years.
Mistakes that cost less than you think
These appear on nearly every checklist. Each is worth fixing if it is free, and none is worth a project.
| Commonly flagged | Actual cost |
|---|---|
| Missing meta descriptions | Low. Not a ranking factor; Google writes one. Matters only for CTR on important pages |
| Missing alt text | Low for ranking, real for accessibility. Fix it for the right reason |
| Multiple H1 tags | Near zero. Long since handled by parsers |
| Keyword not in URL | Marginal. Readable and stable beats exact-match |
| Perfect PageSpeed score | Diminishing past a point. Field data matters; the lab score is a proxy |
| Broken external links | Low for ranking. Fix for credibility |
| No blog | Zero, inherently. A blog is one way to cover queries, not a requirement |
The reason these dominate audit reports is that they are countable. An audit that reports 340 missing alt attributes looks thorough; an audit that reports “your three most valuable target queries are unwinnable for eighteen months” looks thin and is worth a hundred times more.
New mistakes specific to AI search
Assistants retrieve pages before answering, and they behave differently from Googlebot in ways that create genuinely new failure modes.
Content that only exists after JavaScript runs. Google will usually render it eventually. AI crawlers largely will not execute scripts, so a client-side rendered page can be effectively empty to them. This is the single biggest new technical mistake, and it is invisible in every conventional SEO tool. Test by fetching your key pages with JavaScript disabled.
Writing passages that cannot stand alone. An assistant lifts a section, not a page. A paragraph whose meaning depends on the three above it cannot be quoted usefully, so it will not be — regardless of how good the page is as a whole.
Hedging every claim. Vague ranges and heavy qualification make text unquotable. “Typically six to twelve weeks” gets cited; “this can vary considerably depending on circumstances” does not.
Assuming Search Console covers it. An AI answer that names you produces no impression, no click and no row in any report you currently read. Brands routinely discover months late that assistants have been describing them inaccurately, or recommending a competitor, in the exact research conversations that precede a purchase. The measurement gap is covered in monitoring AI brand mentions.
The mistake behind the mistakes
Nearly everything above shares a cause, and naming it is more useful than the list.
Sites optimise what they can see, and the important things are not visible by default.
An audit tool shows missing alt attributes, so alt attributes get fixed. Nothing shows that forty pages are unindexed unless you go and look. Nothing announces that three of your posts are competing for one query. Nothing flags that your content only exists after JavaScript runs, or that an assistant has been recommending a competitor for six months. These problems are silent, and silent problems lose to noisy trivial ones every time.
This is also why the cheapest possible intervention — one afternoon, once a quarter, deliberately looking at the things no tool reports — outperforms continuous work on the things tools do report. The sequence below is that afternoon.
A related pattern worth naming: the mistakes that hurt most are the ones committed by people outside the marketing team. A developer changes a URL structure. A designer removes text that was carrying the page. A CMS update alters canonical behaviour across the site. Someone leaves a staging noindex in place. None of those people did anything unreasonable, and none of them knew the consequence, because nothing in their workflow surfaces it.
The defence is not technical. It is a habit: whenever URLs, templates or rendering change, someone checks indexation and organic traffic for the following fortnight. That single practice prevents more damage than everything in the moderately-costly list combined.
Finding yours in an afternoon
In order, because each step can make the later ones unnecessary.
- Search Console, Pages report. Compare indexed against published. A large gap is your problem and nothing below matters until it closes.
- Check for recent URL changes. Any launch in the last three months — verify old URLs return a 301 to the right place, not a 404 or a homepage redirect.
- Cannibalisation sweep. For your ten most important queries, search your own domain and count competing pages.
- Export by impressions. Find high-impression, low-CTR pages. Those are title rewrites and they are the fastest wins available.
- Fetch three key pages with JavaScript off. If the content disappears, you have an AI retrieval problem that no SEO tool has told you about.
- Read the SERP for your three main targets. Decide honestly whether they are winnable this year.
That sequence finds the expensive problems. It will not produce a score out of 100, which is precisely why it is more useful than the reports that do — a graded checklist rewards the mistakes that are easy to count, and the ones costing you money are not among them.
Find the expensive problems, not the countable ones
Klepha checks indexation, cannibalisation and striking-distance pages, and tests whether AI assistants can actually retrieve and cite your content.
Run a free visibility reportFrequently asked questions
What is the most common SEO mistake?
Assuming pages are indexed without checking. Sites routinely publish for months while a large share of their pages sit unindexed, which makes every downstream effort worthless for those URLs. It costs nothing to check in Search Console and it is the single most frequent cause of work producing no result.
What SEO mistake costs the most traffic?
A site redesign or migration that changes URLs without correct redirects. It can remove most organic traffic within weeks, and unlike gradual problems it happens all at once. The damage is also the hardest to reverse, because the accumulated signals attached to the old URLs are discarded rather than moved.
Does keyword cannibalisation really hurt?
Yes, and it is one of the most common self-inflicted problems on sites older than about two years. When several pages target the same query, search engines split the signals between them and often rotate which one they show, so none establishes a stable position. Consolidating them into one strong page regularly produces bigger gains than writing new content.
Is publishing too much content an SEO mistake?
It can be. Volume without editorial control produces overlapping pages that compete with each other, dilutes the average quality of the domain, and spreads crawl attention across pages that do not deserve it. Ten pages that each own a specific question outperform sixty that half-answer the same few.
How do I know which SEO mistakes my site has?
Check four things in order: how many of your pages are indexed versus published, whether multiple pages target the same query, whether your highest-impression pages have deliberately written titles, and whether any recent URL changes were redirected. That sequence finds the four most expensive problems in about an afternoon and requires no paid tools.