Category: SEO

A roofer calls in May, after the first storms, wanting to be visible for the season. The season is already running. Whatever we build now lands in August.

This is the single most consequential decision in seasonal trades marketing and it is almost never treated as a decision. It is treated as when somebody got round to it.

The arithmetic

Work takes weeks to be crawled and reflected, months to settle into position, and longer again in a competitive local market. Call it three months before anything is meaningfully working, on a site that is otherwise healthy.

Now put that against a demand curve that is not flat. Roofing spikes after storms and through late summer. Heating spikes on the first genuinely cold night. Cooling spikes in the first real heat.

Start three months before the spike and you arrive as demand does. Start at the spike and you arrive as it ends, having paid for the whole thing and captured the tail.

The honest version of the advice: the best time to start is the quiet season, which is precisely when nobody feels like spending money on marketing. That tension is the whole problem and it is worth naming rather than talking around.

What to do if the call comes in May anyway

Because it will, and “you should have started in February” is not advice.

What to do in the quiet months

This is the part that gets skipped, and it is where the year is actually won.

Publish the content that takes time to mature. The pages you want ranking in October should exist in June.

Collect reviews while the work is fresh. A trade that goes quiet for months has a review timeline with a visible gap in it, and BrightLocal’s 2026 survey of 1,002 US consumers found 74% look for reviews written in the last three months. Ask every customer during the busy season, not after it.

Photograph the jobs. Nobody wants to stage a photograph in November. The gallery you need next year is on somebody’s phone right now.

The thing I would stop doing

Switching the whole programme off in the off-season and on again in the spring. It is intuitive and it wastes the previous year’s work. Position decays, the review timeline gaps, and the rebuild starts partway back down the hill.

A smaller continuous programme beats a large intermittent one in every seasonal account I have seen. The spend curve should be flatter than the demand curve, not matched to it.

The other reason to be careful about arriving at the peak: succeeding at the wrong moment is its own problem. I have ended campaigns because they worked. A business that cannot answer the phone during a storm week does not benefit from being first in the pack during it.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Roofing and HVAC SEO covers seasonality, the profile and the phone. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

I have watched a firm spend a year improving its visibility while losing people at the front door, and the marketing was working perfectly the whole time.

That is the uncomfortable version of this job. Everything I control ends at the moment somebody calls. Everything that decides whether the money was well spent happens after.

The number that makes it an SEO problem

MIT’s Professor James Oldroyd, working with InsideSales.com, analysed more than 15,000 leads and 100,000 call attempts across six companies and found a twenty-one-fold drop in the odds of qualifying a prospect when response time went from five minutes to thirty.

Five to thirty. Not five minutes to the next day.

Now put that next to what a personal injury lead costs. First Page Sage tracked 49 PI firms across 36 states and put the average at $442 through Google Search Ads and $183 through SEO. Their data, their clients, and they sell SEO. I quote it with that caveat attached and I am also an SEO consultant quoting it to you.

Either number, multiplied by a twenty-one-fold drop in qualification odds, is the most expensive half hour in the firm.

It also decides whether the case for organic makes sense at all. The whole argument for building search over buying clicks rests on a cheaper cost per lead compounding over time, and a lead that is never contacted costs the same as one that signs. Intake is the multiplier sitting underneath both columns.

Why it is worse in personal injury than almost anywhere

The caller is shopping in parallel. Somebody in the days after an accident is not calling one firm. They are working down a list, and whoever engages first has an enormous advantage that has nothing to do with quality.

The competition answers. The firms spending five figures a month on ads have staffed intake, because they have already done this arithmetic. You are not competing against nobody.

The calls do not arrive at eleven on a Tuesday. Accidents happen at weekends and at night. The hours when your visibility is worth most are the hours the office is closed.

What I check before agreeing a campaign

I ask to call the firm, from an unknown number, at a time nobody is expecting it. It is a five minute test and it has changed the shape of more engagements than any crawl I have run.

What I would rather do than sell a campaign

If the answers are bad, fix that first. It is cheaper than SEO, it works immediately, and it improves the return on every channel the firm is already paying for.

Doubling visibility for a firm that misses half its calls doubles the number of people who now have a poor opinion of them. That is not a neutral outcome. I have stopped campaigns over exactly this, in trades rather than law, and the mechanism is identical.

The part that is genuinely not my job

I can tell a firm their intake is losing cases. I cannot staff it, and I have no useful opinion on whether they should hire, use an answering service, or take calls themselves.

What I will not do is take the money, produce the leads, and let the firm conclude twelve months later that SEO does not work for them. It worked. It arrived at a front door nobody was standing behind.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Personal injury SEO covers visibility, and it is worth nothing without the intake behind it. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

Somebody hit by a truck on the interstate is not searching “personal injury attorney”. They are searching what happened to them.

Most firm sites have one Personal Injury page with a bulleted list underneath: car accidents, truck accidents, motorcycle, slip and fall, dog bites, wrongful death. That page is trying to be the answer to fifteen different questions asked by fifteen different people in fifteen different states of mind.

Why the single page loses

The head term is also the most expensive and most defended query in the vertical. Competitive personal injury clicks run into the hundreds of dollars precisely because everybody is fighting over them, and a general PI page is fighting there with no differentiation.

The case-type queries are a different market. Lower competition, far higher intent, and the person searching has already self-selected into a case you can evaluate. “Truck accident lawyer” is a person with a case. “Personal injury attorney” is often a person browsing.

There is also less room on the page than there used to be. SparkToro’s analysis of Similarweb’s US clickstream found 68.01% of Google searches ended without a click in early 2026, up from 60.45% two years earlier. Competing broadly for a shrinking share of clicks is a worse bet every year. Competing specifically, where the query is unambiguous, is not.

Where the line is

A page per case type is legitimate. A page per case type per city is usually not, and Google’s spam policies name the difference: doorway abuse includes “pages targeted at specific regions or cities that funnel users to one page” and “substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy.”

A truck accident case and a dog bite case have nothing in common: different defendants, different insurers, different regulations, different evidence, different timelines. Writing both is writing two documents.

“Car accident lawyer Springfield” and “car accident lawyer Shelbyville” are the same document with a noun swapped. Same test as always: could you write the second page without looking at the first?

Which case types get a page

For most firms that is five to eight pages, not fifteen.

What goes on one

The questions in the order the person has them. Who pays. How long this takes. What happens to my medical bills in the meantime. Do I have to go to court. What is my case worth, all answered honestly, including why nobody can tell them precisely.

Then the part only you can write: what you do first on this case type, what evidence you go after, what you have seen go wrong. On a truck case that might be the preservation letter and the electronic logging data. That paragraph is the difference between a page and a brochure.

What this does not fix

Better pages produce more leads, and leads are not cases. The economics only work if the intake works. A firm converting well at $442 a lead is in a better position than one converting badly at $183, and no amount of page structure touches that.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Personal injury SEO covers the page set, the profile and the economics. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

Both are criminal defence. On a website they are usually the same page. As businesses they have almost nothing in common, and treating them as one thing is why so many defence sites read as generic.

The difference runs through everything: how the client searches, how long they take to decide, what they need to read, and what evidence of competence they will accept.

How the client arrives

DWI. Fast, high volume, mostly mobile, often within a day or two of the arrest. The search is the charge plus the place. The person is anxious but the situation is comprehensible. They know roughly what a DUI is, they want to know what happens now.

Federal. Slow, low volume, and frequently the search starts before any charge exists: a target letter, a subpoena, a visit from agents. The queries are different in kind: not “federal defence lawyer” but the specific situation. The person may research for weeks and consult several firms.

One of those is a decision made in an evening. The other is a decision made over a month with a family involved.

What that changes on the site

DWI pages should be fast, direct and mobile-first. Deadlines up top. The DMV hearing window is the classic one and it is genuinely urgent. Short paragraphs. A phone number that works with a thumb. Nobody facing a DUI reads three thousand words at eleven at night.

Federal pages should be long, and it is the one place I would say that without hedging. Somebody deciding who defends them against a federal indictment will read everything you have written. Depth is the trust signal. This is the opposite of my usual position on length and it is the audience that justifies it.

The evidence each audience accepts

DWI clients behave like ordinary local consumers. They read reviews. BrightLocal’s 2026 survey of 1,002 US adults found 97% read reviews for local businesses and 31% will only use one rated 4.5 or above, and a DWI client is comfortable leaving one afterwards, because the outcome is a relief rather than something to bury.

Federal clients accept almost none of that. They want to know which courts you are admitted to, whether you have tried a case in that district, what you have published, whether you have been a prosecutor. The substitutes for testimony matter far more at that end of the docket.

Which produces the pattern I wrote about in the parent post: review volume tracks charge severity inversely. The firms with the most reviews are usually doing the least serious work.

The map pack splits them too

DWI is a local pack business. Distance matters, proximity matters, the profile matters.

Federal is not. The relevant court may be a hundred miles away and the client will travel or hire remotely. Optimising a federal practice for the local pack is effort spent in the wrong competition entirely. the website and the pack are two different contests and federal lives almost entirely in the first.

Google’s local ranking guidance makes the reason plain: the pack weighs relevance, distance and prominence, and distance is “how far each business is from the customer who’s searching.” For a federal client that measurement is close to meaningless, because they are not choosing a lawyer by driving time, so the entire mechanism the DWI side depends on simply does not apply.

What I would actually do

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Criminal defense SEO covers both ends of the docket. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

A defence firm covering four counties wants a page for each. Reasonable instinct, because the courts genuinely are different, and clients search that way.

It is also the single easiest way to build a doorway set without meaning to, and the difference between the two versions is not subtle once you know what to look for.

What Google actually prohibits

Worth quoting rather than paraphrasing. Google’s spam policies define doorway abuse to include “having multiple domain names or pages targeted at specific regions or cities that funnel users to one page” and “creating substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy.”

Notice what is not prohibited: having a page about a place. What is prohibited is a set of near-identical pages whose only function is to catch a place name and hand the visitor to the same destination.

So the question is never “are location pages allowed”. It is “is this page a document about something, or a net”.

The test

Could you write the second page without looking at the first?

For two genuinely different courts, yes. Easily, and you would struggle to make them similar even if you tried. For two towns in the same judicial district served by the same courthouse under the same DA, no. You would be writing the same page and changing a noun.

Which points at the right unit. The page should be about the court, not the town. Four counties with four different courthouses is four pages. Twelve towns inside those four counties is still four pages, because the twelve towns share four courts and nothing else distinguishes them.

That reframing usually reduces a proposed forty-page project to six, and makes the six defensible.

What genuinely differs, and therefore what goes on the page

This is the part that makes the page real, and a defence lawyer already knows all of it. It just rarely gets written down.

Any single one of those makes the page non-substitutable. All four make it the most useful page on the subject anywhere, which is the actual goal.

The pack question, which people conflate with this

A county page will not put you in the map pack for that county. The pack ranks your Business Profile, and distance is measured from where your office actually is, and a page cannot move it.

What the county page does is compete in the ordinary results, where there is no distance constraint. That is the right expectation to set before anyone writes six pages hoping for a pack position they cannot get.

Google is explicit that the pack runs on relevance, distance and prominence, and adds that there is “no way to request or pay for a better local ranking on Google.” A page is not one of those three inputs. It is worth saying out loud to a firm that has been sold county pages as a map pack strategy, and worth reading alongside what actually decides the pack in this vertical, which is rarely what people expect.

How many, honestly

As many as you have distinct courts and real knowledge of. Usually three to six for a firm that genuinely practises across a region.

If the answer is “we would take a case anywhere in the state”, that is not a content strategy, and forty pages asserting it will read exactly like what it is. Write the ones you can write properly and let the rest be covered by the practice’s main pages.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Criminal defense SEO covers the page set and the local layer. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

Almost every piece of advice about demonstrating experience and trust assumes you can point at satisfied customers. Testimonials, case studies, named results, before-and-afters, a wall of reviews.

A criminal defence firm can do almost none of that. The better the outcome you delivered, the more your client needs no permanent public record that they were ever your client.

So the standard playbook is unavailable, and the honest question is what actually replaces it.

The scale of what you are giving up

Worth being clear-eyed about the cost before working around it. BrightLocal’s 2026 Local Consumer Review Survey, covering 1,002 US adults in February 2026, found 97% read reviews for local businesses, and that 31% will only use one rated 4.5 or above.

Those respondents were answering about restaurants and plumbers, and I have argued elsewhere that the finding does not transfer cleanly to a serious criminal charge. But it is not zero either. Some proportion of people facing a charge will look you up the way they look up anything, and find less than they expected.

What actually substitutes

Five things, in rough order of how much work they do.

Demonstrated reasoning about the law. This is the strongest and the most under-used. A page explaining how a particular charge is actually defended, covering what the prosecution has to establish, where those cases typically break down and what you look at first, demonstrates competence directly rather than reporting it second-hand. It is also the thing nobody else can copy, because it is your judgement.

Verifiable facts about the lawyer. Bar admissions, years in practice, courts admitted to, former prosecutor experience, teaching, published writing, bar association roles. These are checkable, which is precisely what makes them useful in the absence of testimony.

Case results, within your bar’s rules. Most jurisdictions permit outcomes without identifying details and with a disclaimer. “Felony assault charge dismissed at preliminary hearing, 2024” names nobody. What your bar allows varies enormously and it is worth getting that answered precisely rather than avoiding the category entirely out of caution.

The reviews you can get. Not every client is unnameable. A DWI client who is relieved and unembarrassed will often review you happily. Ask those clients. A firm with fifteen genuine reviews from the cases where reviewing is comfortable is in a better position than one with none on principle.

Being visibly a real person. A photograph of the actual lawyer, a real biography, a phone number that reaches somebody. Sounds trivially obvious; a surprising number of defence sites are anonymous stock photography and a form.

What Google is actually asking for

Worth reading the source rather than the commentary. Google’s helpful content guidance frames this as a set of self-assessment questions: does the content demonstrate first-hand expertise, would a reader trust it, is it clear who produced it and what their credentials are.

Nothing in that requires customer testimony. It requires evidence of expertise and clarity about who is speaking. A defence firm can satisfy both completely without naming a single client.

What I would stop doing

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Criminal defense SEO covers trust signals in a practice where clients cannot be named. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

Nobody arrested last night is searching “criminal defense attorney”. They are searching the thing that happened to them: the charge, by name, usually with the city attached.

So the page that should meet them is a page about that charge. Most defence firm websites have one page called Criminal Defense with a bulleted list of everything the firm handles, and it meets nobody.

Why the practice area page loses

Three reasons, and the third is the one people miss.

Where the line is, because this can go wrong

A page per charge is legitimate. A page per charge per city usually is not, and the difference matters because Google names the failure mode directly.

Google’s spam policies define doorway abuse as, among other things, “having multiple domain names or pages targeted at specific regions or cities that funnel users to one page” and “creating substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy.”

Read that against what a charge page is. DUI and assault are genuinely different: different statute, different penalties, different process, different questions. Twelve charge pages are twelve different documents. That is a browseable hierarchy of real distinctions.

“DUI lawyer Springfield” and “DUI lawyer Shelbyville”, identical but for the place name, are substantially similar pages funnelling to one contact form. Same law, same process, same everything. That is the pattern the policy describes.

The test I would apply: could you write the second page without looking at the first? For two charges, yes. For two towns, usually not.

Which charges get a page

Not all of them, and this is where the discipline is.

Charges you actually take. Obvious and routinely ignored. A page for a charge you would refer out is a page that generates calls you do not want and dilutes the ones you do.

Charges with enough volume to be worth the page. DUI, drug possession, domestic violence, assault, theft. Below a certain volume the charge belongs as a section on a broader page, not a page of its own.

Charges where you have something specific to say. If the page would be a restatement of the statute, it is not a page. It is a Wikipedia entry with your phone number on it.

On most firms that is somewhere between six and twelve pages. Not forty.

There is a second reason to keep the number honest, which is that the result page is getting less generous. SparkToro’s analysis of Similarweb’s US clickstream found 68.01% of Google searches ended without a click in January to April 2026, up from 60.45% in 2024. Thin pages competing for a share of a shrinking click are not a volume play any more. Fewer, better-answered charges is.

What goes on one

The questions the person actually has, in the order they have them. What am I facing. What happens next and when. What are the deadlines. What are the realistic outcomes. What does this cost.

Then the firm-specific part, which is the only bit a competitor cannot copy: how you approach this charge, what you look at first, what you have seen work. In a practice where clients can rarely be named, that judgement is doing the work testimonials do everywhere else.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Criminal defense SEO covers the page set, the profile and the local pack. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

I once fixed a heading problem on a client’s site, confirmed it, wrote it up, and moved on. Three weeks later it was back. Nobody had touched the page. A theme update had restored the original template and taken my change with it.

Nobody noticed for months, because heading markup does not show on screen. That is the whole trap: the fixes most likely to be silently reverted are the same ones that were invisible in the first place.

Why it happens

Most site owners edit in the interface. The interface writes to whatever layer is easiest: a page, a block, a builder module, sometimes a database row that overrides the template.

The template underneath still says what it always said. So the fix and the source of the problem now disagree, and the next thing that touches the template wins. A theme update, a builder version bump, someone duplicating a page from the old layout, a section reset.

The change was never wrong. It was just made in the wrong layer, and the wrong layer is the default one.

Which fixes are at risk

Anything invisible. Headings, canonicals, meta robots, structured data, alt text on template-driven images. If reverting it produces no visible change, nobody will spot the revert.

Anything you did in a page builder on one page that is really a template problem. Fixing seven blank H1s on one home page is a page edit. The builder emitting a heading per section container is a template problem, and it will emit them again on the next page anyone builds.

Anything a plugin also has an opinion about. Two things writing the same tag is a fight you will lose at an unpredictable moment.

Visible things are self-policing. If a theme update moves your logo, someone will mention it that afternoon.

This is why heading problems in particular survive for years. They are emitted by a template, they are invisible on screen, and any fix made in the editor is sitting in the layer that loses.

The aggregate picture is consistent with things quietly going backwards. WebAIM crawled the top million home pages in February 2026 and found skipped heading levels on 41.8% of them, up from 39% the year before, with detected errors per page rising 10.1% year on year. Whatever is happening across the web, it is not a slow accumulation of fixes holding.

The scale of the exposure

HTTP Archive’s 2025 Web Almanac puts 54% of the web on a CMS, WordPress at 64.3% of those, and roughly 60% of WordPress sites running a page builder, with Elementor at 43%. So a very large share of sites are built in exactly the arrangement where this happens, with an editing layer sitting on top of a template layer, updated independently, by different people, on different schedules.

The same data shows WordPress passing mobile Core Web Vitals on 45% of sites against 85% for Duda. That gap is not the platform being slow. It is what years of additive edits in the top layer accumulate to.

Making a fix that holds

The unglamorous part that actually works

Keep a change log. What changed, when, why, and where in the stack.

I keep one for every site I touch and it sounds like bureaucracy until the first time a theme update reverts something and you genuinely cannot remember whether you fixed it, meant to fix it, or dreamt it. On a site where three people make changes, the log is the only thing that distinguishes “this was never done” from “this was done and undone.”

It also makes the revert diagnosable. A dated list next to a theme update date usually identifies the culprit in about a minute.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Technical SEO covers themes, templates and the markup they emit. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

Most of an audit on a headless or React site is the same audit. The questions do not change. Is it crawlable, does it say what it is, is anything removing pages from search. What changes is that half your usual checks return the wrong answer if you run them the way you normally would.

So this is the list of things I do differently, rather than another explanation of what headless means.

Your tools will lie to you by default

Turn on JavaScript rendering in the crawler before you start. Screaming Frog and Sitebulb both crawl raw HTML unless told otherwise. On a client-rendered site that produces a crawl of empty shells with no headings, no content and almost no internal links, and a report that looks catastrophic and is meaningless.

The reverse mistake matters too: once rendering is on, the crawl shows you the rendered page and hides the very gap you are trying to find. Run both. The difference between the two crawls is the finding, and it is the single most useful artefact you can produce on one of these sites.

Expect it to be slow. A rendered crawl is an order of magnitude slower. Budget for it or sample.

There is a reason for that, and it is the same reason Google rations rendering. HTTP Archive’s 2025 Web Almanac puts the median mobile home page at 632 KB of JavaScript against 22 KB of HTML. Rendering means fetching and executing all of it, per URL, and neither your crawler nor Googlebot does that for free.

The head is assembled at runtime

On a traditional site the <head> is written by a template and you can read it. On a headless build it is often composed by a component, which means it can differ by route, by data state, and occasionally by whether an API call succeeded.

What I check: the canonical, the title, meta robots and hreflang, in the raw response, on at least one page of each route type. Not one page. One of each type, because the composition logic is per-route and the bugs are per-route.

The 2025 Web Almanac found an invalid element inside the <head> on 10.3% of mobile pages, most often a stray <img> or <div>. That matters more here than anywhere, because the head closes at the first invalid element, and a head assembled from components is exactly where a stray wrapper div gets introduced by someone who has never thought about it.

Routing questions that do not exist elsewhere

Who you are actually briefing

The thing that most changes the engagement is not technical. On a WordPress site the person implementing is often the person you are talking to. On a headless build it is a development team with a sprint cadence, a backlog and their own priorities.

That changes what a good recommendation looks like. “Add a canonical” is not a task. “The canonical is composed in the SEO component and returns undefined on the product route when the API response lacks a slug” is a ticket someone can pick up. Three well-specified tickets beat sixty findings, and on these projects that is not a preference, it is the difference between shipping and not.

What is usually fine

Worth saying, because these audits attract alarmism. Google renders JavaScript and has for years. A well-built Next or Nuxt site with server-side rendering has no inherent disadvantage, and some of them are the fastest sites I look at.

The problems are not caused by the framework. They are caused by rendering decisions made by people who were not thinking about crawlers, which is a reasonable thing for a front-end developer not to be thinking about.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Technical SEO covers rendering, routing and the crawl layer. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.

There are two versions of every page: the one your server sends, and the one that exists after JavaScript has run. Google reads both, but not at the same time and not with the same certainty, and almost every rendering problem I find lives in the gap between them.

The practical version of this: view the raw source, not the inspector. Your browser’s element inspector shows you the rendered DOM, the page after everything has executed. It will happily show you a canonical tag, a meta description or an H1 that was not in what the server sent.

The gap is measurable

HTTP Archive’s 2025 Web Almanac found a canonical tag in the raw HTML of 64.3% of mobile pages and in the rendered HTML of 66.1%.

That difference, call it two pages in a hundred, is sites where a directive Google uses to decide which URL to index does not exist until a script puts it there. It usually works. Google renders. But you have taken something that could have been a static line in the <head> and made it conditional on a script executing correctly, on a crawler’s schedule, with a rendering budget attached.

When it fails there is no error anywhere. You get a page canonicalising to nothing and you find out months later.

Why the second pass matters

Google crawls the HTML first, and renders later. Sometimes seconds later, sometimes considerably longer, depending on how much it wants your pages. Anything only present after rendering is discovered on that second visit, if it happens.

Which produces a specific and common symptom: the page is indexed but thin. Google took the raw HTML, found a shell, and indexed that. The content arrived afterwards and may or may not have been picked up. The page ranks for nothing in particular and the owner cannot work out why, because when they look at it, it is full of content.

The same applies to links. A link that only exists after JavaScript runs is a link Google may not follow on the first pass. If your entire navigation is client-side rendered, your internal linking is conditional.

Headings behave the same way, and that is the version I find most often. A page whose heading structure is assembled by a component can have a perfect outline in the inspector and nothing but a shell in what the server sent. The audit finds no H1, the owner opens the page and sees one, and everybody concludes the tool is broken.

The scale of what has to be fetched

The Almanac’s page weight data, from 16.2 million sites, puts the median mobile home page at 632 KB of JavaScript against 22 KB of HTML. Twenty-eight times more script than document.

That is the modern web and it is not automatically a fault. But it explains the crawl stats I keep finding: sites where the majority of Googlebot’s requests are scripts and stylesheets rather than pages. Google is not refusing to index the content so much as never getting a clear run at it.

Checking yours, in about five minutes

view-source: and Ctrl+F. Look for your H1, your canonical, your main body copy and your primary navigation links. If they are not in what the server sent, they are conditional.

URL Inspection in Search Console, then “View crawled page”. This is Google telling you what it actually got. It is the closest thing to ground truth available and it is free.

Disable JavaScript and reload. Crude, fast, and it shows you the shape of the shell.

Compare word counts. Raw versus rendered. A large gap tells you where the content is coming from.

What I would actually change

Not “stop using JavaScript.” That is not advice, it is nostalgia.

The line I would draw: if Google needs it to decide what the page is, send it. If a person needs it to use the page, render it.

The short version

---

If a site should be ranking and it isn’t, that’s the work I do. Technical SEO covers crawling, rendering and indexing. If you’re not sure which of several plausible problems is costing you, that’s what an SEO audit is for.