Practice Area Pages That Actually Rank: 12 Winning Pages, Measured

A practice-area page is the one that has to win “atlanta car accident lawyer” or “phoenix dui lawyer” and then turn the click into a call. It is the most commercially important page on a law firm website and the hardest to write well.

There is plenty of advice about them. What follows instead is what the pages currently winning those searches actually contain, measured, plus the two things on nearly all of them that stopped working this year.

The shape of a page that wins

Start with the dimensions, because they settle the brief before anyone writes a word.

Three of those numbers are worth more than the others.

The heading count, not the word count. Sixteen headings across 2,800 words means a new heading every 175 words. These pages are built to be scanned by someone who is upset, not read start to finish. A 2,800-word page with four headings is a wall, and one of the pages measured is exactly that.

A third of headings phrased as a question. This is the single clearest pattern in the set and the one most guides miss. The pages winning are organized around what the visitor is asking, not around what the firm offers. A few run to half their headings as questions.

Calls to action around three per thousand words. The heaviest page in the set runs almost twelve, which is a call to action every eighty-five words. That is not persuasion, it is interruption, and it is worth knowing the winning pages mostly do not do it.

What these pages answer, and what they skip

This is where the opportunity is, because the pattern is lopsided. The pages are good at the firm’s story and thin on the visitor’s questions.

Read the bottom of that table again, because it is the whole argument for writing a better page.

Only half name the filing deadline. It is the most consequential fact on the page. A visitor who leaves without learning there is a clock may lose the claim entirely, and the firm that told them has done something no competitor did. Deadlines differ by state and by claim, so the figure belongs on the page with its jurisdiction attached rather than as a general statement.

One page in four addresses whether the visitor has a case. That is the question they arrived with. Everything else on the page is downstream of it. A page that opens by describing the firm’s thirty years of experience has answered a question nobody asked.

One page in four tells the visitor what to do right now. Say nothing to the other driver’s insurer. Photograph the scene. Get the medical record. Those instructions are free to give, immediately useful, and they establish competence faster than any list of credentials.

The readability problem nobody mentions

Here is the finding that surprised the measurement most.

These pages score a median of 44 on the Flesch reading-ease scale, and the hardest in the set scores 19. Only one clears 60, which is roughly where plain English begins.

For comparison, a score in the 60s is conversational prose a general audience reads comfortably. The 40s is where academic and professional writing sits. Under 30 is dense technical material.

So the pages winning searches made by people in acute distress are written at the level of a professional journal. Two things follow, and they point in opposite directions.

The first is that readability is evidently not a ranking requirement. These pages rank while being hard to read, so anyone claiming a plain-English rewrite will lift rankings is overstating it.

The second is that ranking and converting are different jobs. A frightened person who cannot follow your page calls the next firm, and the ranking bought nothing. This is the cheapest available advantage on a practice-area page, because the field has left it on the floor.

Two pieces of standard markup that stopped working

Every page measured carries structured data, which is the one area where the field is thorough. Two of the common choices are now actively wrong, and both are checkable against Google’s own documentation.

FAQ markup earns nothing now

More than half the pages measured include FAQPage markup, and Google has retired the feature it was for.

Its documentation is explicit: the FAQ rich result “will no longer appear in Google Search starting May 7, 2026.” Google removed the FAQ documentation outright that June, and FAQ no longer appears anywhere in its list of supported structured-data features.

Note what this does and does not mean. The markup is not harmful and it does not need stripping out in a hurry. It simply earns no result, so any vendor still selling FAQ schema as a route to extra space in the results is selling something Google switched off.

One wrinkle worth getting right. Google still supports a feature called Q&A, which is a different thing: it is for pages where users submit competing answers to a question, like a forum thread. A list of questions the firm wrote and answered itself is not that.

Marking up your own reviews is a risk, not a win

Half the pages measured carry AggregateRating markup, usually alongside LegalService. That combination is the one Google’s review-snippet guidelines rule out.

The guidelines are direct about self-serving reviews. Pages using “LocalBusiness or any other type of Organization structured data are ineligible” for the star feature where the reviewed entity “controls the reviews about itself.” The example Google gives is exactly this case: “a review about entity A is placed on the website of entity A.”

LocalBusiness review markup is permitted “only for sites that capture reviews about other local businesses.” A law firm page rating the law firm is not that, and LegalService is a subtype of LocalBusiness, so the rule reaches it.

The consequence is worse than getting nothing. If a site breaches the guidelines, “Google may take manual action against it,” and when that happens “the structured data on the page will be ignored.” Not the rating. All of it, including the markup that was working.

One thing in the markup is easy to get backwards, so it is worth stating plainly. Firms reach for LegalService far more often than LocalBusiness, and that is the right call, because LegalService is the better fit for a practice. The type to avoid is the one actually named Attorney. Despite the name it is not a type for a person at all. It sits under LocalBusiness as a kind of business, and schema.org marks it deprecated on the grounds that “LegalService is more inclusive and less ambiguous.” So the pages using a plain Person entry for a named lawyer have it right. A vendor offering to add Attorney markup to an attorney bio page is putting a retired business type on a page about a human being.

Settlement figures, which fewer pages carry than you would expect

Under half the measured pages name a settlement or verdict figure, which is lower than the reputation of this market suggests.

Part of that is practice area. A bankruptcy or immigration page has no settlement to report, so the absence is structural rather than a choice. But among the injury pages, where a recovery figure is the obvious proof, several still do not publish one.

The likely reason is the rule rather than the reticence. Publishing case results carries requirements that differ by state. Texas requires the amount the client actually received to be as prominent as the verdict. Louisiana requires the associated attorney fees in the same ad. Florida requires the figure be objectively verifiable.

Which makes this a decision to take deliberately rather than a box to fill. If you publish figures, publish the disclaimer your state requires at the prominence it requires, and keep the underlying verification. If that is more work than the page is worth, describe the kind of outcome without the number. That carries no disclaimer obligation, because it makes no claim.

Internal links, and the one thing almost nobody cites

The measured pages carry a median of about 115 internal links, with the heaviest at nearly 900. Much of that is navigation rather than editorial, so treat it as an upper bound on effort rather than a target.

The editorially interesting number is elsewhere. Just over half the pages link to any authority at all: a statute, a court, a government agency, a safety body. Nearly half cite nothing.

That is a strange omission on a page whose entire claim is expertise. A sentence citing the actual statute, with a link to it, does three things at once. It proves the claim and gives the visitor somewhere real to verify it. It is also exactly the kind of passage a search engine or an assistant can lift and attribute.

One more thing the internal-link figures hint at. The pages with the largest counts are not linking editorially; they are carrying a mega-menu that lists every practice and every city. That inflates the number without helping a reader, and it spreads the page’s own authority thinly across hundreds of destinations. The links worth counting are the ones inside the copy, placed where a reader would want them. The related claim type, the attorney who handles this work, the article answering the question the section just raised.

The page template the measurement implies

Put the findings together and a structure falls out. This is not a creative exercise; it is the shape of the pages that win, with their gaps filled.

• 
  Open with whether they have a case, not with the firmThree pages in four skip this, and it is the question that brought the visitor. Two or three sentences on what generally makes a claim viable in this area, then the honest caveat that it depends on facts you would need to hear. That is a stronger opening than any credential.
• 
  Put the deadline high, with its jurisdiction namedHalf the field omits it. Name the limitations period for this claim in this state, cite the statute, and link to it. If the visitor reads one thing and leaves, this is the one worth them having.
• 
  Say what to do in the next 48 hoursAlso missing from three pages in four. A short list of concrete instructions, including the things not to do. This is the section a visitor screenshots and sends to a relative, which is distribution nobody is competing for.
• 
  Answer cost plainly, since almost everyone else doesThis is table stakes rather than an advantage. A contingency percentage, an hourly range, or the honest reasons it varies. A page that routes the cost question to a phone call is now the unusual one, and not in a good way.
• 
  Then the process, in order, with a heading per stageMost pages cover this and it is what gets the heading count up where it belongs. Filing, investigation, negotiation, and what happens if it does not settle. A heading every couple of hundred words, phrased the way a client would ask.
• 
  Keep calls to action to about three per thousand wordsThat is roughly one per major section. It is what the winning pages do, and the page running one every eighty-five words is the outlier rather than the model.

Two structural notes on that template. Phrase roughly a third of the headings as questions, which is what the ranking pages do and what makes the page quotable. And run the finished draft through a readability check, because the field’s median is 44 and clearing 55 would make your page the easiest to read in the search results.

Auditing a page you already have

Most firms reading this have the pages already and want to know which to fix first. The measurement gives an order, cheapest first.

Open the page and count the headings. Fewer than one per 200 words and that is the first fix. It needs no new research, only breaking the existing copy up and phrasing the new headings as questions.

Then search the page for a year. If the filing deadline is not on it, add it with the statute cited, because half the field is missing the most consequential fact on the page.

Then run the readability check. Under 40 means the copy needs a pass for sentence length and plain words, which is the change most likely to lift calls without touching rankings.

Then view the page source and search for AggregateRating. If it is there, remove it, because it earns no stars and risks the rest of the markup. FAQPage can stay or go as convenience dictates, since it now does nothing either way.

That sequence costs nothing but attention and it addresses, in order, the four things the winning pages most often get wrong.

What changes for a city version of the page

Most firms building depth are building a page per practice per city, which raises a question the template does not answer: what actually differs between them.

The honest answer is less than vendors imply and more than nothing. The law is the same across a state, so the deadline, the elements and the process do not change between two cities in it. Swapping the city name through an otherwise identical page produces exactly the thin duplication that makes these builds fail.

What genuinely differs is local and checkable. Which courthouse hears the case and where it is. The local filing quirks. The intersections or facilities that generate these claims in that city. Whether the firm has an office there, and who works in it. If a city page cannot carry two or three paragraphs of that, it should not exist. That is the same test that decides which law firm website pages are worth building at all.

Common Questions

Where these figures come from

The pages were collected in October 2026 by running the commercial searches a firm competes for. Five practice areas across five metros: car accident in Atlanta, divorce in Chicago, DUI in Phoenix, immigration in Houston, and Chapter 7 bankruptcy in Philadelphia. Only pages on a firm’s own site were used, because the question is what a firm publishes. Directories and aggregators were never in the list.

Sixteen were collected and twelve measured. One returned a 404 and one rendered its content in JavaScript, reading as 194 words. Two more were byte-identical duplicates of pages already in the set, which a checksum caught rather than the eye. Dropping duplicates matters more than it sounds, because counting one page twice would have shifted every figure.

The counts behind the plain-language columns above, out of the twelve pages measured:

Readability is the Flesch reading-ease score, computed on the visible body text, which is the same measure and the same implementation used to grade the articles on this site. That makes the comparison like for like rather than a claim about someone else’s writing against a standard this site does not meet.

The Google positions are quoted from its own developer documentation: the FAQ deprecation notice, the list of currently supported structured-data features, and the review-snippet guidelines. All three are linked below and all three are short.

Two limits worth stating. Measuring fetched HTML means a page assembled in the browser reads shorter than it looks, which is why the one JavaScript-rendered page is excluded rather than counted as thin. And twelve pages across five practice areas is a small sample taken on one day; it is enough to show a pattern and not enough to settle a number.

AttorneyPlot builds websites for law firms and writes practice-area pages. So a page arguing that most of them are hard to read and badly structured is a page arguing for its own services. The measurement is reproducible: the searches are named above, and anyone can run a readability check in about two minutes.

1. Google Search Central — the FAQ rich result deprecation notice
2. Google Search Central — structured data features currently supported
3. Google Search Central — review snippets and self-serving reviews
4. Google Search Central — local business structured data
5. Google Search Central — structured data general guidelines
6. Google Search Central — creating helpful, people-first content
7. Schema.org — LegalService, a subtype of LocalBusiness
8. Schema.org — Attorney, a deprecated business type
9. Schema.org — Person, the correct type for a named lawyer
10. Cornell Legal Information Institute — statute of limitations
11. American Bar Association — 2023 Websites & Marketing TechReport
12. Clio — 2025 Legal Trends Report
13. W3C — the Web Content Accessibility Guidelines