Prepared by PI Marketing. Observed 2026-11-03; scored by AI agent sessions under a written rubric, not yet re-scored by a person.

How were the 30 pages chosen?

The pages came from 20 searches for “birth injury lawyer” plus a US metro or state, run once on 2026-11-03 in one search tool (not Google), organic results only: 3 multi-market firms plus 27 drawn in a fixed order, one page per firm. It is not a ranking or a compliance audit; no firm is named or was contacted.

These counts are preliminary. On 15 of the 30 pages, most items were read from page source (the HTML the server sends), not from a phone render (the page as drawn on an emulated phone).

A range in parentheses shows how far a borderline, disputed or source-only score could move a count (full table on the method page).

Where should a firm start?

How many of 30 birth injury practice pages met each yes-or-no criterion, observed 2026-11-03 Horizontal bars, one per yes-or-no criterion, grouped by intake, content, trust and technical. Bar length is the count over its n. Intake: phone number or labelled call button on the first phone screen: 22 of 30; tap-to-call link: 29 of 30; qualifying contact form: 29 of 30; chat launcher visible on a phone: 13 of 30; after-hours promise: 15 of 30; Spanish option: 15 of 30; specific response-time promise: 3 of 30. Content: links to 3 or more condition pages: 7 of 30; cites a qualifying medical source: 10 of 30; names an author or reviewer in a byline: 4 of 30; visible labelled date: 5 of 30. Trust: attorney advertising label: 9 of 30; past-results statement within one phone screen of every dollar-results block: 1 of 17 pages with scorable dollar results; asserts a guarantee (fewer is better): 1 of 30; privacy link near the form: 6 of 29 pages with a form. Technical: HTTPS with http:// redirected: 24 of 30; viewport tag and no sideways scroll at 390 px: 26 of 30; LegalService or Attorney structured data: 20 of 30; robots.txt and a reachable sitemap: 30 of 30. n = 30 pages, one per firm, observed 2026-11-03. Preliminary counts: the first scorer’s, except the first-screen phone, chat and sideways-scroll items on 15 pages, which come from a same-day render pass, and the past-results item, which was re-scored on all 30 pages by both scorers. Intake Phone on first screen 22 of 30 Tap-to-call link 29 of 30 Contact form 29 of 30 Chat launcher shown 13 of 30 After-hours promise 15 of 30 Spanish option 15 of 30 Response-time promise 3 of 30 Content 3+ condition pages 7 of 30 Medical source cited 10 of 30 Author or reviewer 4 of 30 Visible labelled date 5 of 30 Trust Advertising label 9 of 30 Statement at every dollar-results block 1 of 17 Asserts a guarantee (fewer is better) 1 of 30 Privacy link by form 6 of 29 Technical HTTPS with redirect 24 of 30 No sideways scroll 26 of 30 Legal structured data 20 of 30 robots.txt + sitemap 30 of 30 n = 30 pages, one per firm, observed 2026-11-03. Bar length shows x of n; n is 30 unless stated. Preliminary counts. Not a ranking.
Pages meeting each yes-or-no criterion. n = 30 pages, observed 2026-11-03; past results out of 17, privacy link out of 29. Preliminary counts. Method.

The order starts with a family’s first contact, not with measured impact.

  1. Treat phone basics as the minimum. Put the number, or a “Call” button that dials, on the first screen. If you use call tracking, check that the tracking number shown dials.
  2. Publish a response time only if intake keeps it. Measure answer and callback times first.
  3. Put a past-results statement next to each set of results. Ask counsel whether your state requires one, and its wording.
  4. Show who reviewed the page, and when. Display an update date stored in your site’s code only if it marks a real content review.
  5. Link the conditions your copy already names. Then have your web team confirm the technical checks.

No count here shows a change’s effect on signed cases; measure that across your birth injury marketing funnel with the cost per signed case calculator. Ask any agency how it will verify changes on a phone (see evaluating an agency).

To check your own page, load it on a phone, wait 5 seconds and look before you scroll or tap. Then work through the 20 checks below.

Intake: can a family reach you, and do you say when you will respond?

Intake self-check: 1 to 7

  1. Phone on the first screen. After 5 seconds, unscrolled, is your number visible, or a “Call” button that dials?
  2. Tap-to-call. Does tapping the number open the dialer? That takes a tel: link, code that makes a number dial.
  3. Form. Is a contact form visible at phone width? How many fields show in its first step, consent boxes included?
  4. Chat. Does a chat button appear within 15 seconds, with no scroll or tap?
  5. After hours. Does the page say a family can reach you outside business hours, using a listed phrase such as “24/7”? Is it true?
  6. Spanish. Is there a visible Spanish link, “Se habla español” text, or a Spanish version declared in the code?
  7. Response time. Does the page say how soon a family will hear back? Can intake keep that?

Intake results

29 of 30 pages had a tap-to-call link, and 22 of 30 showed the number, or a “Call” button that dials, on the first screen; 4 of the other 8 showed only a phone icon.

3 of 30 said when the firm would respond (range: 2 to 4). 15 of 30 promised after-hours availability (range: up to 16). 15 of 30 offered a Spanish option (range: down to 13).

13 of 30 showed a chat button within 15 seconds (range: up to 15). 8 more loaded chat code but showed no button in that time; 3 of those waited for a scroll or pointer movement. Ask your chat vendor what triggers the button on a phone.

Presence is not performance. A promise on the page does not show that anyone answers or keeps it; see the intake capacity planner and the intake capacity ceiling.

29 of 30 had a contact form (range: 28 to 29), with a median of 6 fields in its first step. Leaving out one multi-step form and one form hidden on phones, forms showed 4 to 9 fields. Fewer fields was not scored as better. See also the lead quality scorecard and intake optimization.

Visible fields in the contact forms on 29 birth injury pages, observed 2026-11-03 Column chart of visible fields in the first step of each page’s contact form. 0 fields: 1 page; 1 field: 1 page; 2 fields: 0 pages; 3 fields: 0 pages; 4 fields: 5 pages; 5 fields: 4 pages; 6 fields: 8 pages; 7 fields: 4 pages; 8 fields: 3 pages; 9 fields: 3 pages. The 0 is one form hidden at a 390 px phone width; the 1 is one multi-step form, first step only. Median 6. n = 29 of 30 pages had a form; 1 page had none. Observed 2026-11-03. Visibility at phone width was checked on the 15 pages rendered during scoring. Preliminary first-scorer counts. 1 0 1 1 2 3 5 4 4 5 8 6 4 7 3 8 3 9 median Visible fields in the form’s first step 0: one form hidden at phone width 1: one multi-step form, first step Bars count pages; n = 29 of 30 pages had a form, and 1 had none (not shown). Observed 2026-11-03. Phone-width visibility checked on 15 pages. Preliminary: first scorer’s counts.
Visible fields in each form’s first step. n = 29 pages with a form, observed 2026-11-03. Preliminary counts. Method.

Content: does the page show who checked it, when, and what it relies on?

Content self-check: 8 to 11

  1. Condition pages. How many of the 10 conditions in the chart below link to their own page on your site?
  2. Medical sources. Does the main content cite a qualifying source, such as a peer-reviewed journal or a children’s hospital, linked or named with an article title or year?
  3. Author or reviewer. Does visible text name who wrote or reviewed the page, with any medical credential?
  4. Date. Is a month and year shown with a label such as “Updated”? A copyright year does not count.

Content results

None of these items assesses accuracy. 16 of 30 pages linked no dedicated condition page from the list of 10; 7 of 30 linked 3 or more.

Dedicated condition pages linked from 30 birth injury pages, observed 2026-11-03 Two panels. (a) Column chart of how many of a fixed list of 10 conditions each page links to a dedicated page for: 0 linked on 16 pages; 1 linked on 5 pages; 2 linked on 2 pages; 3 linked on 1 page; 4 linked on 2 pages; 5 linked on 2 pages; 6 linked on 1 page; 7 linked on 1 page. 3 or more: 7 of 30. (b) Bars for how many pages link each condition: Cerebral palsy 12 of 30; Erb’s palsy 7 of 30; HIE, birth asphyxia or oxygen deprivation 6 of 30; Brachial plexus injury 6 of 30; Shoulder dystocia 5 of 30; Kernicterus or jaundice 5 of 30; Brain bleed 1 of 30; PVL 1 of 30; Neonatal seizures 0 of 30; Skull fracture or cephalohematoma 0 of 30. n = 30 pages, observed 2026-11-03. Only links to the firm’s own condition pages count. Preliminary first-scorer counts. (a) Condition pages linked per page Distinct conditions, from a fixed list of 10 16 0 5 1 2 2 1 3 2 4 2 5 1 6 1 7 3 or more: 7 of 30 Conditions linked; bars count pages, of 30 (b) Pages linking each condition Cerebral palsy 12 of 30 Erb’s palsy 7 of 30 HIE or asphyxia 6 of 30 Brachial plexus 6 of 30 Shoulder dystocia 5 of 30 Kernicterus/jaundice 5 of 30 Brain bleed 1 of 30 PVL 1 of 30 Neonatal seizures 0 of 30 Skull fracture 0 of 30 n = 30 pages, observed 2026-11-03. Counts only links to the firm’s own pages. Preliminary: first scorer’s counts.
Condition pages linked per page, and pages linking each condition. n = 30 pages, observed 2026-11-03. Preliminary counts. Method.

A condition page should explain, not accuse: a diagnosis such as cerebral palsy is not by itself evidence that anyone was negligent.

  • Medical sources. 10 of 30 cited one (range: up to 11); 4 of the 10 relied on one link to the same children’s hospital health library.
  • Bylines. 4 of 30 named an author or reviewer, all attorneys (3 legal reviewers, 1 author); none named a medical reviewer. 4 more named an author only in structured data, code that describes a page to search engines and AI assistants, unseen by readers.
  • Dates. 5 of 30 showed a labelled date (range: down to 4). Yet 18 of 30 stored an update date in structured data, and 15 of those showed none. 9 of the 15 were dated 2026.
Visible and machine-readable dates on 30 birth injury pages, observed 2026-11-03 Unit chart, one square per page. 5 of 30 pages show a visible labelled date. 18 of 30 carry a machine-readable dateModified in structured data. 15 of 30 carry a machine-readable date but show no visible one, and 9 of those 15 are dated 2026. n = 30 pages, observed 2026-11-03. dateModified was recorded, not scored. Preliminary first-scorer counts. Visible labelled date 5 of 30 Date in structured data 18 of 30 In structured data, not visible 15 of 30 Of those 15, dated 2026 9 of 15 n = 30 pages, observed 2026-11-03. Structured data: JSON-LD dateModified, recorded but not scored. Preliminary counts.
Visible dates and stored update dates; stored dates were not scored. n = 30 pages, observed 2026-11-03. Preliminary counts. Method.

See also birth injury SEO and AI assistants in attorney discovery.

Trust: where do results disclaimers and advertising labels sit?

These checks record placement, not compliance. Attorney advertising rules vary by state, so ask your own counsel.

Trust self-check: 12 to 15

  1. Advertising label. Is attorney advertising wording on the page itself, not only behind a disclaimer link?
  2. Past-results statement. Is a statement such as “past results do not guarantee a similar outcome” visible within one phone screen of each set of dollar results, with nothing tapped open?
  3. Guarantee language. Search the page for “guarant”. Does any use assert a guarantee rather than deny one?
  4. Privacy link. Is your own privacy policy linked in the form or just below the submit button?

Trust results

29 of 30 pages loaded for this check; 18 of the 29 showed a dollar result at phone width, and 17 of those could be scored. One phone screen is 844 CSS pixels tall at a 390-pixel width.

1 of 17 put a statement within one phone screen of every set of results. Of the other 16, 2 covered only some sets, 3 placed it about 5 to 11 phone screens away, 5 only in the footer or on a linked disclaimer page, and 6 had none on the page or any linked disclaimer or terms page.

If one page whose results appeared under only one scroll method drops out, the count is 1 of 16 (range table). No disputed score changes the 1.

Where 30 birth injury pages placed a past-results statement near dollar results, observed 2026-11-03 Unit chart, one square per page. Of 30 pages, 29 loaded when the item was re-scored (1 returned a server error). 18 of those 29 showed a visible dollar result at a 390 px phone width (11 showed none). 17 of those 18 could be scored (1 linked disclaimer page never rendered its text). Of the 17: 1 placed a past-results statement within one phone screen of every results block; 2 placed one next to some blocks but not others; 3 had one on the page but more than one screen from every block; 5 had one only in the footer or on a linked disclaimer page; 6 had none on the page or on any linked disclaimer or terms page. Observed 2026-11-03. Placement observed, not a compliance finding; rules vary by state. Preliminary counts. Pages observed 30 pages Loaded at the re-score 29 of 30 Dollar result at phone width 18 of 29 Could be scored 17 of 18 Past-results statement, 17 scored pages Within one screen of every block 1 of 17 Next to some blocks, not others 2 of 17 On the page, over one screen away 3 of 17 Footer or linked page only 5 of 17 None on the page or linked pages 6 of 17 n = 30 pages, observed and re-scored 2026-11-03. One screen = 844 px at a 390 px phone width. Placement observed; not a compliance finding. Rules vary by state. Preliminary counts.
Where pages placed a past-results statement. n = 30 pages, re-scored 2026-11-03. Placement, not compliance. Preliminary counts. Method.
  • Advertising labels. 9 of 30 carried attorney advertising wording (range: down to 8), all in the footer. 14 of the other 21 only linked to a disclaimer page.
  • Guarantees. 1 of 30 asserted a guarantee, about privacy, not an outcome or a fee. 9 of 30 used the word only to deny one.
  • Privacy near the form. 6 of 29 pages with a form put a privacy link in or just below it (range: down to 4); one of the 6 linked a CAPTCHA provider’s policy, so 5 of 29 linked the firm’s own.

For disclosures from ad to landing page, see compliant ad creative.

Technical basics: what should your web team confirm?

A viewport tag tells a phone to fit the page to its screen. A robots.txt file tells search crawlers what they may read; a sitemap lists the pages a site wants found.

Technical self-check: 16 to 20

  1. HTTPS with redirect. Have your web team request the http:// address outside a browser (browsers upgrade it). Does the server redirect to https://?
  2. No sideways scroll. On a phone, does the page stay put when swiped sideways? Does its code carry a viewport tag?
  3. Document size. In a desktop browser’s developer tools, open the Network tab and reload. What is the document row’s uncompressed size?
  4. Structured data. Does a structured-data testing tool find the LegalService type, or the older Attorney type?
  5. robots.txt and sitemap. Do both load?

Technical results

30 of 30 pages loaded over HTTPS and had robots.txt and a reachable sitemap. 24 of 30 also redirected the http:// address. 26 of 30 had a viewport tag and no sideways scroll at a 390-pixel phone width (range: up to 27). 20 of 30 carried LegalService or Attorney structured data (range: up to 25). 5 of the 10 without it were checked in source only, which misses markup added after loading. The HTML document alone had a median of 292,098 bytes, a proxy for page weight, not speed.

HTML document size of 30 birth injury pages, observed 2026-11-03 Strip plot, one dot per page, of the HTML document size in bytes for 30 pages. Median 292,098 bytes; middle half 163,598 to 497,803; range 48,068 to 1,102,830. 7 pages are over 500,000 bytes and 3 under 100,000. HTML document only, uncompressed; a proxy, not page speed. n = 30 pages, observed 2026-11-03. First scorer’s fetch, preliminary. median 292,098 bytes middle half 163,598 to 497,803 0 250K 500K 750K 1M HTML document size, bytes (K = 1,000) n = 30 pages, observed 2026-11-03. HTML document only; a proxy, not page speed. Range 48,068 to 1,102,830 bytes. Preliminary.
HTML document size, one dot per page. n = 30 pages, observed 2026-11-03. Preliminary counts. Method.

How far can these counts be trusted?

The sample leans toward firms ranking in one search tool’s organic results, not Google’s, with no ads or map listings. Firms whose sites blocked a plain fetch were excluded (4 in the draw, 1 more checked during design). Every value reflects one day. 9 of the 15 pages read mostly from source were outside the second scorer’s 10-page check; on them, any text item could fall outside the ranges, especially form visibility, the Spanish option and the privacy link. One page moves a count out of 30 by about 3 percentage points, and the past-results count out of 17 by about 6: treat a one- or two-page difference as noise.

  • Who scored. AI agent sessions, following the written rubric; no person has re-scored any page. The first scorer rendered 15 pages and read 15 from source, with a same-day render for the first-screen phone, chat and sideways-scroll items.
  • Agreement. A second scorer rendered and scored 10 pages on all 20 criteria, and 192 of 200 score pairs agreed; two AI sessions can share blind spots. None of the 10 passed the response-time, byline or date checks, so those passes were not cross-checked.
  • Past results. This item fell below the method’s 90% agreement bar, so it was rewritten and both scorers re-scored all 30 pages from their own renders. They now agree on 29 of 30; a third reviewer, also an AI agent session, settled the one difference.
  • Status. Four rulings (form length, Spanish option, medical sources, privacy link) are pending; any change will be logged under corrections.
  • Approval. Spencer Moeller, PI Marketing’s founder, approved the article outline on 2026-11-03 and its publication on 2026-11-04.

Questions about the benchmark

What did 30 birth injury law firm practice pages show?

On 2026-11-03, 29 of 30 pages, one per firm, had a tap-to-call link, but 3 of 30 said when the firm would respond. The pages came from one search tool’s organic results, not Google. Counts are preliminary.

What should a birth injury law firm’s practice page include?

Among 30 pages from one search tool’s organic results (not Google), observed 2026-11-03, most had tap-to-call (29 of 30) and a contact form (29 of 30). Few said when the firm would respond (3 of 30), named an author or reviewer (4 of 30), showed a labelled date (5 of 30) or put a past-results statement within one phone screen of every set of results (1 of 17). Counts are preliminary; ask your own counsel about disclosures.

How can a firm check its own birth injury page?

Start on a phone: load the page, wait 5 seconds and look before scrolling or tapping. Then run the 20 checks in four groups (intake, content, trust, technical), and ask your own counsel about the trust items. The comparison counts, from 30 pages observed 2026-11-03, are preliminary.

Do birth injury law firm websites put a disclaimer next to case results?

Rarely, among 30 pages from one search tool’s organic results, not Google. On 2026-11-03, of the 17 pages showing dollar results that could be scored, 1 put a past-results statement within one phone screen of every set of results; 6 of 17 had none on the page or its linked disclaimer and terms pages. This records placement, not compliance; rules vary by state, so ask your own counsel.

Are these numbers final, and who scored them?

No, they are preliminary. Every scorer was an AI agent session following a written rubric; no person has re-scored any page. A second scorer agreed on 192 of 200 score pairs across 10 pages; four rulings are pending.