The 9pm Test
The 9pm Test is a dated, reproducible check of what a person who needs a lawyer at 9pm actually finds on a law firm's public pages, scored across seven signals. In September 2026 it was run on 308 personal injury firm websites in Los Angeles and Houston: 281 were readable, the median score was 2 of 7, and no firm scored above 5. The same method works for any business that is reached after hours. It was created by Juan Camilo at Wealth Marketing AI in 2026. Every signal carries its evidence and the date. Nothing is called and nothing is submitted.
Method version 1.0 · Last updated · Run on 308 firm websites across two metros
Free, on your own site, in about thirty seconds.
Why 9pm and not another hour?
Because 9pm is the hour no published study covers and almost no business staffs. Every published test of business responsiveness we could find either ran inside office hours or never says what hour it ran. 9pm is not the busiest hour for calls. It is the hour where the published record is empty and the person who was hurt that morning finally sits down with a phone.
Hennessey Digital's 2025 Lead Form Response Time Study submitted its forms to 1,333 law firms between 10am and 12pm in each firm's own time zone, and says so plainly in its method. Clio's 2024 Legal Trends Report had a third party phone and email 500 US firms and found that "only 40% of firms answered the phone calls" and "48% did not do either, meaning they were essentially unreachable by phone", but it does not publish the hour of day any of that happened. Both are careful pieces of work. Neither can tell you what happens at 9pm.
The only after-hours count anyone has published is a FindLaw audit of 100 firms, on a page last updated in March 2019, which reported that "73 of them didn't have any way to answer the phone after business hours". One audit, seven years old, in English only.
So the after-hours gap is not a secret. It is simply unmeasured, and it has stayed unmeasured because measuring it properly means either paying people to make calls at night or finding something honest to measure instead.
Nor is 9pm the peak. Market My Market's 2022 look at 30,000 calls from its own legal clients found the after-hours peaks clustered right around the edges of the working day, 34.22% between 5pm and 6pm and 28.54% between 8am and 9am, and that roughly 60% of after-hours and weekend callers were calling for the first time. 9pm sits well past that peak on purpose. It is late enough that no reasonable reading of "business hours" covers it, early enough that a person is still awake and searching, and it is the same hour every time, which is what makes results from different cities and different months comparable.
The second reason for 9pm is language. In the Los Angeles metro, 34.4% of people aged 5 and over speak Spanish at home; in Houston, 30.2% (US Census Bureau, American Community Survey 2024 1-year, table S1601). After hours is where a bilingual promise is cheapest to make and hardest to keep, so 9pm is where the gap between the page and the reality is widest.
What are the seven signals?
Seven, because seven is what a person can check from the outside in a quarter of an hour without touching the business.
Each one is pass, fail or unknown. Unknown is a real answer and is never counted as a fail.
-
The Spanish page
A real page written in Spanish, linked from the home page, with Spanish navigation and one consistent voice. Not a translation widget, and not English boilerplate run through a machine.
-
Tap to call
A phone number on that page that dials when you tap it on a phone. Not text, not a script, not an image.
-
The form and its promise
A form on the Spanish page, in Spanish, with copy beside it or on the thank-you page saying in Spanish what happens after submit and how soon.
-
After hours, and who answers
The site says someone takes contact outside business hours and names a Spanish speaker. Both halves, or it fails.
-
The public reply badge
The "typically replies" badge on the business's public Facebook Page, when the Page shows one.
-
Paid acquisition markers
A Google Ads tag, remarketing tag, Meta Pixel, Microsoft UET, Local Services badge markup or a call-tracking script in the page source. A proxy for "buys traffic", not proof of it.
-
Qualifying questions
Whether the form asks anything about the situation itself, when it happened, who was at fault, whether there was treatment, or whether there is already a lawyer, before the enquiry reaches the intake desk.
Signal 5 is the weak one and we say so on every run. Facebook does not put the reply badge in the static page source for most Pages, so in practice it returns unknown far more often than pass or fail. In both metros published below it returned unknown for every single business, which means the highest score anyone could have earned was 6, not 7. The free self-serve tool skips signal 5 altogether rather than guess at it, and reports the other six.
Can I run a 9pm Test myself?
Yes, by hand in about fifteen minutes with a phone and a browser that can view page source. Work through the seven signals in order, write one line of evidence and the date beside each, and total the passes out of seven. The method is published in full on this page so it can be repeated without us.
You need a phone, a browser that can view page source, and the website address. Do not call. Do not submit anything.
-
Find the Spanish page
Open the home page on a phone and look for the Spanish link. Record one of three things: a real Spanish page, a translation widget only, or nothing. Then read the Spanish page aloud. If the voice slides between usted and tú, if the navigation still says Home and Contact, or if a phrase reads as a literal translation, it is translated boilerplate, not a written page. Write one sentence of evidence and the date.
About 2 minutes. Home page, then the Spanish page.
-
Tap the number
Tap the number on that page. It passes only if tapping dials. Plain text, a script-drawn number, or a number set inside an image all fail.
About 1 minute. The same page, on a phone.
-
Read the form, do not submit it
Find the lead form on the Spanish page or the contact page. Do not submit it. It passes only if the form sits on the Spanish page, its labels are in Spanish, and something beside it or on the thank-you page says in Spanish what happens after submit and how soon. A Spanish form with an English promise fails.
About 3 minutes. Spanish page, contact page, thank-you page.
-
Search for the after-hours promise
Search the home and contact pages for 24, horas, after hours and español. It passes only when both halves are true: someone takes contact outside business hours, and a Spanish speaker is named. "Open 24 hours" with nobody named in Spanish is a fail.
About 2 minutes. Home page, contact page.
-
Look for the reply badge
Open the business's public Facebook Page and look for the "typically replies" badge. Pass only if you see it. If there is no badge, record unknown, never fail.
About 1 minute. The public Facebook Page.
-
View the page source
View source on the home page and search for gtag, googleadservices, fbq, connect.facebook.net, uetq, callrail and calltrackingmetrics. Any hit means the business is set up to buy traffic and track calls. If only a tag manager is present, record unknown, because a tag manager can load the rest out of sight.
About 3 minutes. View source on the home page.
-
Read the form fields again
It passes only if the form asks something about the situation: when it happened, who was at fault, whether there was medical treatment, whether there is already a lawyer. Name, email and phone alone is a fail, because every click then lands on the intake desk unfiltered.
About 2 minutes. The best lead form on the site.
-
Score it and date it
Mark each signal pass, fail or unknown, write the date beside each, and total the passes out of seven. Publish the date and the evidence line with every score. A score without a date is not a 9pm Test.
About 1 minute. Your own notes.
Does the 9pm Test call the business or submit its forms?
No. The 9pm Test reads public pages with ordinary GET requests and nothing else. No number is dialled, no form is submitted, no message is sent, no accident or enquiry is invented, nobody poses as a customer, and nothing behind a login is touched. That is the whole point: it costs the business no staff time and it can be repeated by anyone from the outside.
Posing as a prospective client is the single thing business owners name most often as disqualifying, and they are right. The restrictions are the method. Remove them and the result stops being reproducible, starts costing other people's time, and becomes something nobody should publish.
Does the 9pm Test name businesses that fail?
Never. Failures are reported only in aggregate, as counts across a metro. Names appear only beside a pass. A business that wants its own scorecard can have it privately and is free to publish it, but we do not publish a named failure and neither should anyone running the method.
What else it will not do
- No ranking and no league table. A score of 2 is not a verdict on a business. It is a description of what its public pages did on one dated day.
- No claim about what actually happens. This is the public half. It measures what the pages promise, not whether a person answers. Anything interactive would have to be disclosed, dated and counted separately, and it is not part of this method.
- Not advice. Nothing here is legal, medical or financial advice.
May I publish my own 9pm Test results?
Yes. Anyone may run and publish a 9pm Test; please cite the method. Cite it as: The 9pm Test, a method created by Juan Camilo at Wealth Marketing AI, 2026, https://wealthmarketing.ai/9pm-test.
Two conditions, both of which exist to protect the people being measured rather than us: publish the date every signal was checked alongside the score, and name only what passes. A 9pm Test published without dates, or with a named failure, is not one.
Where has the 9pm Test been run?
Los Angeles and Houston, both crawled on 15 September 2026: 308 personal injury firm websites in total, of which 281 were readable. Sixteen of 139 readable Los Angeles sites and 23 of 142 readable Houston sites had a real Spanish page. The median score in both metros was 2 of 7, and no firm in either metro scored above 5.
The numbers below are ours. Method: public pages only, ordinary GET requests with a browser user agent, twelve second timeout, nothing called, nothing submitted, every signal stamped with the date it was read.
| Metro | Crawled | Sites read | Readable | Real Spanish page | Tappable number | After hours and a Spanish speaker named | Median score |
|---|---|---|---|---|---|---|---|
| Los Angeles, CA | 160 | 139 | 16 | 132 | 47 | 2 of 7 | |
| Houston, TX | 148 | 142 | 23 | 132 | 38 | 2 of 7 | |
| Both metros | 308 | 281 | 39 | 264 | 85 | 2 of 7 |
Read that table with three caveats attached, because they change what it means. First, 21 Los Angeles sites and 6 Houston sites could not be read at all (DNS failures, 403s, 500s and bot walls) and are counted as unreachable, never as failures. Second, signal 5 returned unknown for all 308, so nobody could have scored more than 6. Third, no business in either metro scored above 5, and the most common score in Los Angeles was 2 while in Houston it was 2.
The pattern that survives all three caveats is the Spanish one. Across 281 readable sites, 39 had a page actually written in Spanish. Far more had something: 87 had a machine-translated page and 39 had a translation widget and nothing else. That is the gap the test exists to measure, and it is the same gap in both cities.
The full write-ups, with the charts, the breakdowns by signal and a downloadable aggregate CSV, are published at the Los Angeles study and the Houston study.
The two personal injury runs are also set beside 30 sourced industry numbers, each carrying its publisher, its year and a link, in the law firm intake statistics. The same engine, with the signals rewritten for the market, produced the Stairlift Enquiry Study across 596 US mobility and home medical equipment dealer sites, and the Coach Funnel Review across 483 US coach and consultant sites. All of them were read on 15 September 2026, on public pages only.
Which markets does Wealth Marketing AI apply the 9pm Test to?
The method is not specific to law firms. It asks what a person looking for help at 9pm finds on a business's public pages, and that question is the same in every market we work in.
Both published runs so far are personal injury. The other markets are listed here because the same seven signals apply, not because we have published a run in them yet.
- Personal injury law firms, US: can a Spanish-speaking accident victim reach a person who answers in Spanish, after hours?
- Coaches and consultants, US: does the page say what the offer is and what happens after someone books, at the hour a buyer actually reads it?
- Real estate agencies, Dubai and Abu Dhabi: what does a buyer enquiring at 2am see, and does anything answer before the morning?
- Mobility and home medical equipment dealers, US: does the page ask who it is for, which product and what timeline, before the phone rings?
Where does every number on this page come from?
Six sources: one primary, four vendor-published, and our own crawl. All five third-party sources were re-fetched on 15 September 2026, and each is listed below with its publisher, its year and its link.
The Los Angeles and Houston figures were produced on 15 September 2026 by Wealth Marketing AI. Public pages only, ordinary GET requests with a browser user agent, a twelve second timeout, eight concurrent workers across businesses and one request at a time per host. No form was submitted, no number was called, no message was sent, nothing behind a login was read. Every signal carries the URL it was read from and the date it was read. Sites that could not be fetched are counted as unreachable, not as failures.
- Primary. US Census Bureau, American Community Survey 2024 1-year estimates, table S1601 (Language Spoken at Home), 2024. Spanish spoken at home, population 5 years and over: Los Angeles metro 34.4%, Houston metro 30.2%. Table pulled 2026-09-14, endpoint re-checked 2026-09-15. data.census.gov, table S1601
- Vendor-published. Clio, 2024 Legal Trends Report, November 2024, reported by the Illinois Supreme Court Commission on Professionalism. Secret shop of 500 US law firms run by a third party (the write-ups do not state the hour of day): "Only 40% of firms answered the phone calls (down from 56% in 2019)", "Just 33% of firms responded to the emails", "48% did not do either, meaning they were essentially unreachable by phone". Vendor-published but shopped by a third party and re-reported by a bar body. 2civility.org
- Vendor-published. Hennessey Digital, 2025 Lead Form Response Time Study, 2025. 1,333 law firms, forms submitted between 10am and 12pm in each firm's local time zone during Q1 2025, seven day cutoff. Median response time 13 minutes; "Twenty-six percent of law firms don't respond to online leads." Business hours only, which is why it does not answer the 9pm question. hennessey.com
- Vendor-published. FindLaw (Thomson Reuters), Strategist blog, page last updated 21 March 2019. Audit of 100 firms: "73 of them didn't have any way to answer the phone after business hours" and "half of all the firms took more than 24 hours to respond, or simply never responded at all." The underlying audit page no longer resolves and the post shows no original publication date; it is cited here only as the one after-hours count anyone has published. findlaw.com
- Vendor-published. Market My Market, 24 January 2022, using CallRail data from its own legal clients. 30,000 calls over six months: of the after-hours calls, "34.22% of the calls occurred between 5:00 p.m and 6:00 p.m., and 28.54% occurred between 8:00 and 9:00 a.m.", and roughly 60% of after-hours and weekend calls were first-time callers. Its own client base, not a random sample. marketmymarket.com
- Ours. Wealth Marketing AI, The 9pm Test, Los Angeles and Houston, crawled 15 September 2026. 308 personal injury firm websites, 281 readable, public pages only. Method as described above.
Statistics we saw repeated online and deliberately did not use, because no primary source exists for them: "78% hire the first firm that answers", "responding within five minutes lifts conversion roughly 400%", and "42% of leads come after hours". If you find a primary source for any of these, send it and this page will be updated with a dated correction.
Run it on your own site.
Paste your website into the free tool and it returns the seven signals, pass or fail, with the evidence line and the date beside each one, in about thirty seconds. Email optional. If you would rather have it run across a whole market than one site, book a call.