An audit of 149 ABA clinic websites found families can rarely learn how fast a provider will answer. The agency corrected the study during Acuity’s review.
Key Takeaways
- Most ABA clinic websites say nothing about response speed. An audit of 149 clinic sites crawled on July 30 found 9 percent state a bounded response time, 16 percent offer chat or text, and 15 percent allow real booking.
- The agency corrected the study twice during Acuity’s review. Digital Authority Partners removed one organization under its own eligibility rule and extended a scoring rule it had applied to only one of six signals.
- The evidence trail is thinner than the published method claims. The per-clinic page inventory was never retained in the delivered files, so figures for four of the six signals cannot be traced back to specific pages.
- Scale buys families information rather than speed. National operators named a payer on every site while fewer than six in ten independents did, yet independents lead on true booking, 18 percent against 12.
A family that has just received an autism diagnosis arrives at a clinic website with one question the site is rarely built to answer. Not whether the provider is good. Whether anyone will answer.
An automated audit of 149 applied behavior analysis and autism therapy clinic websites, crawled in a single pass on July 30 and released by the healthcare marketing agency Digital Authority Partners, found that 9 percent of those sites state a bounded timeframe for replying to a family inquiry. Sixteen percent offer a chat or text surface a visitor can use immediately. Fifteen percent let a family choose a real appointment slot. Ninety-two percent carry a contact form and 74.5 percent name at least one payer, which the study treats as evidence that those two signals describe the field rather than separate it. No clinic in the sample passed all six of the signals scored.
The finding lands in a market where the front door has become an operating problem rather than a marketing one. The Centers for Disease Control and Prevention’s most recent surveillance estimate, released in April 2025, put autism identification at about 1 in 31 eight-year-olds, up from 1 in 36, with a median age of diagnosis of 47 months. Demand has outrun the workforce trained to meet it, and the result is a queue at every stage. Private equity has moved into the same gap, with researchers at Brown University counting 574 private-equity-owned autism therapy centers across 42 states as of the end of 2024, in a research letter published in JAMA Pediatrics in January.
What the ABA Intake Benchmark Measured, and What It Deliberately Left Out
The study scores six signals, all observable on a public website without contacting anyone: an inquiry form, insurance clarity, true booking, a response-time promise, instant contact, and whether structured data lets a search or answer engine read the intake path. Scoring ran off a published codebook, written by Rhiannon Cecil, the agency’s Head of Content, that draws each line explicitly. A response-time promise passes only where the site commits to a unit of time, so “within 24 hours” and “within 24 to 48 hours” clear the bar and “we respond promptly” does not. Hedged wording passes if the number survives it. Fifteen rejected phrasings sit in the dataset with the reason each one failed.
Instant contact passes on a live chat, a chatbot, or a text-to-connect widget, and fails on a contact form styled to look like chat. Detection ran against vendor signatures in the raw markup rather than page text, because widgets injected by JavaScript rarely appear in extracted copy. A 40-clinic pilot that checked text only returned 5 percent on the same signal.
The quality-assurance record is unusually full for agency research. A second detector, built on deliberately different logic, ran across all 150 clinics rather than a sample, agreeing with the first on 145 of 150 for the response-time promise and 126 of 150 for instant contact. Adjudicating the disagreements by hand surfaced five missed response-time promises and seven missed chat widgets, and both detectors were corrected.
The sample was drawn from a query set of 96 searches across 64 cities in all four Census regions, written down before collection, with ranked “best ABA clinic” roundups excluded on the grounds that such lists select for marketing sophistication. The study says plainly that no clinic was contacted, which is the limitation an operator should read first. A site with no chat widget may still answer the phone in twenty seconds. What the audit measures is a publishing decision.
Cecil and Ally Ipsen, the agency’s Vice President of Marketing, sat down with Acuity in September to discuss the study, both on the record. Why response time carried the headline is a question Ipsen answers from the other side of the form. “As a parent myself, if I am searching for ABA therapy care at 10 o’clock at night, I forget,” she says. “So I need follow-ups. I need someone to call me. I need someone to nurture me back in, because you’re not the only ABA clinic that I’m reaching out to.”
ABA Operator Size and Intake Readiness: Scale Buys Information, Not Speed
The segment results are the part most likely to interest operators and investors, because they cut against the intuition that consolidation fixes the front door. National operators, defined as sixteen or more locations or a presence in four or more states, named a specific payer on every site in the sample. Fewer than six in ten independents did. On machine-readable intake the same gradient holds.
On booking it reverses. Independents let a family select a slot at 18 percent, regional groups at 15, and national operators at 12. Stated response times sit near the floor across all three, at 7, 8 and 12 percent. Instant contact is the one speed signal where scale shows up, at 9, 12 and 35 percent. Read together, the pattern suggests the largest providers in the field have invested in telling families more rather than in answering them faster. The scheduling gap in particular has drawn founders arriving from outside the field, who tend to find the problem less technical than organizational.
That spread is the reason Ipsen gave Acuity for publishing the benchmark at all. “Not every ABA clinic has the same budget that can afford a full digital marketing agency,” she says. “We want to make sure that we’re giving them a running chance to beat those competitors, or at least be on the same track as the bigger ABA therapy centers.” A small operator, she says, generally knows only that it has a lead form.
The vendor data points the same direction. Of the 24 clinics offering instant contact, seven carried the signature of LeadTrap AI, more than any other vendor detected. LeadTrap rebranded to Carelu this month and sells intake automation specifically into ABA, where Chief Executive Yoni Belson says roughly 80 percent of families who inquire qualify for services while only about 10 percent of all inquiries reach the end of intake. Those figures are the company’s own. What the audit adds is an outside count of how thinly the category is deployed: the most common instant-contact vendor across 149 clinics appeared on seven sites.
The machine-readability finding runs on a similar split. Sixty-seven percent of the sample carries valid structured data, but much arrives by default from a website builder or search plugin, so the agency reports it both ways: 66 of the 81 sites carrying an injecting plugin pass, or 81 percent, against 34 of the remaining 68, or 50 percent. Question markup, the format that lets an engine lift an answer and attribute it, sits at 17 percent, and that matters more each quarter as families begin their search through an assistant rather than a results page.
Two Corrections to the ABA Website Study, and a Thinner Evidence Trail
Two changes were made to the study while Acuity was reviewing it, both identified by the agency and both disclosed before publication.
The first is an eligibility removal. Mountaineer Autism Project, a West Virginia organization that reads as an advocacy and referral body rather than a direct provider, sat in the sample as an independent. The study’s own eligibility rule excludes advocacy bodies, and its exclusion log shows a comparable organization removed on exactly that basis. The agency pulled the row. The sample moves from 150 clinics to 149, total passes from 405 to 404, average readiness from 2.70 to 2.71 out of six, and the independent segment from 56 clinics at 2.27 to 55 at 2.29. Acuity recomputed each figure from the row-level file and they reconcile.
The second is a scoring rule that was written once and applied once. The codebook says a signal scores a failure only where its relevant page was retrieved, and anything else is recorded as not assessed and dropped from that signal’s denominator. The agency applied that to the inquiry form, which is why it carries a base of 146 rather than 150, but not to the response-time promise, so four clinics where no contact page came back were counted as failures rather than set aside. The correction extends the rule to response time and true booking, both of which depend on a page the crawler may never have reached. The agency estimates the true base is lower still, nearer 141, once the thinnest crawls come out as a class. The headline survives all of it: 13 passes against 150 is 8.7 percent, against 146 is 8.9, and against 141 is 9.2, and every one rounds to nine.
A third issue is narrower but goes to how much weight the study can carry. The published method states that per-clinic page URLs and raw captures are retained, and that any figure can be traced back to a specific string on a specific page on a specific date. When the agency went looking for the page inventory at Acuity’s request, it was not in the delivered files. It had lived in a working session that resets. The workbook records how many pages came back per clinic, which is where the study’s count of 761 retrieved pages originates, but not which pages they were. For the response-time promise and instant contact, an evidence log carrying a verbatim quote or a vendor signature for every pass preserves the trail. For the other four signals it does not. The agency says its codebook is being revised to describe what the delivered files actually hold.
Crawl Depth and the ABA Intake Data: One Question That Stays Open
Whether crawl depth is depressing the two headline numbers remains unresolved, and it matters because both describe how thinly the field performs. Clinics where the crawler retrieved three or four pages average 2.02 out of six on the composite score, against 3.31 for clinics at seven or eight. Eleven of the 13 response-time passes were found on a contact or intake page, so a crawl that never reached one cannot register the signal. That is the case for a measurement artifact.
The case against it rests on instant contact, which is detected from vendor code sitting on the homepage. Every crawl retrieves a homepage, so depth should not cost a clinic that signal, and the rate still drops from 20 percent in the deepest bucket to 6 percent among clinics at three to four pages. Acuity tested that by isolating only the passes found on a homepage, and the gradient flattens rather than climbing: 33 percent among clinics at one or two pages, 6 percent at three to four, 14 percent at five to six, and 6 percent at seven to eight. The thinnest bucket carries the highest rate in the study, several of them large operators whose sites returned little to the crawler. Pages retrieved is a rough proxy for how built-out a site is, not a clean one. Cecil says the crawler matched internal pages by URL and link text against a list headed by contact, get started, admissions, new patients and intake, which put contact pages first in priority, and in the five clinics whose retrieved pages survive in working files the contact page came back inside the first three. She attributes the thinnest crawls to navigation built in JavaScript, which exposes few links to a crawler reading raw HTML. With 24 passes spread across four buckets, neither reading is distinguishable from noise, and the inventory that would settle the question no longer exists.
What ABA Intake Friction Means for Operators and Buyers
For operators, the number to read is not the composite on its own. The codebook asks that the score out of six be published with the per-signal pass rates alongside it, because the pass rates stay comparable across versions of the study and the composite does not: the 40-clinic pilot that preceded this one scored out of five and never collected the sixth signal. The finding that matters is that three of the four signals a family uses to judge responsiveness sit near the floor across every operator size, which makes them cheap to move. Stating a response time costs nothing but a commitment.
For buyers, intake has already migrated into diligence. Referral conversion, the share of inquiries that become authorized and delivered care, is now a standard line in ABA valuation work, and it is measured against exactly the friction this study describes. Practice management vendors have moved into the same gap, and operators weighing those tools have been advised to ask hard questions before signing.
Digital Authority Partners is a full-service digital marketing agency, working in search optimization, paid media and website builds. Ipsen told Acuity the healthcare practice narrowed onto ABA as the firm took on clients in the space, including one that arrived with two clinics in Chicago and now operates around 17 locations across the Midwest. The study discloses on page 15 that one clinic in the sample is a client, entered through the same random draw as every other row with no score adjusted.
Asked in the interview how she wanted the commercial relationship framed, Cecil, who designed both the codebook and the sample frame, answered by narrowing the claim rather than defending it. She rejects the idea that the audited sites fall short. “A lot of the websites in the audit are beautiful websites that offer a lot of information,” she says. “So it’s not so much about falling short as it is about where can you improve.” The signals the study scores are, by design, ones a clinic can fix without hiring anybody. “To put a response time on your website doesn’t require us to rebuild your website,” she says. “It doesn’t require us to even be involved with you.”
Neither executive frames the study as a call to action, which is unusual for agency research. “It’s not so much about wanting them to do something,” Cecil says. “It’s about them having information that is available to them, so that they can make small improvements if they want to.” A stated response time, she notes, is a single sentence on a page. Ipsen puts the same point as an objective: “Helping more parents find you. That’s my main message to ABA clinics.”
