The State of Addiction Treatment Search 2026

Executive summary

This is the inaugural edition of the State of Addiction Treatment Search, an annual benchmark of how people actually search for treatment in the United States. Most marketing advice in this field is anecdotal. This is not: it is built from a systematically expanded universe of 2,410 distinct addiction-treatment search terms, each measured for demand, cost-per-click, ranking difficulty and live SERP features, then benchmarked against the wider healthcare market.

One structural decision shapes everything that follows. Searches for a treatment service and searches about a substance or medication behave nothing alike, so we separate them. The 1,916 service-seeking terms are what fill beds. The 494 research terms carry far more raw volume (1,540,080 searches a month) but almost no commercial intent. Averaging the two together, as most keyword studies do, buries the commercial picture under pharmacology traffic.

Four findings define 2026. Treatment search is overwhelmingly local: 69.8% of service-seeking terms return a local pack, and 95.4% of "near me" terms do. Paid search is expensive relative to the rest of healthcare, with high-intent clicks averaging $9.24 and the category as a whole running 4.5 times cosmetic surgery. Demand is genuinely transactional: a third of service-seeking queries are people trying to find and contact a provider now. And the AI-search story is sharper than a single number suggests. AI Overviews have already taken the education layer, appearing on 57.5% of substance and medication research queries, while barely touching the money terms at 0.5% of geographic searches. Centers that publish education content are already feeding answers they will not get clicks from; the transactional window is still open, and closing.

Key statistics

• 2,410 distinct addiction-treatment search terms measured, split into 1,916 service-seeking terms and 494 substance and medication research terms.
• Service-seeking demand totals 913,900 searches a month across 48 states.
• Median cost-per-click $6.00; average $7.78; high-intent terms average $9.24.
• The top decile of terms costs $15.00 or more per click; 2.2% cost $25 or more.
• Addiction clicks run 4.5x cosmetic surgery and 22x consumer-health and wellness search.
• Only insurance ($22.28) and personal-injury law ($14.39) out-price addiction in our comparison set.
• 69.8% of treatment searches return a Google local pack, rising to 88.0% for geographic queries and 95.4% for "near me" queries.
• 22.4% of demand explicitly names a city or state; 37.1% carries explicit local intent.
• Only 9.2% of service-seeking searches show an AI Overview, versus 95.0% of adjacent consumer-health queries.
• But 57.5% of substance and medication research queries already show one. The education layer has already gone to AI.
• AI Overviews on geographic treatment queries are effectively 0.5%, a blind spot directly over the money terms.
• Transactional queries are 32.7% of service-seeking demand, not a fringe.
• Detox carries the highest cost per click ($11.08); sober living the lowest.
• The treatment gap remains the backdrop: roughly 77% of U.S. adults with mental illness went untreated in 2024 (SAMHSA).

1. Market structure: fragmented and local

Addiction-treatment search is not a handful of head terms; it is a long tail of thousands of modality, substance, program and city-level queries. After removing non-treatment "rehab" and "detox" meanings, dropping low-confidence entries and separating research queries from service-seeking ones, we measured 1,916 service-seeking terms with real monthly demand, 22.4% of which explicitly name a city or state.

The head is modest. The largest service-seeking term in the category draws 23,000 searches a month, and no term in the category clears 25,000. That is the whole strategic point: there is no single keyword that wins this market.

2. Search intent

Classified by primary intent, service-seeking demand divides between people learning about treatment and people trying to find it. Informational queries lead at 47.9%, with transactional queries at 32.7%. That second figure is the one worth sitting with: roughly one query in three is someone actively looking for a provider to contact, which makes this a far shorter funnel than the category is usually credited with.

3. The local battleground

69.8% of service-seeking terms return a Google local pack, climbing to 88.0% for explicitly geographic queries and 95.4% for "near me" queries. The map pack and Google Business Profile routinely sit above both organic results and ads.

Inpatient and outpatient programmes show the strongest local-pack presence, which follows from how people search for them: a bed or a day programme is something you travel to, so the query almost always carries a place. Detox sits lower because a large share of detox queries are still informational.

4. Paid-search economics

Across 787 terms with cost data, the average cost-per-click is $7.78 and the median is $6.00. The high-intent terms that actually drive admissions average $9.24. The top decile starts at $15.00, and 2.2% of terms cost $25 or more.

A note on what this number is. These are modelled average costs per click, not top-of-page bid ceilings. The two are routinely confused, and the difference is large: a bid ceiling answers "what is the most anyone has paid", an average answers "what does a click typically cost". Ceiling figures for this category run several times higher than the averages below. Read these as a relative index of competitive pressure between terms and verticals, not as a bid guide, and expect live auction costs in the most contested metros to exceed them.

Most expensive keywords

Restricted to terms drawing at least 200 searches a month, so that thin, unstable long-tail entries do not dominate the list.

5. How addiction CPCs compare to the rest of healthcare

"Expensive" is relative, so we benchmarked addiction treatment against other high-value healthcare categories and the two priciest verticals in all of search, measured the same way, in the same month, with the same metric. Addiction treatment averages $7.78 per click, roughly 4.5x cosmetic surgery and 3.3x dental, and it out-prices LASIK, fertility and major elective surgery. Only insurance and personal-injury law, the perennial CPC champions, sit above it.

6. Demand and cost by substance

Segmenting service-seeking terms by the substance they reference shows where demand and competition concentrate. Most treatment search does not name a substance at all: people search for a service first and disclose the substance later.

7. Demand and cost by modality

By level of care, demand and cost diverge sharply. Detox commands premium clicks; sober living and housing draw comparable volume at roughly a third of the price.

Outpatient and IOP is the notable outlier on AI Overviews at 34.1%, far above every other level of care. The reason is that "IOP" and "intensive outpatient program" are terms people need defined before they can shop for one, and definitional queries are exactly what Google is confident summarising.

8. Specialty audiences

Queries that specify an audience are smaller in volume but often higher in intent and value. Two segments carry real weight: adolescent and teen treatment, which draws the largest specialty volume, and luxury or executive care, which carries the highest cost per click of any segment in this report.

9. Geographic landscape

Treatment demand is measurable across 48 states. The table below shows the states with the most keyword coverage in our dataset. State-level keyword counts reflect how deliberately we sampled each state as well as underlying demand, so read the cost column rather than the coverage column for competitive signal.

New Jersey is the most expensive state market measured, at $19.36 per click, ahead of Texas, Florida and Colorado. California and Florida carry the deepest coverage, which matches the concentration of licensed capacity in both states.

10. The AI-search shift: the defining story of 2026

The headline

AI Overviews have already transformed adjacent health search but have barely arrived in transactional addiction treatment. In our data, 95.0% of consumer-health and wellness queries (the "how to lose weight", "gut health" world) now return a Google AI Overview, versus 9.2% of service-seeking treatment queries.

Where AI Overviews appear in treatment today

AI Overviews in treatment are not random. They concentrate in informational and navigational queries and are almost entirely absent from transactional, local "near me" and "[city] rehab" searches.

The transactional blind spot

The queries that actually produce admissions are an AI blind spot. AI Overview prevalence on non-geographic queries is 11.5%, but on geographic city and state queries it is 0.5%, and on "near me" queries 3.9%. Google still hands those to the local pack and organic results, not an AI summary.

Treatment queries already triggering AI Overviews

These are the leading edge: service-seeking terms where Google already shows an AI answer. They are the queries to optimise for citation first.

One row in that table matters more than the rest. "intensive outpatient program near me" is a transactional, local, bottom-of-funnel query, and it already returns an AI Overview. That is the pattern arriving at the money terms, on a term with real volume behind it.

What this means, and what wins AI citations

The window is open but closing, and it is closing from the top of the funnel down. The education layer has already gone. As Google grows confident summarising higher-stakes health topics, AI Overviews will keep moving toward "[city] drug rehab" and "alcohol detox near me". Centers that wait will find the answer already written, and themselves uncited.

What wins AI citations: clinically reviewed, genuinely authoritative content; explicit, structured answers to the questions patients and families ask; schema and entity clarity; consistent NAP and local signals; and third-party corroboration through reviews, directories and citations. This is Answer Engine Optimization, and 2026 is the year to build it, while the transactional SERPs are still uncrowded.

11. Ranking difficulty

Organic difficulty across the universe is more attainable than operators assume, because so much volume sits in specific, local and long-tail terms rather than a few hyper-competitive heads. 48% of service-seeking terms score 20 or below on a standard 0-100 keyword difficulty scale.

The distribution is genuinely bimodal. There is a large accessible tail and a hard core of 292 contested terms, mostly the unqualified category heads ("drug rehab", "alcohol treatment") and the major metro markets. The winnable volume is in between, and it is where most centers are absent.

12. What it means for operators

Win local first. With 69.8% of service-seeking searches showing a local pack and 95.4% of "near me" searches doing so, Google Business Profile and local SEO are the highest-return investments available.

Do not rely on paid alone. At $9.24 per high-intent click, and multiples of that in contested metros, paid search is a tax you pay forever. Compounding organic and content assets lower blended cost per admission.

Cover the matrix. Demand is fragmented across modalities, substances, audiences and cities, and the biggest term in the category draws under 25,000 searches a month. Systematic coverage beats chasing a few head terms.

Change what education content is for. More than half of research queries already return an AI Overview. Publishing explainers to earn clicks is a losing trade; publishing them to be the cited source, with the clinical review, schema and entity signals that earn a citation, is not.

Build AEO now. AI Overviews have reached one transactional term with real volume in this dataset. They will reach more. Structured, authoritative, citable content is the cheapest insurance you can buy this year.

13. Methodology

Collection. We built the keyword universe by expanding 19 seed groups spanning treatment modalities (detox, inpatient, outpatient, IOP, PHP, MAT, residential, sober living, halfway house), substances (alcohol, opioids, heroin, fentanyl, meth, cocaine, benzodiazepines, marijuana, prescription drugs), specialty audiences (teen, adolescent, veterans, men, women, executive, faith-based, luxury, LGBTQ) and geography (all 50 states plus the largest metros). Seeds are multi-word phrases anchored to treatment intent rather than single words like "rehab" or "detox", which pull in physical therapy, wildlife rescue and detox teas. Each term was enriched with search volume, cost-per-click, keyword difficulty and live SERP features for the United States, compiled August 2026 from industry keyword research and SERP measurement platforms.

Filtering. 4,819 raw terms were reduced to 2,410. We removed non-treatment senses of "rehab" and "detox" (88 terms: physical, cardiac and skilled-nursing rehabilitation, wildlife rescue, detox teas and cleanses, drug-test evasion products, entertainment and jobs), terms with no addiction or treatment signal (284), and 1,449 low-confidence entries carrying neither a difficulty score nor live SERP features. That last filter matters: such entries arrive in banded groups of identical volume across dozens of city variants, which is a fingerprint of estimate clustering rather than measured demand.

Segmentation. The 2,410 treatment terms are split into 1,916 service-seeking terms (someone looking for a facility, programme or admission) and 494 research terms (someone looking up a substance, medication, symptom or timeline). Commercial figures throughout this report describe the service-seeking segment. Reporting them across both would let 1,540,080 monthly searches of pharmacology traffic, which converts to essentially no admissions, set the category's headline cost and volume.

Prevalence sampling. Percentages (local pack, AI Overview, intent mix, difficulty) are computed on a neutral sample of 1,043 service-seeking terms that excludes the geography-targeted seed groups. Those groups were added to give the state table enough cells, and including them would inflate every place-related finding: measured across everything, "names a city or state" reads 57.8% against 22.4% on the neutral sample. Prevalence measured within the geographic stratum itself (for example, what share of geographic queries show a local pack) uses the full service corpus, since oversampling a stratum does not bias a rate computed inside it.

Reporting floor. Any cell computed on fewer than 25 terms is reported as "n/a" rather than given a number. Several substance, audience and state cells fall below that floor. A percentage derived from nine keywords is not a finding.

Cost metric. CPC figures are modelled average costs per click, not top-of-page bid ceilings, and the two differ by a large multiple. All verticals in section 5 are measured with the same metric, in the same month, so the comparison is like for like. Absolute values should be read as a relative index of competitive pressure, not as a bid guide.

Volume caveat. Search engines group near-identical variants under shared volume estimates, so per-term volume is directional and total-search figures more so. We deliberately report distributions and prevalences rather than a single market-size number, and we exclude the banded low-confidence entries described above, which are the worst offenders. Treatment-gap context is from SAMHSA's 2024 National Survey on Drug Use and Health. All figures are point-in-time and will change; this is the v1 baseline.

Sources:

• Industry keyword research and SERP measurement platforms (keyword expansion, search volume, cost-per-click, keyword difficulty, live SERP features), United States, August 2026.
• SAMHSA, 2024 National Survey on Drug Use and Health (NSDUH).
• Analysis, filtering, segmentation and benchmarking by Addiction Marketing Agency.

14. How to cite this report

What is next (v2)

The 2027 edition adds two original datasets: an operator survey on budgets, channel mix and cost-per-admission, and a first-of-its-kind response-time study measuring how quickly treatment centers actually answer and follow up with inquiries. To take part, or to benchmark your center privately, get in touch.