The full plan to launch the $30,000/month Google Ads structure.

What this plan does — and the one rule behind it

Today, 100% of paid budget runs through a single Performance Max campaign whose efficiency has declined for months on flat spend, while all keyword Search sits paused and brand searches are bought at a premium inside that one black box. This plan splits the same $30,000/month across an intent-tiered structure where every dollar maps to a purpose we can read, steer, and defend.

One rule governs the whole rollout: respect the learning period. Every conversion-goal change, brand exclusion, creative refresh, or bid-strategy switch resets Google's machine-learning for ~1–2 weeks. We flag each such change resets learning and deliberately space them out — because CCG is in an early, trust-rebuilding posture, and we want clean, attributable wins first, not a risky all-at-once teardown.

The $30,000 / month, split into legible tiers

This is the steady-state target (≈ Month 3). Performance Max stays the largest line because it's the only surface driving Maps, local and re-engagement — but it is now capped, goal-locked, brand-excluded and creative-refreshed rather than absorbing everything blind. The other 40% funds the attributable, intent-rich Search layer.

Budget envelope confirmed by the client and matched to the current ~$30K/mo PMax run-rate. This reallocates existing spend; it is not a request for more.

Six rules the audits demand

1. Make spend legible. Split the monolith so each dollar maps to an intent tier we can read and steer — not one black box.
2. Stop paying PMax ~$325/conv for brand searchers. Add a brand-exclusion list, remove the brand & "free therapy" search-theme signals, and run a dedicated Brand Defense campaign that captures that traffic at a fraction of the cost.
3. Re-open high-intent non-brand Search — but only against proven converters, with the ~$28K historical waste list baked in as negatives from day one.
4. Keep PMax, but fix it before scaling. Lock the goal to booked revenue, refresh the creative, and only then introduce a target-ROAS. Never scale budget into a declining ROAS.
5. Weight to where money already works: mobile-first, lean into FL + NY + MD, put DC on a short leash.
6. Respect learning periods. Stage every learning-resetting change — don't stack them.

What changes in PMax — fix it, then scale it

PMax is 60% of spend, so getting it right matters more than anything. It currently spends to a $1,000/day cap, and its ROAS decayed 2.86× → 1.26× from March to June on flat spend — the classic signature of creative fatigue. Changes below are ordered by leverage and grouped so we never reset two campaigns' learning in the same week.

A1 · Measurement & read-accuracy do first

Until these are done, every efficiency read is contaminated. No budget or bid change happens before them.

1. Lock the bidding conversion goal to Won + Consult Scheduled only. resets learning The live campaign already counts deep-funnel revenue correctly (Won 351 / $236K · Won Calls 159 / $115K · Consult Scheduled 46 / $11K). Demote ad-extension taps, page views and "Learn More / Book" micro-clicks to secondary. The 8 paused local campaigns were optimizing to a $0 ad-extension-tap goal — which is why they looked cheap (~$39 "CPA") while booking almost nothing. That goal must never re-enter a bidding column.
2. Stop reading allConversions. 556 real bidding conversions vs 22,404 all-conversions — a 40× gap of micro-actions. Build every dashboard on primary conversions only.
3. De-duplicate conversion actions. 6+ overlapping lead actions and 10+ call actions mean one inquiry can trip several. Define one primary lead action + one primary call action per funnel step; set the rest to secondary. This is the single biggest source of inflated counts.
4. Add a brand-exclusion list + remove brand/free signals. resets learning Apply a BRAND_LIST exclusion (none exists today); remove the counseling center group search-theme signal (feeding own brand tells PMax to chase brand → cannibalization) and the where can i find free therapy signal (CCG is premium self-pay). Then re-read true non-brand ROAS — it will look lower, and that's the honest number. (1,699+ brand clicks run through PMax today with no exclusion.)

A2 · Fix the decay — the revenue problem

1. Refresh creative / activate asset-group diversity. resets learning — the highest-leverage performance fix. 17 asset groups exist; only General Brand is enabled, so 100% of $18K flows through one aging creative pool feeding every geo and modality. That group is well-built (ad strength EXCELLENT) — the problem is monoculture, not quality. Re-enable/rebuild the 16 paused modality & geo groups with fresh video + images, prioritizing the modalities that actually convert in Search (EMDR/Trauma, Couples, DBT — skip fresh IFS/CBT creative). Clean the jumbled naming first (14 groups carry UNKNOWN ad strength; a [LOCAL][DC] campaign contains Bethesda & Arlington groups).
2. Strengthen audience signals. General Brand has 50 search-theme signals but only 1 audience signal. Add remarketing lists (site visitors, blog readers, inquiry-not-booked), Customer Match of past inquiries (hashed email only — no PHI, BAA in place), and in-market "therapy / mental-health." Signals are hints, not targeting — but they steer a value-maximizing PMax toward the right people.

A3 · Bidding, budget & hygiene

• Hold budget flat at $18K — do not scale into a declining ROAS. Let the creative refresh work first (optimization score 0.89, ~11% headroom).
• Introduce a tROAS only after ROAS restabilizes resets learning — set near the trailing achieved ratio (~1.7–2.0×), never a hard target mid-slide (that chokes volume).
• Verify Final URL Expansion in the UI. If ON with no page feed, add a page feed limited to modality + location pages and exclude the non-converting blog/explainer URLs.
• Geo re-weight: lean into Florida (2.69× value/spend); interrogate Washington DC (1.02× on ~$21K/half-year) — fix-or-trim. Remove West Virginia if its $0 targeting is unintended.
• Device: mobile $232/conv vs desktop $933/conv (~4× worse). PMax has no device bid lever, so this is a mobile-first landing-page mandate — page speed, message match, tap-to-call above the fold.
• Placement exclusions: 7,263 placements served, skewed to Display/website (110.8K impressions). Run an account-level exclusion pass on the junk-app / made-for-advertising long tail.
• Extend negatives with the low-intent drift PMax pulls in: bipolar disorder symptoms, ocd symptoms, adhd symptoms in kids, how to deal with a narcissist, neurodivergent.

PMax change sequencing — never stack two learning resets

The Search build — organized by specialty, not location

The data demands a modality-led structure. In the served-keyword history, specialty is a 14× cost lever ($389 EMDR → $5,431 generic online), while geography is only a 1.7× lever ($454–$786). So we run three campaigns by modality, each with one ad-group per geo. This pools conversion signal at the campaign level (so target-CPA bidding has enough data — fixing the lesson where eight $48/day local budgets starved the algorithm), funds the cheap modalities and never the expensive ones, and keeps each geo legible for bid weighting and geo-specific negatives.

Which specialties we fund — and which we kill

Match-type policy and the shared negative list

Match-type policy (account-wide)

Historical: Broad $477/conv (best) · Phrase $596 · Exact $698 (worst). Counterintuitive — but the Exact set skewed to high-CPC head/generic terms that don't convert.

• Default to Broad + Phrase on local-intent terms ([modality] near me, [modality] [city/state]) — the proven winners. Broad is safe only because the negative footprint below is heavy and Smart Bidding controls it.
• Use Exact only for proven city-modality combos (trauma therapist dc, emdr therapy dc, washington dc couples therapy).
• Never run broad/phrase on generic single-modality heads (emdr, cbt, therapy) without the negative list.

The shared negative-keyword list — built once, attached to all three campaigns

Today's negatives are uneven (NJ 47, NY 28, MD 10, DC 10, VA 6, FL-Boca 0) and ~$23–28K of historical spend converted zero. One shared list closes that, plus the geo-specific negatives above.

• Dead modalities: ifs, internal family systems, cbt, cbt counseling, cbt treatment, accelerated resolution therapy.
• Generic low-intent: online therapy near me, online therapist near me, online therapy, generic teletherapy, bare family therapy / family counseling services (keep specific parent therapy, which converted in MD).
• Self-pay protect (CCG is out-of-network): free, free therapy, low cost, sliding scale, medicaid, medicare, insurance, in network, takes insurance, aetna/cigna/bcbs/united.
• Wrong service / safety: psychiatrist, medication, meds, crisis, hotline, suicide, emergency, 988.
• Anti-persona / non-commercial: child, kids, teen, adolescent (unless a pediatric line is intended), jobs, careers, salary, degree, course, certification, training, definition, symptoms, quotes.
• Geo-specific: emdr therapy near me & dbt therapy near me as negatives in the DC ad-groups.

Cross-campaign settings (all three Search campaigns)

• Brand as negative: add CCG brand terms as campaign negatives on every non-brand campaign so non-brand never absorbs brand clicks. This is the read-accuracy backbone — brand and non-brand never blend.
• Ads: 2–3 RSAs per ad group (≥10 headlines / 4 descriptions). Lead with modality specificity + locality + the evidence-based, short-term wedge. Honest self-pay framing; no outcome guarantees, no "best/#1," no insurance language.
• Assets: Sitelinks (Book a Consult, Our Therapists, [Modality] page, Locations, Pricing/Insurance FAQ); Callouts (Evidence-Based, Short-Term Focus, In-Person & Virtual, Specialist Matching); Structured Snippets (DBT, EMDR, Gottman, Trauma…); Call asset (CallRail number); Location assets (the 16 GBP locations — ties to the local work).
• Landing pages: route each ad group to its matching modality × location page (not the homepage); confirm HTTP 200 and a working mobile booking/consult CTA before launch.

Search bidding ramp (per campaign)

1. Launch on Max Conversions, no tCPA — let it find conversions for ~2 weeks / ~15–30 conv.
2. Switch to tCPA at the modality target ($400 EMDR / $425 Couples / $500 DBT). resets learning
3. After 3–4 weeks at target, tighten toward the 3:1 LTV:CAC line (~$590 ceiling, push to ~$450) if volume holds.
4. DC fix-or-trim gate at 6–8 weeks: if DC ad-groups can't clear their tCPA, pause and reallocate to VA/NY.

Brand Defense Search — $1,500 / mo

PMax currently harvests 1,699+ brand clicks at ~$325/conv blended economics. A dedicated brand campaign captures them at low CPC / very high ROAS and frees PMax to chase incremental non-brand once brand-excluded. It's demand-constrained, so it won't spend the full $1,500 unless conquested.

• Bidding: Max Conversions → Target Impression Share (90%+ top-of-page) once stable.
• Keywords (Exact + Phrase): counseling center group, counseling center of maryland/new york/[state], ccg therapy, counseling center group [city]; near-brand bethesda counseling services.
• Negatives: jobs, careers, login, patient portal, competitor brand names.
• Ads: brand RSAs with sitelinks to Book, Our Therapists, Locations, Modalities; call asset. A light competitor-conquest variant stays in Reserve until brand is stable.

Reserve / Test — $1,500 / mo, from Month 2–3

Held back at launch; deployed only once PMax has restabilized and the three Search campaigns are at tCPA. Pick one to start:

The 90-day launch sequence

A deliberate, staged rollout — sequenced so we never blind two campaigns at once. Because CCG is in a trust-rebuilding posture, the fastest-to-prove wins (Brand Defense + EMDR/Couples Search) come first.

KPIs & the CAC sanity-check

Client-confirmed, all-time unit economics give us a confident basis for "good":

The 2.8–3.7× ratio is already healthy (3:1 is the standard target), so the job is to hold CAC at/under ~$631 while scaling intakes, pushing toward 3:1+ (≤ ~$590). At $30K/mo and the current ~$585–$631 cost/booked-intake, that's ~48–51 paid+lifecycle intakes/mo — under the ceiling, with room to grow. If paid + lifecycle + organic together approach the intake ceiling, that's a hiring trigger, not a reason to cut spend.

Compliance & brand guardrails

• Self-pay / out-of-network framing throughout — transparent on cost, never implying insurance coverage or in-network status.
• No outcome or "cure" guarantees — evidence-based, "designed to" language only; APA-ethics and Google health-policy compliant.
• No crisis or fear-based messaging — crisis searches routed to 988, never positioned as CCG.
• HIPAA-first — no client or clinical information in any tag, audience, Customer Match upload (beyond hashed email), feed or report; a BAA with every vendor before anything goes live.
• Brand self-claims attributed, not asserted — and every CPA, ROAS and dollar figure here is a sourced historical actual from the PMax & Search deep dives (Jan–Jun 2026, account 6356233337), or a clearly-labeled planning target. None are invented.