# Always-On Front Desk — Go-To-Market Owner: Mike Rodgers Product: AI receptionist for dental practices Status: pre-first-customer. Nothing in this document claims a customer we do not have. --- ## 1. ICP **Primary ICP — the owner-operator general practice** | Attribute | Target | |---|---| | Practice type | Solo or 2-doctor general dentistry, fee-for-service or PPO-heavy | | Locations | 1 (2-3 is a bonus, not required) | | Annual production | $800K – $2.2M | | Operatories | 3 – 8 | | Staff | 4 – 12, with 1 or 2 front desk seats | | PMS | Open Dental, Dentrix, Eaglesoft, Curve, Denticon | | Geography | US metro and suburban. Dense metros first: Denver, Phoenix, Dallas, Charlotte, Tampa, Nashville | | Decision maker | The dentist who owns the building or the practice. One signature. No committee. | | Buying budget | Already spends $4K–$5.5K/mo on a front desk seat. This is a reallocation, not a new line item. | **Why this ICP and not a DSO first.** A solo owner can say yes in one call and there is no procurement, no security review, no pilot committee. The DSO is a better contract but a 4-month sales cycle. Land solo practices, use the operating data to walk into DSOs in month 4+. **Disqualify immediately:** - Practices under ~$500K production. The math genuinely does not work and I will say so on the call. - Medicaid-primary practices. Volume is high, per-patient value is low, ROI is thin. - Practices where the office manager is the owner's spouse. The buying decision becomes emotional and the product reads as a threat to a family member's job. Not worth the cycle. - Anyone mid-PMS-migration. Come back in 90 days. --- ## 2. Trigger event **The primary trigger: a live front desk job posting.** When a practice posts "Dental Front Office Coordinator" or "Patient Coordinator" on Indeed or ZipRecruiter, three things are simultaneously true: 1. They have an unfilled or about-to-be-unfilled phone seat. Calls are being missed **right now**. 2. They have already decided to spend roughly $4,700/mo on this problem. Budget is approved. 3. They are 3–8 weeks from a hire plus 90 days of ramp. That is a 5-month gap I can fill this week. That is the cleanest buying signal in this market. It is public, dated, and provable, and it lets the first line of a cold email be about them instead of about me. **Secondary triggers, in priority order:** | Trigger | Where it shows | Why it converts | |---|---|---| | Google reviews mentioning "no one answers", "left three messages", "couldn't get through" | Google Business Profile, last 12 months | Public, specific, and embarrassing. Quote it back verbatim. | | Practice added a second location | Google Maps, practice website | Front desk load just doubled with the same staff | | Just bought the practice (new owner listed on state board) | State dental board license lookup, DentalPost | New owners rebuild ops in the first 6 months | | Runs after-hours emergency ads | Google Ads "emergency dentist" in their zip | They are paying for after-hours calls and then not answering them | | Answering-service vendor logo in their footer | Website | Already pays for a partial version of this | **The tightest segment:** posted a front desk job in the last 14 days AND has at least one review in the last year complaining about the phone. That is a sub-100 person list per metro and it should convert far better than a broad blast. --- ## 3. Cold email sequence Five emails, 18 days. Plain text, no images, no tracking pixel, no unsubscribe footer theater. Sent from mike@rodgers.systems, not a lookalike domain. Merge fields: `[First]` `[Practice]` `[City]` `[JobTitle]` `[Days]` `[ReviewQuote]` --- ### Email 1 — Day 0 **Subject:** your front desk posting **Body:** [First], I saw [Practice] posted for a [JobTitle] [Days] days ago. Between now and the day that person is fully ramped, you are probably looking at four or five months of calls going to voicemail at lunch and after close. That gap is the expensive part, not the salary. I build an AI receptionist that answers every call and text 24/7 and books straight into your schedule. It can be live in about a week and it runs alongside whoever you hire. Worth 12 minutes to see if the math works for a practice your size? Mike Rodgers mike@rodgers.systems --- ### Email 2 — Day 3 **Subject:** the lunch hour number **Body:** [First], quick arithmetic, then I will leave you alone about it. A practice taking ~120 calls a month typically loses somewhere north of 25% of them to lunch, after-hours and the times both front desk lines are already busy. Call it 30 missed calls. If 1 in 5 of those was a new patient, that is 6 new patients a month. At roughly $950 in first-year production each, you are looking at about $5,700 a month, or $68,000 a year, that never becomes a chart in your system. You cannot see it, which is exactly why it never gets fixed. There is no report in Open Dental for the patient who called at 9pm and went somewhere else. Do you know what your after-hours call volume actually looks like? I can walk you through pulling it in about 10 minutes. Mike --- ### Email 3 — Day 7 **Subject:** what week one looks like **Body:** [First], In case the concern is disruption, here is the whole implementation: We spend two hours together. Your hours, providers, operatory templates, appointment lengths, in-network carriers, fee schedule, and your emergency protocol. That is the only real time you spend. We configure and test it against your live schedule for about a week without answering a single real patient. You review the transcripts and sign off on the triage rules in writing. Then you forward on no-answer only. Day one it picks up nothing but the calls that were already going to voicemail. Your front desk keeps doing exactly what they do now. I will run that as a 30-day pilot. If it has not booked patients by day 30, you turn off the forwarding and you are exactly where you started. Want me to put together the numbers for [Practice] specifically? Mike --- ### Email 4 — Day 12 **Subject:** pricing, plainly **Body:** [First], I would rather you know the number than wonder about it. $2,500 one time to set it up and configure it against your PMS. $890 a month after that. Unlimited calls and texts, no per-minute billing, month to month, cancel with 30 days notice. Compare that to the seat you are hiring for. A front desk salary in [City] runs about $3,300 a month, plus roughly $900 in payroll tax and benefits, plus PTO coverage, plus the amortized cost of doing this hiring cycle again in 18 months. Call it $4,700 fully loaded for coverage of about 40 hours out of the 168 in a week. $890 covers all 168. This is not an argument for firing anyone. Most owners keep their front desk person and stop paying them to be a switchboard. 12 minutes this week or next? Mike --- ### Email 5 — Day 18 **Subject:** closing the loop **Body:** [First], I will stop here. If the front desk role got filled and the phone is handled, genuinely good, that is the right outcome and you do not need me. If it is still open, or the new hire is three months from being useful, the offer stands and takes about a week to turn on. I will leave the demo here in case it is useful later: rodgers.systems/front-desk-ai Good luck with the hire. Mike --- ## 4. LinkedIn posts ### Post 1 — the number Your practice's best referral source is the phone. It goes to voicemail at 12:15pm. Here is the arithmetic nobody runs. Take a practice doing 120 inbound calls a month. Lunch hour, after close, weekends, and the moments both lines are already busy. Realistically a quarter of those calls do not get answered by a human. That is 30 calls. Say 1 in 5 was a new patient. Six new patients. At roughly $950 in first-year production, that is $5,700 a month. $68,000 a year. Now here is the part that makes it stick: none of that shows up anywhere. There is no report in Open Dental called "patients who called at 9pm and went to the practice down the street." The loss never becomes a record, so it never becomes a problem anyone is assigned to fix. You will happily spend an afternoon negotiating a supply contract to save $4,000 a year. The phone is quietly costing 15 times that. If you own a practice: do you actually know your after-hours call volume? Not your answered calls. Your inbound total. Most owners I ask have never seen the number. ### Post 2 — the build What happens when a call hits my AI receptionist at 9:40pm on a Saturday. Speech to text on the caller's first sentence. Intent classification. Booking, insurance question, emergency, reschedule, or a human handoff request. Deterministic scoring, not a vibe. Then triage runs, and this is the part I care most about. Before anything gets scheduled, the system checks for the four things that are not appointments: avulsed permanent tooth, facial swelling with fever or difficulty swallowing, uncontrolled bleeding, facial trauma. Any of those and it stops selling, pages the on-call doctor, and gives the caller the actual first-aid instruction. Put the tooth in milk. Do not scrub the root. Reimplantation window is about an hour. If it is a real booking, it queries live operatory availability, respecting appointment length, provider, lunch, and block scheduling. It writes back to the PMS and texts the confirmation. Here is the deliberate design choice. The language model is never allowed to invent a booking. It cannot generate a time. It can only choose from slots the scheduler already returned, and it cannot alter any date, time, provider or dollar figure in a reply. The LLM handles language. The scheduler handles truth. That constraint is the entire reason I would put this on a real practice's phone line. ### Post 3 — the contrarian take Hiring a second receptionist will not fix your phone problem. Not because the person is bad. Because of queueing. A human front desk is serial. One conversation at a time, by definition. Calls are not serial, they arrive in bursts: right after a lunch break, right after the 5pm ads run, the Monday after a long weekend. When three calls land in the same ninety seconds, a second receptionist does not answer all three. She answers the second one. Caller three still hears voicemail, and caller three does not call back. They hit the back button and call the next practice on the map. You are not short a person. You are short parallelism. That is why practices add a front desk seat, feel better for a quarter, and end up with the same review that says "I called three times and nobody picked up." The bottleneck was never headcount. It was that the queue depth is one. The honest test: pull your inbound call log for last month, not your answered calls. Look at how many came in between 12:00 and 1:00, and how many came in after close. If that number does not bother you, ignore me entirely. If it does, the fix is capacity that does not have a lunch break, not another chair at the front. --- ## 5. Pricing and the ROI math Full detail in `PRICING.md`. The version I say out loud on a call: **Price:** $2,500 setup, $890/month, month to month. **What it replaces:** one front desk seat, fully loaded. | Line | Monthly | |---|---| | Base salary (US metro dental front office) | $3,300 | | Payroll tax + benefits (~27%) | $900 | | PTO / sick / coverage | $300 | | Turnover + retraining, amortized | $200 | | **Fully loaded** | **$4,700** | | Hours covered | ~40 of 168 | **Always-On Front Desk:** $890/mo, 168 of 168 hours. **Direct monthly delta: $3,810. Annual: $45,720.** **The revenue side, on a 120-call/month practice:** ``` 120 calls/mo × 25% unanswered = 30 missed calls 30 missed × 20% were new patients = 6 lost new patients 6 patients × $950 year-one value = $5,700/mo = $68,400/yr ``` Recovering only **half** of those calls is $2,850/mo of production against an $890/mo cost. That is a 3.2x return on the software line alone, before the salary displacement. **Payback on the $2,500 setup:** roughly 26 days at half-recovery, from recovered production alone. **The honest version:** at under ~60 calls/month the recovered-revenue case gets thin and the pitch becomes purely a labor-cost argument. I say that on the call. It costs one deal and buys the credibility to keep the other nine. --- ## 6. Top 3 objections ### Objection 1 — "My patients will hate talking to a robot." *What they are really saying: I have a relationship business and you are about to make it feel like a call center.* **Answer:** That is the right instinct, and it is the wrong comparison. You are not comparing this against your front desk person being warm on the phone. At 9pm on a Saturday you are comparing it against a voicemail beep. Nobody has ever loved a voicemail beep. Two things I build in for exactly this reason. First, if a caller asks for a person, it stops immediately, does not try to talk them out of it, and either transfers during office hours or puts them at the top of your 8:02am callback list with the full transcript attached so they never repeat themselves. Second, it never claims to be human. If someone asks, it tells them. And here is the honest framing: your existing patients mostly text and use the portal. The people hitting your phone at 9pm are strangers with a problem who are going to call three practices. The one that picks up wins. Warmth you never got to deliver is not warmth. ### Objection 2 — "What if it tells someone the wrong thing about their insurance and we eat the bill?" *What they are really saying: I have been burned by a front desk quoting coverage wrong, and now you want to automate that mistake at scale.* **Answer:** It cannot freelance on insurance, structurally. In-network status, coverage percentages, annual maximums, deductibles and waiting periods come out of a table you review and sign off on during setup. That table is what generates the answer. The language model does not have an opinion about Delta Dental. If a caller names a carrier that is not on your list, the correct behavior is refusal, and that is what it does: it says you are not a provider for that plan and offers the self-pay fee instead of booking someone into a surprise bill. I would rather lose that booking than hand you an angry patient at checkout. Same architecture on scheduling. The model is not permitted to generate a time. It can only select from slots your scheduler returned, and a validation pass rejects any reply where a date, time, provider or dollar amount was altered. That is why double-booking is structurally prevented rather than caught later. For the first 30 days you can run review mode, where bookings land in a holding column and your front desk approves them with one click. ### Objection 3 — "We already have Weave / an answering service." *What they are really saying: I already pay for something in this category and I do not want a second bill.* **Answer:** Keep both. They do different jobs and this one runs alongside them. Weave and the tools like it are strong at the patients you already have: reminders, two-way texting, recall, reviews. They are not holding a booking conversation with a stranger at 9pm. An answering service takes a message. Think about what that message actually is. It is a task for somebody at 8:30 tomorrow morning to call back a person who, last night, was in pain and calling three practices. By the time your front desk dials, that patient is already scheduled somewhere else. You paid to be told about the patient you lost. The difference here is completion. The call ends with an appointment in your schedule, a confirmation text sent, intake forms delivered, and insurance verification queued. Nobody has a task tomorrow. Straight test: pull your answering-service messages from last month and count how many became a booked appointment. That conversion number is the entire argument. --- ## 7. First 10 prospect types and exactly where to find them Ordered by expected conversion, not by list size. ### 1. Solo GP with an open front desk req, posted 0-14 days ago $900K–1.4M production, 3-5 ops, owner is the only doctor. **Find:** Indeed search `"dental front office" OR "patient coordinator" OR "dental receptionist"` filtered to Posted: Last 14 days, radius 25mi of target metro. Exclude employers whose name contains "Dental Group", "Smiles", "Partners" (those are DSO-owned). Cross-reference the practice name in Google Maps to confirm it is a single location. ### 2. Practice with a public "nobody answers the phone" review in the last 12 months **Find:** Google Maps search `dentist` in target zip, then filter reviews by keyword using the in-listing review search for: `answer`, `voicemail`, `call back`, `phone`, `messages`, `reach`. Any 1-3 star review containing those words is a scripted first line. Quote it verbatim. ### 3. Newly acquired practice, owner licensed under 12 months at that address **Find:** State dental board license lookup (Colorado DORA, Texas TSBDE, Arizona ADBOHC all have public searchable rosters). Filter by license issue date or address change in the last year. Cross-reference to the practice website's "Meet Dr." page. ### 4. Two-location practice with one shared front desk **Find:** Google Maps: same practice name returning 2 pins within 20 miles. Then check whether both listings show the same phone number. Same number across two locations means one overloaded desk. ### 5. Practice buying Google Ads on "emergency dentist" They are paying for after-hours intent and then not answering the phone after hours. **Find:** Google search `emergency dentist [city]` between 7pm and 11pm local, collect the paid results. Or Meta Ad Library / SEMrush for the paid keyword set. ### 6. Practice whose website footer shows an answering service or "call service" vendor Already paying for a partial version of this, so the category is pre-sold. **Find:** Google `site:*.com "dental" "[city]" "answering service"`, or scrape target practice websites and grep for known vendor strings. ### 7. Apollo-built list: dentist-owners at 1-location practices in target metros **Find:** Apollo people search — `Title: Owner OR Dentist OR "Practice Owner" OR DDS OR DMD` `Industry: Hospital & Health Care / Medical Practice` `Employees: 5-25` `Keywords: dental, dentistry` `Location: [metro], United States` Then enrich for direct email and mobile. **Verify each is single-location in Google Maps before sending** — Apollo mislabels DSO-owned offices constantly. ### 8. Office managers in dental Facebook groups complaining about phone volume Not the buyer, but the best intel source and a warm path to the owner. **Find:** Facebook groups "Dental Office Managers", "Dental Peeps", "Front Office Rocks Community". Search the group for `phones`, `voicemail`, `short staffed`, `covering the front`. Do not pitch in-group. Read, then reach the owner directly. ### 9. Practices hiring for the SECOND front desk seat The strongest signal in the whole list. They have already diagnosed the problem as capacity and are about to spend $4,700/mo on the wrong fix. **Find:** Indeed job description text containing `"join our front office team"`, `"second"`, `"additional patient coordinator"`, `"growing team"` for practices whose Google listing shows one location. ### 10. Regional DSO ops directors, 4-15 locations (month 4+, not month 1) Longer cycle, much larger contract. Do not start here. **Find:** LinkedIn Sales Navigator — `Title: "Director of Operations" OR "Regional Manager" OR "VP Operations"` `Industry: Hospital & Health Care` `Company headcount: 51-500` `Keywords: dental, DSO` Plus the Group Dentistry Now DSO directory and ADSO member list. --- ## 8. Channel sequencing **Weeks 1-2 — proof of message.** Build 100 contacts from prospect types 1, 2 and 9. Send the 5-email sequence manually, in small daily batches, from a warmed domain. Manual so I actually read the replies and can hear which sentence is landing. Target: 3 booked calls. **Weeks 3-4 — LinkedIn in parallel.** Publish the 3 posts, one per week, Tuesday or Wednesday morning. Connect with every practice owner who reacts. No pitch in the connection note. The posts are for the people already in the email sequence, so the name is familiar when email 3 arrives. **Weeks 5-8 — the local wedge.** One metro, in person. Dental study clubs and local dental society meetings. A solo operator with a working demo on a laptop beats a national vendor with a brochure in this market. One referral from one respected owner is worth 300 cold emails. **Month 3+ — the operating-data pitch.** Once real practices are live, the pitch changes from arithmetic to evidence: "here is the answer rate and the after-hours booking count from a practice your size." That is when DSO conversations become worth starting. **Gate-D:** every email send, LinkedIn post and connection request in this plan requires typed approval before it goes out. Nothing in this document has been sent.