Hospital patient acquisition funnel across multiple units

Every multi-unit hospital group I have worked in has a funnel chart. It sits on slide four of the monthly review, it has five stages, and the numbers move by a percent or two each month in a direction nobody can explain. The chart is not wrong, exactly. It is just describing a thing that does not exist.

There is no group funnel. There are eleven unit funnels, or six, or twenty-three, each shaped by its catchment, its age, its TPA mix and which three consultants happen to be the reason anyone drives there. When you average them you get a number that is true of no unit and actionable for none. The mature flagship pulls the conversion rate up; the two-year-old unit in the Tier 2 city pulls the enquiry volume up. Neither is visible.

What follows is how I have learned to map this honestly. It is less tidy than the slide. It is also the only version that has ever led to a decision anyone could execute.

Where demand actually originates

Start by accepting that most of your patients did not come from marketing. In a mature unit, the bulk of volume arrives through channels that predate your department: the referring GP two kilometres away, the diagnostic centre that sends chest CTs to your pulmonologist, the corporate empanelment that makes you the default for a software park, the family who came for a delivery in 2016 and now comes for everything. Walk-ins from the neighbourhood. A sticker on an autorickshaw.

Digital sits on top of that base. It is the layer that captures demand which has become unmoored — people who have moved cities, who do not have a family doctor, who are searching because a relative has been diagnosed with something frightening and nobody has told them where to go. That is a real and growing layer. It is not the base.

The practical consequence: any funnel that begins with “impressions” is starting three steps too late and in the wrong place. A usable funnel for a hospital unit starts with a question — how many people in this catchment had a reason to need us this month, and how many of those reasons did we intercept? You cannot measure that precisely. You can estimate it well enough to stop mistaking share shift for market growth.

Service lines do not share a funnel

This is the single biggest source of nonsense in group reporting. Rolling cardiac science, orthopaedics, IVF, paediatrics and preventive health into one conversion rate is like averaging the sale cycle of a flat and a phone recharge.

Emergency and trauma. No funnel. Proximity, ambulance tie-ups, and whether the traffic police know your gate. Marketing’s job here is to not be invisible, and to make sure the emergency number resolves in one ring.
Planned surgical — joints, spine, bariatric, cosmetic. A genuine consideration cycle of weeks to months. Multiple opinions. Price sensitivity. This is where content, second-opinion journeys and consultant reputation actually convert.
Oncology. The longest and most emotionally loaded cycle, almost always mediated by a family member rather than the patient, and heavily influenced by whether the first phone call was handled with patience. Digital’s role is largely to get you into the consideration set during the two days after diagnosis.
Maternity. Nine months of predictable decision-making, then loyalty for a decade. The only service line where a drip programme genuinely works as designed.
IVF and fertility. Behaves like a direct-response consumer category. High intent, high competition, expensive clicks, brutal lead quality problems, and a consultation-to-cycle conversion that has nothing to do with marketing and everything to do with counselling.
Preventive health checks. Volume-led, price-led, seasonal, and the only line where you can move numbers in a fortnight. Which is exactly why it is over-used as proof that digital works.

If your dashboard does not split at least into emergency, planned surgical, maternity, oncology and health checks, you are not looking at a funnel. You are looking at a weighted average of five unrelated businesses.

Unit maturity changes the shape of everything

A unit in its first eighteen months has a different problem from a unit in its eighth year, and they need opposite things from you.

The new unit has no referral base, no alumni, no autorickshaw stickers, and a catchment that does not know the building has opened. Its funnel is almost entirely top-heavy: awareness, search for the brand name plus locality, and a desperate dependence on whichever two consultants arrived with a following. Conversion looks terrible because the enquiries are exploratory. Cost per acquisition looks terrible because you are buying attention, not intent.

The mature unit has the inverse problem. Its catchment is saturated, its brand search is already strong, and incremental media buys mostly harvest demand that would have arrived anyway. Its funnel leaks at the bottom — slot availability, waiting time, the contact centre, the fact that its best surgeon is booked for eleven days.

I got this wrong for most of a year. I held every unit to the same cost-per-appointment target, which meant the new units were systematically starved — their numbers always looked worse — and the mature units were systematically over-funded, because their numbers always looked good. We were paying a premium to buy back our own demand in the places that least needed it, and under-investing in the only places where new demand had to be created.

Why group-level funnel maths misleads

Three specific failure modes, all of which I have presented to a board before I understood them.

Simpson’s paradox, in plain clothes. Group conversion improves while every unit’s conversion declines, because volume shifted towards the unit with the best conversion. Or the reverse. The group number is a mix effect masquerading as a performance effect. If you cannot decompose month-on-month change into mix versus rate, do not report the change.

Double-counted enquiries. A patient who fills a form, then calls the central number, then walks into the unit is three records in three systems. Group totals are inflated by exactly the amount of duplication you have not deduplicated, and the duplication rate is not constant — it rises with campaign spend, which makes your best months look even better than they were.

The denominator nobody owns. Units count footfall. The contact centre counts calls handled. Digital counts leads. Finance counts registered patients. Four numbers, four owners, none reconciling to another. I have sat in reviews where the same month had four different “total enquiries” figures and nobody flinched, because each owner was confident about their own.

The leaks that are structural rather than creative

When conversion is poor, the reflex is to blame the creative, the landing page or the agency. Occasionally that is right. Usually the leak is in the operating model, and no amount of better copy touches it.

Response latency. An enquiry answered in four minutes and an enquiry answered in four hours are different products. Most groups have no service-level agreement on this, and the ones that do measure it at the aggregate and not at the unit-and-shift level where it actually breaks.
Slot scarcity as a marketing constraint. You can generate all the interest you like for a consultant who has no slots for a fortnight. The enquiry does not wait; it goes to whoever can see them on Thursday. Marketing spend against a full OPD calendar is a transfer of money to your competitor’s conversion rate.
Language mismatch. A Telugu-speaking enquirer routed to a Hindi-speaking agent converts worse, and the loss never appears as a language problem. It appears as a bad lead.
Payer friction. Cashless eligibility, TPA empanelment gaps and pre-authorisation timelines kill more planned surgeries than price does. The patient says they will think about it. They have in fact gone to the hospital where their policy works without a fight.
Routing to the wrong unit. Group campaigns that send a patient in one part of the city to the unit forty minutes away because that unit had the budget. They do not travel. They drop.
Consultant-level dependency. One doctor’s exit can collapse a service line’s funnel in a quarter, and your dashboard will read it as a digital performance problem for three months before anyone says the real reason out loud.

The funnel model I would actually build

Not five stages. Seven, defined by who owns them, so that a decline has an owner rather than a discussion.

Addressable need in the catchment, estimated annually, by service line. Crude is fine. Directional is enough.
Intercepted intent — search, referral, empanelment, walk-in. Split by whether you created the demand or harvested it.
Identified enquiry — a contactable human, deduplicated across channel. This is the first number you can defend.
Contacted within the agreed window, measured at unit and shift level.
Appointment booked, with the slot offered recorded alongside the slot requested.
Appointment honoured. The no-show gap is the most under-examined leak in Indian hospital marketing and frequently the largest.
Treated and revenue-recognised, linked back to stage three by identity, not by attribution model.

Stages three through six belong to operations as much as to marketing. That is the point. A funnel that only measures the part marketing controls will always conclude that marketing is performing and something else is broken.

If you are starting this next quarter

In order, and do not skip ahead.

Pick one unit and one service line. A mid-maturity unit and planned orthopaedics is a good pair: long enough cycle to study, short enough to see the end.
Reconcile the denominators. Sit the unit head, the contact centre lead and your analyst in one room with the same month’s data and do not leave until there is one agreed enquiry count. This takes longer than you think and is worth more than anything else on this list.
Instrument deduplication by phone number. Imperfect, legally straightforward, and it will immediately tell you your volumes were overstated.
Measure response latency at the shift level for four weeks before you change anything. You will find that the problem is two specific shifts.
Overlay slot availability on enquiry volume. If the two lines diverge, stop buying traffic for that consultant and say so in writing.
Then replicate to two more units — one newer, one older. The contrast is what teaches the group, not the single case.
Only after that, rebuild the board slide. With mix and rate separated, and with one number per unit rather than one number for the group.

What to tell the board

That the group funnel is a reporting convenience, not an operating reality, and that you are replacing it with a set of unit funnels that will initially look worse. Say the word “worse” out loud before someone else does. The numbers will drop when you deduplicate, and if you have not pre-announced that, the improvement in data quality will be read as a decline in performance.

Then show them the one thing a group funnel can never show: which unit’s problem is demand, which unit’s problem is capacity, and which unit’s problem is the contact centre. Those three need different money.

A funnel is not a measurement system. It is an argument about where the next rupee goes, and averaging eleven hospitals into one curve is how that argument gets won by whoever shouts loudest.

Questions people ask

What is a patient acquisition funnel for a multi-unit hospital group?

It is the sequence from a person in the catchment needing care to that person being treated and billed, split by unit and service line. The version most groups run — impressions to admissions on one slide — describes nothing real, because a mature flagship and a two-year-old Tier 2 unit have funnels of completely different shapes. A usable funnel has seven stages, each with a named owner, so a decline has an owner rather than a discussion.

Why does a group-level hospital funnel mislead the board?

Because it averages units that have nothing in common. Group conversion can improve while every unit’s conversion declines, simply because volume shifted towards the unit that converts best — a mix effect read as a performance effect. Add enquiries counted three times across form, phone and walk-in, and four departments each reporting a different “total enquiries” figure, and the slide becomes a number that is true of no unit and actionable for none.

Where does most demand for a hospital actually come from?

Not from marketing. In a mature unit the bulk of volume arrives through the referring GP nearby, the diagnostic centre that sends scans, corporate empanelment, families who came for a delivery years ago, and walk-ins from the neighbourhood. Digital sits on top of that base, capturing demand that has become unmoored — people new to the city, without a family doctor, searching after a frightening diagnosis. It is a real and growing layer. It is not the base.

Should every service line share one hospital funnel?

No. Emergency has no funnel — proximity and ambulance tie-ups decide it. Planned surgery has a consideration cycle of weeks to months. Oncology is the longest and is mediated by a family member. Maternity is nine predictable months followed by a decade of loyalty. IVF behaves like a direct-response consumer category. Health checks move in a fortnight. Rolling these into one conversion rate is a weighted average of five unrelated businesses, not a funnel.

How should a new hospital unit’s funnel differ from a mature unit’s?

A unit in its first eighteen months has no referral base and a catchment that does not know it exists, so its funnel is top-heavy: awareness, brand-plus-locality search and dependence on two consultants. Conversion looks poor because enquiries are exploratory. A mature unit’s catchment is saturated and its funnel leaks at the bottom — slot scarcity, waiting time, the contact centre. Holding both to the same cost-per-appointment target starves the new unit and over-funds the old one.

What are the leaks in a hospital acquisition funnel that better creative cannot fix?

Response latency — an enquiry answered in four hours is a different product from one answered in four minutes. Slot scarcity, where spend goes against a consultant with no availability for a fortnight. Language mismatch between caller and agent. Payer friction from cashless eligibility and TPA gaps. Routing a patient to the unit forty minutes away because it had budget. And consultant dependency, where one exit collapses a service line and the dashboard blames digital for three months.

How long does it take to rebuild a hospital group’s funnel honestly?

Plan on two to three quarters before the board slide changes. One unit and one service line take a quarter: reconciling denominators, deduplicating by phone number, four weeks of response latency at shift level, and overlaying slot availability. Replicating to a newer and an older unit takes another quarter, and the contrast between them is what teaches the group. Only then rebuild the board reporting with mix and rate separated.

What does fixing the hospital funnel cost, and where does the money go?

Very little in media and quite a lot in people’s time. The expensive parts are an analyst who can decompose mix from rate, deduplication tooling in the CRM, and the hours a unit head, contact centre lead and analyst spend in one room agreeing a single enquiry count. The saving is larger: you stop paying a premium to buy back your own demand at saturated units and redirect it to units where demand has to be created.

Who should own each stage of a hospital patient acquisition funnel?

Each stage needs one named owner. Marketing owns addressable need and intercepted intent. Digital and the contact centre share the identified enquiry and the contact window. Operations owns booking against slot availability and the appointment being honoured. Finance owns treated-and-billed, linked back by identity. Stages three to six belong to operations as much as to marketing — that is the point. A funnel measuring only what marketing controls always concludes something else is broken.

What does IT need to do to deduplicate hospital enquiries across channels?

Start with the phone number. Matching on mobile number is imperfect, legally straightforward under the consent already collected for contact, and immediately shows how overstated volumes were. The harder work is getting the web form, contact centre and unit front desk into one identity so a patient who does all three is one record. Expect volumes to fall when this lands, and expect the duplication rate to have been rising with campaign spend.

What should a CFO ask before funding more media for a hospital unit?

Three questions. Is the unit’s problem demand, capacity or the contact centre — because those need different money. Is the consultant being promoted actually available within a week, or is the spend transferring patients to a competitor who can see them on Thursday. And has the enquiry count been deduplicated, or is cost per appointment being calculated on a denominator inflated by campaign-driven double counting. A group funnel cannot answer any of these.

How do you tell the board the funnel numbers are going to look worse?

Say the word “worse” before someone else does. Explain that the group funnel is a reporting convenience being replaced by unit funnels, that deduplication will drop reported volumes, and that this is data quality improving rather than performance declining. If you do not pre-announce it, the improvement will be read as a decline. Then show what the group funnel never could: which unit’s problem is demand, which is capacity, and which is the contact centre.

Does this funnel approach apply to a single hospital or only to groups?

A single hospital still has the service-line problem — emergency, planned surgery, maternity, oncology and health checks do not share a funnel — and still has the denominator problem where footfall, calls, leads and registrations never reconcile. What it lacks is the mix effect between units, which simplifies the maths considerably. Start with the service-line split and response latency by shift. Those two will explain most of what a single-site funnel is hiding.

What does a senior doctor leaving do to a hospital’s acquisition funnel?

It can collapse a service line’s funnel within a quarter, and the dashboard will read it as a digital performance problem for months before anyone says the real reason aloud. This is why the funnel must be built at consultant level for the lines that depend on named doctors — orthopaedics, oncology, IVF. A medical director who sees enquiry volume tracked by consultant understands the dependency, and the group can plan the second hire before the first one leaves.