Call centers and appointment setters should buy homeowner data on cost per appointment, measured on their own dialer, never on price per record: the list that books more appointments per caller hour is the cheaper list, whatever it costs per row. Almost every quote a floor manager receives is written per record, because that is the number a vendor controls. The number the floor controls is the caller hour, and it is many times larger. This guide is what to measure, how the cost-per-appointment math works, and how to run a head-to-head between two lists that the losing vendor cannot argue with.
Disclosure: Scout Data is our product and sells homeowner data to call centers. The method below works on any vendor, including us.
Why price per record misleads
A record is an input. The output a setter floor sells, or books for its own closers, is a held appointment. Between the two sit the hours a caller spends dialing numbers that do not answer, reaching tenants and adult children instead of the owner, hearing a disconnect tone, and working rows your own compliance scrub throws out before the first dial. Every one of those costs caller time, and caller time is the expensive part of a floor.
That is why two vendors whose prices differ fivefold per record can produce appointments at close to the same cost, and why the cheaper one on the rate card is often the dearer one on the P&L. What a record costs on its own, and why vendors quote it that way, is in homeowner data cost per record. This page is about what happens after the file lands.
The five numbers to measure
Agree on these before the first dial, write the definitions down, and code the dialer’s dispositions to match. If the disposition set cannot tell a wrong party from a refusal, the test cannot be read.
| Measure | How to count it | What it tells you |
|---|---|---|
| Connects per caller hour | Live human conversations ÷ logged caller hours | How much of the hour the list lets your callers spend talking instead of waiting. The single best early read on a list. |
| Owner-reached rate | Connects where the person confirms they own the home ÷ connects | Whether the phone belongs to the person who can say yes. A tenant or a relative is a connect that cannot book. |
| Wrong and dead numbers | Wrong party + disconnected or not-in-service ÷ dials, logged as two separate codes | How the phone was tied to the home, and how old it is. Wrong party points at matching; dead lines point at age. |
| DNC-flagged share | Rows your own scrub removes ÷ rows delivered | Rows you paid for and can never dial. A list that loses a third of its rows to your scrub costs half again as much per callable row. |
| Appointments per 1,000 dials | Set appointments ÷ dials × 1,000 | The outcome, read last and only on enough volume (see below). Track held appointments beside it if your closers report shows. |
Two of these are about the phones and three are about the people. A list can ring well and still lose on owner-reached rate, which is the usual signature of a file that attaches several numbers per address and lets the dialer find out which one is the owner’s. How the four common definitions of “connect” differ is in what is a good contact rate; pick one and use it for both lists.
The cost-per-appointment math
Cost per appointment is everything you spent to produce the appointments, divided by the appointments:
(caller hours × loaded hourly cost + records × price per record) ÷ appointments set
“Loaded” means the caller’s pay plus whatever you spend per seat-hour on the dialer, the lines and supervision. Here is the formula run on two lists of the same 4,000 homes’ worth of territory, each dialed three passes by the same team.
| Example | List A | List B |
|---|---|---|
| Records × price per record | 4,000 × $0.04 = $160 | 4,000 × $0.25 = $1,000 |
| Dials (three passes) | 12,000 | 12,000 |
| Caller hours at 60 dials an hour | 200 | 200 |
| Caller cost at $20 an hour, loaded | $4,000 | $4,000 |
| Connects (6% vs 9% of dials) | 720 · 3.6 per caller hour | 1,080 · 5.4 per caller hour |
| Owner reached (55% vs 80% of connects) | 396 | 864 |
| Appointments (1.5 vs 3.0 per 1,000 dials) | 18 | 36 |
| Cost per appointment | $4,160 ÷ 18 = $231 | $5,000 ÷ 36 = $139 |
Every figure in this table is an illustrative example chosen to make the arithmetic easy to follow. None is a benchmark, a quote or a price from any vendor, Scout Data included. Run the formula on your own floor’s numbers.
List B costs more for the data and produces appointments for about 40% less. On these example numbers the data line barely moves the result; the caller hour moves everything. That is the whole argument against buying on price per record, and it holds on almost any realistic set of inputs, because a caller hour costs more than a lot of rows.
If your floor reads the funnel the other way, starting from the appointments you need and working back to dials, how many dials per appointment has that arithmetic.
How to run a fair head-to-head on your own dialer
A vendor’s sample dialed once by your best caller proves nothing. A head-to-head is a small experiment with one variable, the list, and everything else held still. The checklist:
- Split the same territory. Both lists cover the same ZIP codes, or the test measures neighborhoods. If one vendor cannot cover the other’s ZIPs, cut both down to the overlap.
- Same callers. Every caller works both lists. Never the strong team on one file and the new hires on the other.
- Same hours. Interleave the lists inside every shift, by alternating hour blocks or a mixed campaign, rather than one list in the morning and the other in the evening. Time of day moves answer rates on its own.
- Same dialer settings. Pacing, attempt limits, calling windows by time zone, and the caller-ID pool. A fresh number pool on one campaign beats a tired one regardless of the data.
- Same scrub. Run both files through your own suppression and DNC process before dialing, and record how many rows each lost. That is your DNC-flagged share.
- Separate dispositions for owner reached, wrong party, disconnected, no answer and refusal. Collapsing wrong party into “not interested” hides the most important difference between files.
- Enough volume for the last number. Connects, owner rate and dead lines are readable after a pass or two. Appointments are not: at a handful per thousand dials, a few thousand dials leave the difference inside the noise of one good week. Read the early rates first, and let appointments per 1,000 dials confirm on the full run.
The narrower version of this test, built for a free sample of a thousand rows, is in free lead samples: how to test a vendor. What to check on the first paid delivery, once a list wins, is in how to buy call center leads.
A DNC scrub is not consent
One more measure belongs in the head-to-head, and it is not a rate. The National Do Not Call Registry accepts mobile numbers as well as landlines, and a seller has to sync its lists with the registry at least every 31 days. A scrubbed file has had registered numbers taken out on the day of the scrub. That is all a scrub does.
It does not make dialing mobiles compliant. Under the federal rule, a telemarketing call to a mobile placed with an autodialer or an artificial or prerecorded voice needs the called person’s prior express written consent, and a purchased list does not carry that consent with it. State law can add rules of its own on top. Which rules your floor falls under depends on how your dialer places calls and what your callers say, which is a question for counsel, not for a list vendor. This section is a summary of public rules, not legal advice. The longer version is in TCPA-compliant homeowner leads.
Skip tracing and choosing the homes are different jobs
Many floors buy homeowner data through a skip-trace tool: upload addresses, get back names and phones. That answers one question, who lives here, for addresses somebody else already chose. It leaves the three questions that decide cost per appointment to the dialer: which homes are worth a call for this pitch, whether the person who answers is the owner, and whether the mobile still works. The floor pays caller hours to find out, which is exactly the cost the table above prices.
Skip-trace tools tell you who lives there. Scout Data tells you which homes to call, whether the person is the owner, and whether the mobile is live: one active mobile per home, matched by name to the owner on title, scrubbed against the Do Not Call registry before delivery. How the two products differ in general is in skip tracing vs data append.
Reading the result
| What you see | What it usually means |
|---|---|
| More connects per hour, owner rate flat | Fresher phones, same matching. Worth it if appointments follow. |
| Connects similar, owner rate higher | Better matching. Usually shows up in appointments before it shows up in connects. |
| More connects, more wrong parties | Several numbers per home, address-matched. The dialer is doing the vendor’s matching on your clock. |
| Large share lost to your scrub | You paid for rows you cannot call. Price the list per callable row, not per delivered row. |
| Everything about equal | The files are equivalent on the phones. Terms, pitch fields and price decide it, in that order. |
Keep the write-up: two lists, the same weeks, the callers and hours named, the five numbers side by side. A vendor that wins in one quarter can be held to the same numbers in the next, and the lead source ledger is where that comparison lives once the test is over.
Frequently asked questions
How should a call center compare homeowner data vendors?
On cost per appointment, measured on your own dialer. Split the same geography between the two lists, dial both with the same callers in the same hours, and count connects per caller hour, the share of connects that reach the owner, wrong and dead numbers, the share of rows your own DNC scrub removes, and appointments per 1,000 dials. Price per record is the smallest line in that math.
Why is price per record a bad way to buy call center data?
Because caller hours cost far more than records. A floor that pays a caller for an hour of dialing spends more on that hour than on hundreds of rows, so a cheap list that wastes the hour on wrong parties and dead lines costs more per appointment than a dearer list that reaches owners.
Does a DNC scrub make it legal to dial mobile numbers?
No. A Do Not Call scrub removes numbers on the registry; it does not supply consent. Under the federal rule, a telemarketing call to a mobile made with an autodialer or a prerecorded voice needs the called person’s prior express written consent, and state law can add more. How your dialer places calls decides which rules apply, so take the setup to counsel. This is not legal advice.
What is the difference between skip tracing and buying a homeowner list?
Skip tracing starts from an address you already chose and returns who lives there and numbers that may reach them. A homeowner list built for a call center starts earlier: it chooses which homes are worth a call for your pitch, ties the phone to the owner rather than any resident, and says whether the mobile is live.