Gatekeepers vs decision makers — you were right to ask

884 July transcripts, classified by who actually picked up, joined to line type. Your instinct that reach rate was hiding the truth was correct. The conclusion isn't the one either of us expected.

You said: a gatekeeper always picks up, so raw reach is misleading — and mobiles win because they put you straight through to the person. Half right. Raw reach was indeed misleading. But mobiles don't win. On decision-maker conversations per dial, VoIP business lines beat mobile by nearly 3× — because they answer far more often, and their gatekeepers pass you through 94% of the time.

The number that matters

Line type	Dials	DM convos	DM per 100 dials	GK wall	Machine
Non-fixed VoIP	216	97	44.9	3	8
Mobile	2,228	346	15.5	35	91
Fixed VoIP	26	2	7.7	1	0

A "DM conversation" = the transcript shows you actually talking to the person who handles it, whether direct or after a transfer. 2.9× more of those per dial on VoIP than on mobile.

Why VoIP wins — it's not what you'd think

When someone picks up…	Mobile	Non-fixed VoIP
Straight to the decision maker	226 (32%)	52 (34%)
Gatekeeper, then transferred through	120	45
Gatekeeper blocked you	28	3
Gatekeeper pass-through rate	81%	94%
Hit a machine / voicemail	96 (14%)	8 (5%)

You get a gatekeeper more often on VoIP (32% vs 21%) — exactly as you predicted. But they're the wrong kind of gatekeeper: they almost always hand you over. Only 3 blocked you all month.

The mechanism: "non-fixed VoIP" here isn't Google Voice junk — it's modern small-business phone systems (RingCentral, Dialpad, that class). A real business with a real phone system has someone whose job is to answer it, and in a 20-person shop that person sits ten feet from the owner. Mobile numbers, by contrast, mostly just ring out — 14% straight to voicemail and a far bigger pile of no-answers behind them.

Honest limit: 216 VoIP dials against 2,228 mobile. The gap is big enough that noise is unlikely to explain 2.9×, but treat the exact figure as loose. It's a strong lead, not a settled fact — and it costs you nothing to test.

What actually separates a booking from a callback

I compared phrase usage across 11 booked, 110 callback and 77 rejected conversations. Two moves stand out, and one thing you believe is wrong.

Move	Booked	Callback	Rejected
"Are you still the person who handles…"	27%	3%	16%
Time-boxed permission ("twenty seconds")	27%	13%	16%
Callback frame ("told to call back")	55%	46%	35%
Dollar fee figure	9%	2%	3%
"How you been?"	73%	67%	64%
Naming Paychex early	55%	45%	61%

"Are you still the person who handles it" is your single biggest tell — 9× more common in bookings than callbacks. And "how you been" is pure noise: you say it in two thirds of every category. The earlier note claiming it was 0-for-everything was measuring something else. Naming Paychex early actually skews toward rejection.

The talk track your own bookings are already using

This is reconstructed from the 11 calls that booked, in their actual order. Nothing invented.

1Hey Ernesto, this is Kevin from ADP — how you been?neutral, just warmth

Fine to keep. It isn't doing work, but it isn't hurting either — and it buys you the second before they hang up.

2I don't know if you remember — I called way earlier in the year, towards January. You told me now would be a better time to call back.proven

Own the callback, stay vague on who. Appears in 55% of bookings vs 35% of rejections. Gives them a socially easy reason the call exists.

3Are you still the person who handles the payroll account?strongest signal

The move that separates bookings from callbacks 9 to 1. It qualifies the decision maker out loud, it's a yes/no, and it presumes a prior relationship. If you add one thing today, add this.

4Let me just grab like twenty seconds to see if it's even relevant to you. Is that okay?proven

2× more common in bookings. The self-deprecating framing matters — "see if it's relevant to you" hands them the judgement instead of asking them to sit through a pitch.

5You've got roughly $13,400 a year in plan fees, and they're pulling that from your assets.rare but potent

Only 9% of bookings used a dollar figure, but that's 3× the rate elsewhere. Use it after step 3 confirms you have the decision maker — never as a comeback to a no.

6I'll send a calendar invite now, and I'll call you Wednesday just to confirm the time.closes it

Every booking ended with an invite, and two added a confirmation call. Spell the email back to them — you caught three garbled addresses that way.

One more thing worth stealing from your own tape: two of the eleven bookings came through an AI screener that asked for name and reason. Both times you gave only the name — "Kevin Peak." No pitch, no company. Both got put through. Give screeners the bare minimum; a reason gives them something to reject.

Three things you can run today — pick one

Option A — Swap the list, keep everything else

Zero change to your behaviour. Highest expected value.

Dial 40 Bay Area VoIP numbers instead of mobiles. At the observed rates that's roughly 18 decision-maker conversations instead of 6. You have 594 untouched Bay Area VoIP numbers and 2,386 in total, all with company names.

List saved: ~/voip-dm-list-2026-07-29.csv and OneDrive → Documents → ADP. Bay Area area codes sorted to the top.

Option B — Add one sentence

Zero change to your list. Cleanest experiment.

Keep dialing exactly what you're dialing, but say "Are you still the person who handles the payroll account?" in the first fifteen seconds of every single conversation. It's in 27% of your bookings and 3% of your callbacks. Track how many say yes — that's your real DM-contact count, which nothing in the dialer currently measures.

Option C — Split the block and settle it

Most information, slightly more admin.

20 VoIP and 20 mobile, back to back, same script, and write down decision-maker conversations for each. One session ends the argument about whether the 2.9× is real. If it holds, your whole list strategy changes permanently — and you'd know by tonight.

What I'd actually do

Option A plus the one sentence from B. The list swap is where the leverage is — nearly 3× the decision-maker conversations for no behavioural change at all — and adding the "still the person who handles it" question costs you nothing and is the strongest single predictor in your own data. Save Option C for tomorrow if today's numbers look too good to believe.

And set the target honestly: 40 VoIP dials should produce somewhere near 18 decision-maker conversations and 2–3 callbacks. Not a meeting — you're at roughly 110 dials per booking. Conversations and callbacks are the day's actual output.