8:04 on a Tuesday. The first connect of the morning block comes through, the record for a house on Alder opens on the screen, and in the agent’s ear there is a click and then a woman’s voice saying “Hello?” The agent starts the opener. The woman says “Hello?” again, a little sharper. Then there is a second click, and the line is dead, and the dialer logs a contact with a duration of four seconds.<\/p>\n
The agent said their name. The woman never heard it. The wireless headset, idle since the last call twenty minutes earlier, had gone to sleep to save its battery, and it took most of a second to wake up and open the microphone when the call connected. By the time the agent’s voice reached the woman on Alder, she had said hello twice into silence, concluded she was on the receiving end of a robocall, and hung up, which is exactly what every homeowner in the country has been trained to do. The opening lines<\/a> were fine. They were never delivered.<\/p>\n The Telemarketing Sales Rule defines an abandoned call as one in which a live person answers and is not connected to a representative within two seconds of finishing their greeting, at 16 CFR 310.4(b)(4)<\/a>. The rule is about dialers that connect nobody, and an agent who is on the line but inaudible has not legally abandoned anything. The two seconds matter anyway, for a different reason: that is about how long a person will tolerate silence after saying hello before they decide nobody is there. Every homeowner has hung up on that silence hundreds of times. A headset that adds a second of wake-up delay, a network that adds half a second of buffering, and an agent who says “hello?” back instead of their name have between them spent the whole window, and the call is over before the script begins.<\/p>\n So the audio chain, from the agent’s mouth to the homeowner’s ear, has one job, which is to deliver the first word inside that window, clearly, every time. Everything below is in service of that.<\/p>\n For a calling block, wired beats wireless, and it is not close. A wired headset has no battery to save, no sleep state to wake from, no radio link to renegotiate when the microphone opens, and no other device to have quietly paired itself to. Wireless headsets are designed for people who take a call every twenty minutes and walk around between them. A calling block is the opposite pattern, a call every ninety seconds for an hour at a desk, and the power-saving features that make a wireless headset pleasant for the first person cost the second person the opener on the first connect after every pause.<\/p>\n If wireless is the only option, turn off every power-saving setting the headset and the computer offer, keep the headset on the charger between blocks rather than in a drawer, and make a test call at the start of the block so the first wake-up happens on your own number rather than a homeowner’s. Then accept that it will still happen sometimes.<\/p>\n Two ears or one is a matter of taste, with one honest note: a headset that covers both ears makes you speak more quietly, because you cannot hear yourself, and homeowners on a cold call respond better to a voice at a normal conversational level. Many agents leave one ear open for exactly that reason.<\/p>\n The microphone belongs at the corner of the mouth, an inch or so away, not in front of the lips, where every breath and every hard consonant becomes a pop in the homeowner’s ear, and not down by the chin, where the voice arrives thin and distant. Set it once and leave it.<\/p>\n Noise is the second problem, and the culprits are the keyboard and the room. An agent typing the call record while the homeowner is still talking is transmitting the keystrokes, and a homeowner who hears typing knows they are being logged. Type after the call, per the call record<\/a>, or type quietly, or mute for the two seconds it takes. The room matters more on a team: three agents on a multi-line dialer in one open room produce a background of other people’s openers, and a homeowner who hears a call center hangs up on a call center. Noise suppression in the operating system or the browser helps with steady noise and does little for other voices. Distance and a door do more.<\/p>\n A dialer that runs in the browser carries the call over the same connection as everything else on the machine, and the audio suffers first when that connection is busy. A wired network connection is better than wireless for the same reason a wired headset is: nothing to negotiate, nothing to drop. On wireless, the failure is usually not bandwidth but jitter, the variation in how long each packet takes, which arrives in the homeowner’s ear as a voice that stutters or a half-second of delay that makes both people talk over each other. Close the video call that is still running in another tab. Pause the cloud backup. If the block matters, plug in.<\/p>\n In the browser itself, three things go wrong and each has a thirty-second fix. The browser has to have permission to use the microphone, which it asks for once and remembers, and which another application can silently take away. The dialer has to be pointed at the right input and output devices, which changes every time a headset is unplugged or a monitor with speakers is connected, and the symptom is an agent talking into a laptop microphone from three feet away while the homeowner strains to hear. And a second tab running the same dialer, or a video application still holding the microphone, will produce an echo or a dead mic that looks like a network problem and is not one. Check the device list before the block. It takes ten seconds.<\/p>\n Before every block, not just the first one on a new setup, call your own cell phone from the dialer, put the cell on speaker, and deliver the opener as if it were real. Listen for four things: whether your first word arrived immediately or after a pause, whether your voice is at a normal level, whether there is any hum, echo or keyboard, and whether there is a delay between your mouth and the phone that would make a homeowner talk over you. Then hang up and fix whatever you heard.<\/p>\n That is the whole test, and it catches the sleeping headset, the wrong input device, the video call still holding the microphone, and the congested connection, which between them account for most of the four-second contacts in a week’s records. The setup checklist<\/a> puts this on the list once; it belongs on the list every morning.<\/p>\n On a multi-line dialer the connect arrives as a tone and a record opening, and the homeowner has usually already said hello once by the time the agent hears anything. The instinct is to say “hello?” back, which spends another second establishing nothing. The rule is to say your name into the connect tone, immediately: “Hi, this is [your name] with [your brokerage].” If the homeowner was mid-hello, you talked over half a syllable, which is fine. If they were already reaching for the button, your name arrived in time to stop them. On a single line the same rule applies with less urgency, because you heard the ring and the pickup and you know exactly when the greeting ended.<\/p>\nTwo seconds<\/h2>\n
The headset<\/h2>\n
The microphone<\/h2>\n
The network and the browser<\/h2>\n
The sixty-second test<\/h2>\n
Say your name first<\/h2>\n
\n\n
\n \nWhat the homeowner hears<\/th>\n Likely cause<\/th>\n Fix<\/th>\n<\/tr>\n<\/thead>\n \n Silence for a second, then a voice<\/td>\n Wireless headset waking from idle; agent saying “hello?” before the name<\/td>\n Wired headset; power saving off; name first into the connect<\/td>\n<\/tr>\n \n A thin, distant voice<\/td>\n Laptop microphone in use instead of the headset; boom too low<\/td>\n Check the input device before the block; boom at the corner of the mouth<\/td>\n<\/tr>\n \n Pops on hard consonants<\/td>\n Boom directly in front of the lips<\/td>\n Move it to the side, an inch off<\/td>\n<\/tr>\n \n Stutter, dropped words, talking over each other<\/td>\n Wireless network jitter; another application using the connection<\/td>\n Plug in; close the video call; pause the backup<\/td>\n<\/tr>\n \n Typing<\/td>\n Agent logging the record during the call<\/td>\n Type after; mute for the keystrokes<\/td>\n<\/tr>\n \n Other agents’ voices<\/td>\n Open room, multi-line dialer, no separation<\/td>\n Distance, a door, or a directional microphone; noise suppression will not fix voices<\/td>\n<\/tr>\n \n Echo<\/td>\n Two tabs or two applications holding the microphone; speakers open<\/td>\n One dialer tab; headset output, not speakers<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n The block that follows<\/h2>\n