Cold calling
The phone is where prospecting data goes to die and where deals get made, in that order, and this is where the two halves of a call finally meet. The dialler buys conversations per hour. The recorder turns what was said into something the rest of the business can use. Neither is worth much without the other: volume without a record produces nothing you can coach, and a record of calls nobody makes is an empty archive. Our own volumes, measured: 60 calls a day dialling by hand, 150 to 200 with a power dialler, and up to 500 with one that opens several lines at once. The pick-up rate does not move, and that is exactly the point, because the same share of people answer and more calls mechanically means more conversations. It only pays if the list is big enough to keep those lines busy, which is the condition everyone skips. Expect around 32 dollars per user per month for a team phone system, three times that for the tiers carrying AI agents, and around two hundred euros a seat for a parallel dialler. On the recording side, the thing that makes it pay is asking the same question of every call automatically, so the archive answers instead of waiting to be searched. And the schedule matters more than the software: on roughly four thousand of our own calls, the morning returned 1.85 times what the afternoon did. Protect the morning before optimising anything else.
Protect the morning before you buy anything
Cold calling is where prospecting data goes to die and where deals get made, in that order. The software decides how many conversations an hour is worth. It cannot decide whether the hour was worth calling.
On roughly four thousand of our own calls, the morning returned 1.85 times what the afternoon did. That ratio is free to measure on your own team and it moves more than any licence. The measurement is in the best hours to call, measured.
Our volumes, in the same hands: 60 calls a day dialling manually, 150 to 200 with a power dialler, and up to 500 a day with one that opens several lines at once. The pick-up rate does not move between those three, and that is the whole point, because the same share of people answer and more calls mechanically means more conversations. It only pays if the list is big enough to keep those lines busy, which is the condition everyone skips, and the arithmetic is in parallel dialling maths.
Two different products, then. Allo is a team phone system at 32 dollars per user per month billed yearly, with the AI agents sitting on a 100 dollar tier, and its power dialler tops out around 150 calls a day in our hands. Sonik is 200 euros per user per month for the dialler and buys conversations per hour rather than a licence. Both read 18 September 2026. Louis puts it this way: Allo is a swiss army knife, Sonik is a hunting knife. Both are on Dialers and phone.
All Dialers toolsAsk the same question of every call, automatically
A recording nobody opens is an archive, and an archive answers nothing. What makes this layer pay is asking the same question of every call without anyone deciding to, so that the answer is already there when you go looking.
Praiz publishes thirty euros per user per month on an annual contract, one plan, recordings, transcriptions and agents listed as unlimited, read 18 September 2026. That last part is what decides it: the cost does not scale with how much you actually use the thing, which is rare enough in this category to be the reason to choose it.
One warning that cost us real time. A transcript is a source, not a verdict. A call that shows as recorded can still come back with an empty transcript, so anything automated on top of it has to treat empty as a failure and never as an answer. A scoring agent that reads an empty transcript will happily invent a reason the call went badly.
The other half is who it is for. A rep asked to choose between the next call and writing up the last one chooses the next call, every time, and that is not a discipline problem. The only reliable record is the one that arrives on its own. How we actually use it, and what gets ignored, is in call recording that gets used. What it costs is on Conversation intelligence and call recording.
All Call recording toolsBuy the hours before the software. A dialler makes an hour productive and cannot make it valuable, and the 1.85 ratio between our mornings and our afternoons was free to find. Measure your own version of it before you spend two hundred euros a seat.
The categories in here
What I wrote about them
5 pieces
- The best hours to call, measured on four thousand calls
The morning returned 1.85 times what the afternoon did. Everything else about the dialler matters less than that ratio.
- Parallel dialling, in numbers: when five lines pay for themselves
Two hundred euros a seat sounds absurd until you count conversations per hour instead of licences per month.
- The call archive nobody opens, and the three questions that make it worth paying for
Recording every call is easy and nearly worthless. What makes it pay is asking the same question of every one of them.
- Gong alternatives for small sales teams: who publishes a price, and who makes you ask
The single test to run on your own calls before you sign anything.
- Gong alternative: what we switched to, and what it cost
We run Praiz on client accounts and we left Gong behind.