Nobody has to type the lead in

The case for automatic pipeline capture is not really about pipelines — it is about how fast a lead goes cold once the conversation that created it has ended. The most-cited number here is over fifteen years old and still holds up: the 2007 MIT/InsideSales.com study of six companies and more than 15,000 leads found contact odds fall roughly 100x and qualification odds roughly 21x between a five-minute follow-up and a thirty-minute one. A pipeline board that only fills when someone remembers to type the deal in manually reintroduces exactly the delay that number is about — the enquiry sits in a notebook, or a person’s memory, until the next quiet moment, which for a busy front desk can be hours.

What “fills itself” means concretely is narrower than it sounds, and the limits list above says so on purpose. The assistant opens and advances deals on evidence from a real conversation, inside columns you have explicitly marked as its own — it does not score leads, it does not assign deals to a teammate, and it never sets Won or Lost, because those are calls a business owner makes, not a pattern-match a model should be trusted with. The board is closer to a very fast, very literal note-taker than a sales strategist: it writes down what happened and moves the card the same way a diligent junior would, with the reason attached to every move so it can be checked.

The nurture behaviour — reviving a stalled enquiry with a spec sheet or a comparison rather than another “just following up” message — is deliberately the most restrained part of the product, and that restraint is the whole design. It reaches WhatsApp only, at most three times ever, at least a week apart, sharing one frequency limit with outbound campaigns so a customer cannot receive a nurture message and a marketing broadcast on the same day and call it spam correctly. A hot, live conversation gets none of it, because a live conversation already has the ordinary follow-up a human would give it — automating the part that is already working is how a genuinely useful nudge turns into noise.

For a business already running a CRM, the honest positioning is: this is not a general-purpose CRM. There is no lead scoring, no multi-pipeline support, and no per-teammate identity yet, so a deal’s history says “you” rather than naming which colleague acted. What it is built to do well is the narrower job of turning a conversation an assistant already had into a visible, auditable next step — for a business whose deals mostly start as an enquiry on the phone, WhatsApp or web chat, that is the gap a generic CRM cannot close on its own, because a generic CRM was never listening to the conversation in the first place.

What moves a card, and who is allowed to move it

Action	Assistant	Person
Open a new deal from a conversation	Yes, in a mapped column	Yes, any time
Advance a deal on new evidence	Yes, in assistant-worked columns only	Yes, any column
Move a card manually	Never overrides a human move	Yes — sets a hold on that deal
Mark Won or Lost	Never	Only a person, always
Send a nurture file to a stalled deal	Yes — WhatsApp only, ≤3 times, ≥7 days apart	Yes, any time, any channel

Resolved from Kooday’s own product design (facts inventory §14) — this is not a competitor comparison; it is the accountability split the pipeline enforces regardless of plan.