Inbound leads qualified, researched and routed in four minutes, with the reply still a person’s call.

A B2B software company selling into both enterprise and mid-market had an inbound problem that looked like a volume problem. Web-form submissions arrived in HubSpot all day. A large share were spam and low-quality requests. The rest were real buyers from two very different segments, each with its own reps and its own sales process.

Every submission had to be read by an SDR before anything could happen to it. Is this a real company? Is this person someone who buys? Enterprise or mid-market? Who should take it? Only then a reply, written from a template because there was no time for anything else.

The visible symptom was speed. A lead that came in while the team was on calls waited hours, and inbound buyers do not wait. Time-to-lead was the metric leadership watched, and it moved with how busy the SDRs happened to be.

The expensive symptom was where the SDR hours went. Most of the reading was spent confirming that a submission was junk, which is work nobody should be paid to do. The share that was real got the same generic sequence as everyone else, so the company’s warmest leads received its least specific emails.

The arithmetic was never done, so we did it. At several hundred submissions a week and a few minutes of reading each, the triage alone was on the order of a working day of SDR time every week, by our estimate, most of it spent on submissions that went nowhere. That is before anyone wrote an email.

The team’s enrichment platform could have done some of this, but it synced with HubSpot once an hour. For inbound, an hour is the whole problem. The usual alternative, a rules engine on the form fields, handles the obvious cases and confidently mishandles the ambiguous ones, and the ambiguous ones are the expensive ones.

We mapped what an SDR actually decides between a form landing and a reply going out. Most of it is research: who is this, what do they do, what size, have they been in touch before. Three decisions carry the weight. Is this a real buyer? Who should own it? And what do we say to them?

The first two are high-judgment and low-stakes. Misjudge a lead and it costs an internal hop, not a customer, and the mistake is visible and reversible. That is where a coworker can run on its own, provided it writes down why it decided what it decided. The third is different. An email to a prospect cannot be unsent, and it carries the company’s name. So the coworker drafts it and a rep approves it. Every time.

The coworker picks up a submission the moment it lands, which meant building a direct link between HubSpot and the enrichment platform instead of waiting for the hourly sync. It identifies the person’s role by checking four data sources in turn, because any single source fails on exactly the contacts that matter. Spam and non-buyers are marked and parked. Real buyers get researched: what the company does, recent changes and announcements, hiring, and anything that suggests why they reached out now. It scores fit against the ICP the team already used.

Then it routes. Headcount, fit score and the research together decide enterprise or mid-market, and the lead is assigned to the right rep inside HubSpot with a short brief attached: who this is, what they likely want, and what to open with. For a real buyer it drafts a meeting invitation written from the research. For a submission it cannot place with confidence it drafts a short clarifying email instead of guessing. Both drafts wait in the rep’s queue.

A submission now goes from form to a qualified, researched, routed lead with a draft reply waiting in about four minutes. It used to take hours and depended on who was free. The SDRs stopped reading spam. Each segment’s reps see only their own leads, and the email a buyer receives is about their company rather than about the product.

Against a conservative two-hour baseline, four minutes is about a 98 percent cut in the time a buyer waits, and the wait no longer depends on the calendar. The triage hours come back to the SDRs almost entirely; what they read now is a brief on a real buyer, not a form that might be one. We did not measure conversion for this engagement and will not put a number on it. The reasoning is simple enough to stand on its own: the same lead, reached in minutes with a specific email, converts better than the same lead reached the next day with a template.

The number we watch first is how often reps edit or reject a draft, and what they change. A draft that gets sent unchanged is evidence. A pattern of the same correction is a change to the decision map. Repeat misroutes of the same shape usually mean the segment boundary has moved, and the business moves it more often than anyone announces.

The reply stays behind approval on purpose, and not because the drafts are poor. Enough approved-unchanged drafts would be the evidence to let some categories of reply go out on their own, and the first badly received one would be the reason to pull that back. Autonomy moves on the record, in both directions, and deciding when it moves is part of the monthly service.