Sales Development: Which of These Leads Are Real?

On the Monday of launch week Curtis Nwosu is looking at an empty queue and thinking about a full one. In four days the campaign goes live, and over two quarters it is supposed to deliver 760 MQLs to his team of eleven, of which 190 are meant to become opportunities somebody in sales will work. That is a quarter of them. Which means three quarters of everything his reps touch will not become anything, and how quickly and how honestly they establish that is most of his job.

He has one advantage the previous three launches did not give him. Dana made him sign the MQL definition in week eleven, before a single lead existed, alongside Ryan. He has read it. He agrees with it. That sounds like a small administrative fact and it removes the single most expensive argument in business-to-business marketing.

What lands on Curtis’s desk

From Dana in module 10 comes the plumbing: leads arriving in named queues with the source recorded, the score attached, and — the detail that makes everything in this module possible — which rule fired to make each one an MQL. He also receives the signed definition of what an MQL is, including the exclusions.

From Wes in module 2, carried through Ryan’s plan and published on the website rather than hidden, come two limits. Supply Signal has no native SAP connector at launch; it ships the following quarter. And implementation takes six weeks, not an afternoon. Most sales development managers would treat those as obstacles handed to them by an honest product manager. Curtis treats them as the most useful thing in the folder, and the reason why is worth the whole section it gets below.

What a sales development manager actually does

A sales development representative — an SDR — is the first human voice a prospect hears. They do not close deals. They make contact with people who have shown some sign of interest, establish quickly whether there is a real problem, a real budget and a real timeline, and hand the ones that qualify to a salesperson who will spend weeks on each. Their manager’s job is to design that filter: who gets contacted, how fast, in what words, and what standard a lead has to clear before it costs a salesperson a fortnight.

The whole discipline turns on one asymmetry. An SDR call costs perhaps fifteen minutes. A salesperson working a bad opportunity costs weeks, and worse, it appears in the pipeline while it does so.

The software on Curtis’s desk

The CRM queue is the whole job in one screen, and the field that matters most is the smallest: the plain-English reason this person arrived. A rep who knows someone downloaded the supplier-risk guide opens a different conversation from one who knows nothing — which is the same principle as the alert reason in the first case study, arriving in a different department.

Data providers answer fit before anyone dials, which is what stops a rep spending a morning on a company too small to buy. The sequencer enforces speed and records what was attempted, and it is where the honest agent question sits: an agent can prepare a well-researched first message in seconds, and whether it sends that message as a person is a decision the company should take deliberately rather than inherit from a default setting. The rejection log with its closed list of reasons is what turns SDR frustration into something marketing can act on.

The four decisions

Fit versus intent

These are two different questions and almost every lead-scoring model in existence blends them into one number, which is the commonest mistake in the discipline.

Fit asks: should we sell to them? A discrete manufacturer in Michigan turning over $180m is a good fit whether or not they have ever heard of Cadence. Intent asks: are they ready to buy? Someone who visited the pricing page three times in a week, came back to the supplier-risk guide and registered for the webinar is showing intent, whatever company they work for.

Collapse the two into a single score and something quietly absurd happens. A perfect-fit manufacturer idly reading one article scores 62. A twelve-person machine shop that has consumed every asset on the site and requested a demo also scores 62. The rep sees two identical numbers describing two situations requiring opposite actions, and the score has destroyed exactly the information needed to tell them apart — because a fit problem and an intent problem look the same once they have been added together.

So Curtis keeps them on separate axes, and there are four squares.

High fit, high intent is the rare and obvious one, and it is where speed matters more than anything else in this module. Contact immediately, by phone, and expect a real conversation. Roughly one lead in eight lands here.

High fit, low intent is the largest and most valuable group, and it is the one teams destroy. These are exactly the companies Camille chose in module 5 — mid-market manufacturers whose real competitor is a spreadsheet and who are therefore not looking for anything. They belong in nurture: content, sequence, patience, and a light-touch check every few months. Phone them weekly and you will not accelerate them; you will convert a future customer into someone who will not take your call in eighteen months, which is a cost that never appears in any report.

Low fit, high intent is the seductive square, and it is where SDR time evaporates. The forty-person shop that read everything, the student writing a dissertation, the consultant doing competitive research for somebody else. The behaviour is flawless and the company can never buy. Reps love these leads because they answer the phone and are enthusiastic. Curtis’s rule is a polite decline, a pointer to self-serve material, and no further spend — and he is explicit with his team that ignoring an enthusiastic person is a skill, not a rudeness.

Low fit, low intent gets suppressed. It should never have entered the queue, and if it is arriving in volume from one channel, that is a targeting problem for Ryan, not a work-rate problem for the reps.

The MQL argument, settled

The oldest argument in business software runs like this. Marketing counts leads that match the profile and did something. Sales counts leads that were worth picking up a phone for. Both are counting honestly. The numbers differ by half, and every quarter the two teams meet and each explains that the other does not understand the funnel.

What settles it is not goodwill. It is the signed definition Dana forced in module 10, which has fit criteria, behaviour thresholds and — the part that does the work — a written list of what does not count. Curtis contributed the exclusions, because exclusions are the thing sales knows and marketing cannot see: existing customers whose colleague filled in a form, competitors, the same person submitting from three email addresses, companies below the segment floor.

But a shared definition alone decays, because no definition written in advance survives contact with two thousand real people. What keeps it alive is the second half of Curtis’s decision, and it is the most portable idea in this module: every rejected MQL is rejected with a reason, from a fixed list.

Not free text, which nobody analyses, and not a single “not qualified” flag, which is the same as silence. A closed vocabulary of about ten reasons: wrong company size, wrong industry, wrong role, existing customer, competitor or consultant, no budget this year, timeline too short for a six-week implementation, runs SAP and needs it now, evaluated and staying with the spreadsheet, and unreachable after six attempts.

Look at what that list separates. “Wrong company size” is a targeting failure and belongs to Ryan and the channel plan. “No budget this year” is not a failure at all — it is a good lead badly timed, and it should go back into nurture with a date on it rather than into a bin. “Unreachable after six attempts” is neither; it is an operations failure, possibly a wrong phone number field, possibly a lead that sat in a queue for three days. A single rejection flag merges three completely different problems into one number and makes all of them invisible.

The loop closes monthly. Curtis and Ryan look at rejection reasons cut by channel, and the pattern speaks for itself. If sixty per cent of one channel’s rejections say wrong company size, the advert is reaching the wrong people and no amount of coaching the reps will help. If they say no budget this year, the channel is fine and the nurture track is where the money is. Rejection without a reason teaches marketing nothing, which means the definition never improves, which means the argument recurs next quarter with the same two honest people making the same two honest cases.

Routing and speed

The single most reliable finding in sales development is unglamorous: the sooner you contact someone after they raise their hand, the more likely you are to have a conversation, and the decay is steep. A lead contacted within five minutes converts to a conversation dramatically better than the same lead contacted the next day, and by the following week much of the advantage is gone.

The reason is not mystical. The person filled in that form because at that moment they had a question. Ten minutes later they are in a meeting; tomorrow the question has been absorbed back into a job that contains four hundred other questions. There is also a competitive edge to it: someone researching supplier risk has usually filled in three forms, and whoever calls first gets to frame the category — the second caller is answering questions the first caller planted.

Then comes the operational reality, and this is where most teams quietly fail. Leads arrive at 07:40, at 23:15, and on Saturdays. Reps are on calls, in one-to-ones, at lunch. A five-minute response is not a policy you can announce; it is a rota, an alert that reaches a phone rather than an inbox, a queue somebody is actually watching, and protected time in which the watching rep is not expected to be doing anything else. Curtis builds exactly that: four-hour duty blocks with one named rep on first touch, alerts pushed to a phone, and high-fit high-intent leads jumping the queue automatically because Dana’s scoring writes the square into the record.

And he publishes an honest target rather than an impressive one: under fifteen minutes during business hours, first thing the next working morning otherwise. A team that promises five minutes and averages nine hours has a worse system than a team that promises an hour and hits it, because the second team can see when it is failing.

The obvious question is whether an agent should do the first touch, and Curtis draws a line rather than a conclusion. What agents do genuinely well here is preparation: reading the account, summarising what this person has actually done — which assets, which pages, which webinar — pulling the company’s recent news, and drafting an opening that refers to their situation rather than to Cadence’s product. Done at volume, that turns fifteen minutes of pre-call research into ninety seconds, which is a real gain and is the difference between a rep making thirty prepared calls a day and forty generic ones.

Where he stops is the point at which an agent stops assisting a person and starts impersonating one — sending messages under a rep’s name, holding a back-and-forth exchange with a prospect who believes they are talking to a human being. That may be a defensible thing for a company to do, and it may not, but Curtis’s position is that it is a decision Cadence should take deliberately, with Delia and Miriam in the room, rather than inherit because a vendor shipped it switched on. His own rule for this launch is short: agents prepare, humans contact.

There is a volume argument too, and it is worth stating because it is not obvious. An agent that can send four thousand personalised emails does not make four thousand emails a good idea. The constraint on outbound was never typing speed — it was the recipient’s patience, the sending domain’s reputation, and the fact that a market of 14,000 companies is small enough that burning it is permanent.

What SDRs need from the campaign

Here is a rep’s morning. Forty names on a screen, each with a score of 71 or 84, and nothing else. The call opens: Hi, I’m calling from Cadence Systems, we make supply chain risk software — is that something you’re looking at? It is a fine script and it is a cold call, and the person on the other end treats it as one, because from their side it is indistinguishable from one.

Now the same morning with one field filled in. You downloaded our guide on single-source component risk on Tuesday. Most people who read that one have a supplier they have been quietly worried about for months. That is not a better script. It is a different conversation, because it begins from something true about them, and the honest answer to it is usually the thing the rep needed to know.

So Curtis’s requirement of the campaign is one line: the reason a lead arrived, in plain language, on the record. Not the score. Not the channel code. The human-readable sentence — attended the risk webinar and stayed for the questions, read the SAP integration page twice, requested the pricing sheet, scanned their badge at the booth and asked about board-level reporting.

Readers of the first case in this collection will recognise the principle exactly. Maya’s rule there was that no alert ships without a reason attached, because a churn score of 0.81 tells a customer success manager how worried to be and nothing whatever about what to do. A lead score does the same damage in the same way: it tells a rep how urgently to be generic.

And this is where Wes’s two awkward facts become an asset. Curtis makes both mandatory questions in the first call — what do you run for ERP, and when do you need this live — precisely because they disqualify people. A prospect who runs SAP and needs it working in four weeks is out in minute six, at a cost of six minutes. The same discovery in week six of a sales cycle has consumed a rep, a sales engineer, a demo, a security questionnaire and a buying committee that has already told its own leadership about a product it cannot have. That prospect does not return next quarter when the connector ships; they do not remember a technical constraint, they remember being misled.

At a planned $6,300 per opportunity, an opportunity that cannot close is worse than no opportunity, because it costs money, occupies capacity and inflates a pipeline number that Lena will have to explain in module 12. Early disqualification is a gift, and Curtis says so to his team in those words — then makes the SAP cases into a dated nurture list rather than throwing them away, because next quarter they become good leads with no acquisition cost at all.

Where this goes wrong

The commonest failure is an SDR team compensated on meetings booked. Reps are not dishonest; they are responsive. Meetings get booked with anyone who will agree to one, sales accepts them because they arrived from marketing, and three weeks later the pipeline contains twenty deals that were never real. The number everyone celebrated in month one is the number that produces the miss in month four. What Curtis measures instead is opportunities accepted by sales and still open thirty days later, which is slower to report and much harder to fake.

The second failure is cherry-picking. Given a queue, reps work the leads that answer the phone and leave high-fit, low-intent companies to rot — the exact square containing most of the campaign’s future revenue. The remedy is structural rather than motivational: those leads leave the call queue entirely and go into a nurture track that is somebody’s job to own.

The third is the silent rejection. Leads marked “not interested” with nothing behind it, month after month. Marketing sees a low acceptance rate, concludes sales is not working the leads, and sales concludes marketing is sending rubbish. Both are reasoning from an empty field.

Designing this desk’s agent: the lead triage agent

Curtis’s agent is the one most companies build first and examine least. It is also the only agent here whose errors are invisible by construction — nobody complains about a call they never received.

One concrete design choice. Keep fit and intent as separate scores rather than blending them, and force the reason onto a closed list. A single number destroys the information that distinguishes wrong company from right company, wrong month — and those two need opposite treatment.

{
  "lead_id": "ld-88213",
  "is_valid": true,
  "fit_score": 0.81,
  "intent_score": 0.34,
  "reason": "fit_strong_intent_early",
  "next_step": "nurture_sequence_b",
  "rejected_reason": null,
  "model_version": "triage-2026.06",
  "decided_at": "2026-07-01T08:12:00Z"
}

The metric to track. Precision and recall on qualification — and recall matters more here than precision, which is the opposite of most people’s instinct. A lead wrongly passed to a rep costs six minutes. A lead wrongly rejected is never seen again by anyone, so the error is permanent and unmeasured. Sample rejected leads deliberately and work a portion of them, or you will never discover the segment the agent has quietly been declining.

What Curtis hands on

To sales go accepted opportunities, each carrying the channel it came from, the human-readable reason the person raised their hand, the answers to the ERP and timeline questions, and the record of what has already been said. A salesperson opening one of these does not begin at zero.

To Lena Park in module 12 goes something less obvious and analytically more valuable: the rejection data. Every MQL that did not become an opportunity, with its reason and its channel. That dataset is the only way to tell a campaign that produced the wrong people from a campaign that produced the right people at the wrong time, and those two results look identical in a conversion rate.

The constraint everyone inherits is the definition itself. For the whole two quarters, an MQL means one thing, an opportunity means one thing, and nobody may quietly widen either to make a number look better — which is exactly what makes the final module’s arithmetic worth trusting.

Call, nurture, or decline?

Read each lead and decide what Curtis’s team does, then tap a card to check.