On a clean demo org, lead management looks like a solved problem. There's a Lead object with a Status picklist, a built-in conversion button, and a Web-to-Lead form generator sitting right there in Setup. You wire up a form, leads flow in, reps work them, done. Then you put it in front of a real sales floor and within a month you're staring at a queue of four thousand leads where nobody can tell you which ones are worth calling, half the "Nurturing" ones haven't been touched since the day they were created, and your best rep is fielding angry calls from customers who've already been phoned by two other people.
I've watched this happen often enough to treat lead management as three separate design problems that get conflated into one. Capture is about how leads enter and whether they're clean. Scoring is about which ones get attention first. Status is about the lifecycle and the SLAs that keep leads moving instead of rotting. Each layer has its own gotchas, and a mistake in one quietly poisons the others. The clearest way to talk about it is through a real shape of project: a high-volume consumer-lending operation where each relationship manager handles around a hundred leads a day and the first-contact SLA is measured in hours, not days.
Get these three layers right and the org becomes a machine that routes the right lead to the right person inside the SLA. Get them wrong and you've built an expensive database of names nobody trusts.
Capture: the form is where conversion dies first
Most capture channels are unremarkable to set up. Web-to-Lead generates an HTML form you embed on a site. Email-to-Lead parses messages sent to a dedicated address. Marketing automation syncs engaged contacts in. Reps enter leads by hand after events, ops imports purchased lists with Data Loader, and external systems like social lead-ad platforms push records in over the API in near real time. None of that is hard.
The decision that actually moves the needle is field count on the web form, and it's the one clients fight me on hardest. Marketing always wants more fields because more data sounds better. But conversion drops measurably once a form runs past roughly seven fields, and forms with fifteen-plus fields routinely convert under five percent because people fill in one or two boxes and abandon the rest. The right answer is a short form of five to seven core fields, broken into a couple of short steps rather than one intimidating wall, with the rest collected later through progressive profiling. For a property developer I'll often land on five fields total: name, phone, email, project of interest, and budget range. That's enough to score and route; everything else the rep gathers on the first call.
Capture is also where you decide how granular Lead Source is, and this is worth more thought than it gets. A flat picklist of "Website / Event / Referral" tells marketing almost nothing. Splitting it into Website (Direct), Website (Google Ads), Website (Social), Event, Referral - Customer, Referral - Employee, Data Partner, Broker, and Walk-in lets you actually see which channels produce qualified leads and which produce noise. Add a custom Sub-source field for the campaign name, broker name, or branch code underneath. You will be asked "which channel is worth the spend" within the first quarter, and you want to be able to answer it.
The last non-negotiable at capture is deduplication. A single customer will fill out five forms across three months from three different channels, and they'll use a mobile number on one and a landline on another. If your matching rule is phone-exact only, you get two leads and two reps calling the same person. Turn on Duplicate Management, match on phone or email with a fuzzy name component, and set the duplicate rule to Allow with Warning so the rep sees the existing record and can merge rather than being silently blocked. The branch-walk-in versus web-lead collision is the one that generates customer complaints, so the warning needs to surface "this person was contacted by another rep N days ago," not just "possible duplicate."
Scoring: start transparent, earn your way to AI
Lead scoring exists to answer one question: who do I call first. There are three ways to build it, and the order you adopt them matters more than which one you eventually run.
Start with a manual, rule-based model implemented as a formula field or a Flow on create/update. You assign points to attributes the business already believes predict conversion and sum them to a normalized score out of a hundred. For a consumer-lending client the weights might reward income band, a credit-bureau score above a threshold, a tier-one province, a valid local phone format, and an existing product relationship; for a property developer the heavy weights go on budget, purchase timeline, whether they've visited a show unit, and whether they already hold a deposit elsewhere. The point of starting here is that it's completely transparent. When a rep asks why a lead is an A-tier, you can show them the math, and when the model is wrong you can fix a specific weight.
Behavioral scoring is the second layer, fed from your marketing platform: points for email opens, link clicks, site visits, content downloads, webinar attendance. You sync that score back onto the lead and add it to the attribute score. It's powerful but only as good as your marketing tracking, so it's worth adding once that plumbing is real rather than aspirational.
Einstein Lead Scoring is the third layer, and it's genuinely good when the preconditions are met. It trains on your converted-lead history and predicts conversion probability with a model you don't have to maintain. The catch is that it needs a solid base of converted leads to train on, and it's a black box: you can't tune the weights, so when a stakeholder challenges a score you can only shrug. My standard recommendation is manual scoring for phase one, then Einstein layered on after six to twelve months once the org has accumulated real conversion history. Whatever model you ship, validate it against actual conversion after about three months. The most common scoring failure I see is hand-picked weights that nobody ever checks, producing high scores that don't convert and reps who stop trusting the number entirely.
Scoring only earns its keep when it drives assignment. Tie the score and geography into Lead Assignment Rules that route to queues, and distribute within the queue by round-robin or capacity. Be aware that native round-robin doesn't exist; you'll reach for a Salesforce Labs package or a custom Flow. And don't route on source or score alone the way the first version of every project does, because senior reps end up buried while new hires sit idle. Build a capacity check into the assignment from day one.
Status and SLA: keep leads moving or watch them rot
The default Status picklist of Open / Working / Qualified / Unqualified is too coarse for a real sales floor. I extend it to something like New, Attempting Contact, Working, Nurturing, Qualified, Unqualified, and Recycled. Seven values cover roughly ninety-five percent of call-center situations. Fewer and you can't track where leads stall; many more and reps stop updating it accurately, which is worse than not tracking at all.
Each status needs an SLA attached, and the single biggest mistake is applying one blanket window to everything. A flat twenty-four-hour first-contact SLA across all sources sounds disciplined, but when purchased-list leads arrive in volume at lower quality, a rep physically cannot call a hundred of them a day inside that window, so the SLA becomes a number everyone ignores. Differentiate by tier: A-tier gets a few hours, B-tier half a day, C-tier a full day, D-tier no SLA and an archive timer. The SLA is now aligned with where attention is actually worth spending.
Compute every SLA against Business Hours, not raw elapsed time. A web form submitted at 10pm on a Friday should not start a twenty-four-hour clock that's already half-blown by the time anyone is at a desk on Monday. Configure a Business Hours record with the working window and holidays, store an SLA Due Date that's calculated on each status change, and run a Flow that nudges the rep at fifty percent elapsed, escalates to the team leader at eighty percent, and reassigns at a hundred. For genuinely hot leads you can run a 24x7 Business Hours profile with after-hours coverage, but that's the exception, not the default.
Two status fields deserve to be required, because their absence causes the ugliest failures. Nurture Date must be mandatory whenever a rep sets Nurturing, capped at six months, with a scheduled Flow that auto-recycles past the date. Without it, a large slice of leads get parked in Nurturing and forgotten forever. Unqualified Reason must be a required picklist, because recycling logic depends on it: a "wrong contact info" lead can come back in thirty days if the number gets corrected, a "no need right now" lead in six months, but a "low income" or "bad credit" lead should never recycle. And when leads do recycle, log previous owners and skip reps who've already called, or you recreate the three-reps-one-customer problem you turned on dedup to avoid. Recalculate the score on recycle too, since a six-month-old score is stale by definition.
All of this needs to surface on a dashboard the team leader opens every morning: leads in New past their SLA sorted by score so the hottest stuck leads jump out, leads stalled in Attempting Contact, leads aging in Working, and reps under target. The leader's job becomes unsticking the red leads and coaching the lagging reps, which is exactly the job the system should be making visible.
The thread running through all three layers is that lead management isn't a feature you switch on, it's a flow you keep honest. Capture decides what enters, scoring decides what gets attention, and status decides whether anything actually moves. The orgs that work are the ones where every lead has a clear next action and a clock on it, and the consultant's real job is designing the small constraints, the short form, the validated weights, the required date, the tiered SLA, that keep the whole thing from quietly silting up.
Implementing or optimizing Sales Cloud? Our Salesforce team runs discovery, designs the sales process, and configures Sales Cloud on production engagements. Get in touch ->
See our full platform services for the stack we cover.
