Lead quality for B2B SaaS

Sales says the leads are bad. Marketing says sales is not working them. Both sides are arguing about a word neither has defined, which is why that meeting repeats every quarter and ends the same way.

Lead quality has numbers attached to it. Once you attach them, the argument becomes a spreadsheet problem, and spreadsheet problems get settled. The uncomfortable part comes second: every change that raises quality costs volume somewhere. Your job is not to avoid the trade. It is to pick which volume you are willing to lose, and to prove the swap paid.

What lead quality actually means when you put a number on it

Lead quality is the rate at which leads from a source become qualified opportunities, multiplied by what those opportunities are worth and how long they survive. Title seniority, headcount and budget confirmation are proxies for that, and proxies drift as your market shifts.

One metric alone gets gamed. Lead-to-opportunity rate on its own rewards sources that produce eager small accounts. Average contract value on its own rewards sources that produce two enterprise deals a quarter and nothing else. You need five numbers per source, reported together:

• Lead to opportunity rate, measured on a cohort basis
• Win rate from opportunity to closed won
• Average contract value of the deals that close
• Sales cycle length from lead creation to signature
• Retention at day 90, measured on customers from that source

The fifth is the one nobody tracks and the one that changes decisions. A source can convert well, win well, and then churn in seven weeks because the offer attracted people solving a temporary problem. That source is not producing quality, it is producing a refund queue with extra steps. If you want to turn those five numbers into a single comparable figure per source, the lead value calculator does the arithmetic, but the discipline of looking at all five separately is worth keeping for at least two quarters before you collapse them.

Cohort timing is where most quality reporting falls apart. If your sales cycle runs 70 days, leads created in the last month cannot be judged, and judging them anyway makes fast-closing small deals look like your best source. Report each month’s lead cohort at day 30, day 90 and day 180, and only make budget decisions on the day-90 column.

Build the quality scorecard before you change anything

Put every lead source in one table with the same five columns, on the same cohort window, before you touch targeting. Most teams discover their quality problem lives in two sources that account for 8 percent of spend, and the sweeping change they were about to make would have damaged the other six.

Here is what that table looks like with realistic mid-market numbers. Use your own, but keep the shape.

Read it column by column, not row by row. The gated report produces 420 leads and 8 opportunities. The organic demo request produces 60 leads and 17. Sales feels the first number and reports on the second, which is exactly why the pipeline meeting goes badly.

The purchased list row is the one to kill immediately, and not because of the 0.4 percent. It is the day-90 retention at 79 percent. Those customers cost more to serve, generate support tickets that pull engineering attention, and produce the review scores that damage your organic demo requests eighteen months later.

If you want a structured pass over your own data rather than building the table from scratch, the lead quality audit checklist walks through the fields to pull, the cohort windows to use, and the three joins that usually break in Salesforce.

The four levers that move quality, and what each one costs in volume

Four things change who fills in your forms: the offer, the audience you put it in front of, the form and qualification rules, and how the lead is routed once it arrives. That is the complete list. Everything else is a variation on one of them.

Each lever has a different speed, a different volume cost, and a different failure mode. Ranked by how much quality improvement you get per unit of volume lost:

Lever	Volume cost	Time to effect	Quality lift	Best for
Offer change	High at first, recovers	4 to 8 weeks	Large	Teams whose top asset is a generic ebook
Audience targeting	Direct and proportional	1 to 2 weeks	Large in paid, small in organic	Paid-heavy programmes with loose match types
Routing and speed	None	Days	Moderate to large	Anyone with more than a 15 minute response time
Form and qualification	High across all segments	Immediate	Small	Teams with a genuine spam or bot problem

Notice that routing costs no volume at all. It is the only lever on the list that is free, and it is the last one most teams examine. Notice too that the form, the lever teams reach for first, sits at the bottom.

Offer

Changing what you ask people to raise their hand for changes who raises it. A report called “The State of Workflow Automation” attracts analysts, students, competitors and people building a slide for their boss. An interactive assessment that scores your current setup and emails you a remediation plan attracts people with a current setup and a reason to fix it.

Expect the download count to fall by half and the opportunity count to hold or rise. Give it two months, because the new asset needs to accumulate rankings and internal links before its volume recovers.

Targeting

In paid, this is the fastest lever and the most linear. Tighten match types, exclude job titles, add firmographic layers, drop the placements that produce clicks from countries you do not sell into. The volume drop arrives within days and matches the exclusion almost exactly. The PPC lead quality guide covers the audience mechanics in detail, including the negative keyword patterns that remove student and jobs traffic without touching buyer terms.

In organic, targeting is a content decision measured in quarters. You change what you publish, not what you exclude.

Form and qualification

Adding fields cuts submissions across every segment, including the CFO you wanted. And the low-fit visitor who wants your template will type anything into a field to get it, so you gain fake data rather than filtered leads. Use enrichment on the email domain instead. Clearbit, now part of HubSpot, and similar providers return company size, industry and tech stack from a work email, which covers most of what the extra fields were asking for.

Keep form fields for things nobody can infer: project timing, current tool, who else is involved in the decision. Three fields plus enrichment beats eight fields every time.

Routing

Speed and accuracy of routing do not reduce volume by a single lead. They change conversion outright. More on this below, because it is the most underrated of the four.

A worked example: 40 percent fewer leads, 15 percent more opportunities

Here is the swap that makes the case to a sceptical CRO. The company below cuts monthly lead volume from 1,000 to 600 by retiring two sources and tightening a third, and ends up with more opportunities than it started with.

The arithmetic is simple once the sources are separated. The 400 leads removed were converting at roughly 1.1 percent, producing about 4 opportunities between them. The 600 that remain get the sales capacity those 400 were absorbing, and their conversion rate rises because reps reach them faster and with more context. That second effect is the part teams forget to model, and it is usually worth two to four points of conversion on its own.

Cost per lead went up 44 percent, which is exactly why cost per lead is a diagnostic and not a goal. Cost per opportunity fell 23 percent and new customers rose 26 percent. If your board reports CPL as a headline metric, change the headline metric before you run this play, or you will win the operation and lose the meeting.

Most quality complaints are offer problems wearing a disguise

When a rep says the leads are bad, ask what the lead downloaded. Nine times in ten it is a top-of-funnel asset with no relationship to the product, promoted to an audience defined by job title rather than by problem.

A person who downloads “12 Slack Templates for Remote Teams” has demonstrated interest in Slack templates. They have demonstrated nothing about buying your workforce analytics platform. Passing that person to an account executive and then calling them a bad lead is a category error committed upstream, by whoever decided the asset counted as a lead.

The fix is not better scoring on the same asset. It is an offer that only makes sense to someone with the problem you solve. Three patterns that work in SaaS:

• A tool that requires the reader’s real data to produce output, like a pricing migration estimator or a spend audit
• A teardown of the reader’s own setup, delivered as a recorded walkthrough rather than a PDF
• A benchmark report that requires submitting your own numbers to see where you sit

Each one filters by willingness to expose a real situation, which is a far better predictor than title. And each produces a follow-up conversation with something to talk about, which is why the same lead converts better even before the rep does anything differently.

Once you have the offers right, scoring stops being a filter and starts being a prioritiser. That is the correct order, and it is why a lead scoring model built on top of weak offers just ranks bad leads more precisely.

When “bad leads” is really a routing and speed problem

Before you change a single targeting parameter, measure two things: median minutes from form submission to first human contact, and the percentage of leads routed to the correct owner on the first attempt. If the first number is above fifteen minutes or the second is below 90 percent, you have a routing problem and no amount of qualification tightening will fix it.

The Lead Response Management Study found a roughly 21-fold difference in the odds of qualifying a lead contacted within five minutes versus thirty. That research is old, the effect has been reproduced many times since, and the direction has never reversed. Nothing in a scoring model comes close to that magnitude.

Routing failures that produce the “bad leads” complaint:

• Demo requests landing in a shared inbox checked twice a day
• Round-robin assignment that ignores territory, so the rep has no context and no commission incentive
• Enterprise-sized accounts dropped into a self-serve nurture because the form had no company size field
• Leads that arrive at 4:50pm Friday and get touched Monday at 11am
• Existing customers submitting a demo request for a second product and getting treated as net new

The organisational version of this fix is a written agreement about who touches what, how fast, and what happens when the standard slips. A sales and marketing SLA that names response windows per lead tier, with a monthly review of compliance, removes most of the argument permanently because both sides can see who missed.

Gating harder is the weakest lever, and teams pull it first

Putting a form in front of more content is the intervention that feels like action and produces the least. It cuts volume across every segment uniformly, so your high-fit traffic is filtered out at the same rate as your low-fit traffic, and your ratio barely moves.

Worse, it hides the signal. An ungated comparison page tells you which competitors your visitors are evaluating through page views and scroll depth, feeding your targeting. Gate it and you get 40 form fills a month and lose the behavioural data from the 4,000 people who would have read it.

The gating decision that does work is asymmetric. Ungate everything that builds the case for the category and teaches the reader something. Gate only what requires your effort to produce for them personally: a custom benchmark, an audit, a migration plan. The first group builds the audience. The second group filters it by intent, and the filter is genuine because the ask is real.

Where gating is defensible: you have a compliance or data-residency obligation, you are running a paid programme where the asset is the entire offer, or the asset genuinely costs you money per delivery. Outside those three, ungate and measure what happens to demo requests over the following quarter. At most SaaS companies they go up.

How to run the change so sales and the board see it coming

The technical work is the easy part. The failure mode is political: volume drops in week two, someone escalates in week three, and the change gets reversed in week four, six weeks before the opportunity data would have vindicated it.

What this costs and the three ways it goes wrong

Raising lead quality is mostly analyst time and political capital rather than budget. Expect two to four weeks of one person’s time to build the scorecard properly, because the CRM joins are always messier than they look. The offer work is real production cost, somewhere between $3,000 and $15,000 for an interactive tool or assessment depending on whether you build it in-house.

The failure modes, in order of how often they bite:

You cut a source that was feeding another source. Webinar registrations often convert badly on their own and convert the demo requests that follow them. Before retiring a source, check whether contacts from it appear as an earlier touch on deals attributed elsewhere. Kill it and watch your “good” source quietly decline the following quarter.

Higher quality leaves sales capacity idle. If you cut 400 leads a month and your reps were at 60 percent capacity, you have not freed anything, you have just reduced activity. The swap only pays when reps were saturated. Check rep utilisation before you cut, and if they have spare capacity, fix the offer instead of reducing volume.

You change the definition and quietly move the goalposts. If the new MQL definition is stricter, your historical comparison is meaningless and everyone knows it. Keep reporting the old definition in parallel for two quarters so the trend line stays honest. This is tedious and it is the difference between a credible programme and one nobody trusts.

We spent a year tuning the scoring model. The thing that actually moved opportunity rate was deleting one ebook and answering the phone faster.

What to do in the next two weeks

Start with measurement, not with changes. You cannot defend a cut you cannot quantify, and the scorecard usually reveals that the problem is narrower than the complaint.

Work the list in order. If the routing numbers come back bad, stop there and fix them first, because it is the only lever that improves quality without costing you a single lead. If routing is clean and the scorecard shows two weak sources, retire them and defend the volume drop with the forecast you published in advance.

The broader programme context, including how these sources get built in the first place, sits in the SaaS lead generation hub. Come back to the scorecard every quarter. Quality is not a project that finishes, it is a number you keep on the wall.