Back to Overview
Executive Growth & AI Transformation

Why more leads do not create predictable growth

More contacts in a CRM do not make revenue predictable. Customer fit, accountable handovers, real opportunities and an economic baseline matter more. How CEOs can manage quality rather than lead volume and apply AI only after verification.

Many CEOs see a frustrating pattern: marketing produces more contacts, sales activity is visibly higher, yet revenue and margin remain difficult to predict. The consequences are familiar: overloaded teams, optimistic forecasts and arguments about which channel is at fault. The underlying issue usually sits deeper. A lead is a possible signal, not evidence of demand, buying intent or economic value. Predictable growth only emerges when target accounts, handovers, qualification, sales work and feedback from won and lost business are managed as one accountable value stream. In this article, I set out the decisions leadership must make, how to establish a reliable baseline and why quality after verification matters more than sheer volume at the point of entry.

The convenient metric conceals the management task

Lead volume is attractive because it is visible early and easy to increase. A campaign can generate more form submissions, an event can bring in more contacts and a download can add more names to the CRM. For the leadership team, however, none of this answers the central question: does it reliably create profitable new business with the customers that fit the company?

When an organisation manages to lead volume, it can confuse activity with progress. The sales team then works through many contacts whose origin, need or decision-making capacity has not been adequately established. Marketing optimises for an early response. Sales prioritises on experience or time pressure. Finance sees a margin later on that is difficult to trace back to the original source. No one is acting wrongly within their own function. The system still fails to produce a reliable commercial flow.

I would therefore reverse the question. Not, “How do we generate more leads?” but, “Which verified contacts have a realistic prospect of becoming suitable customers with a sustainable contribution margin and who owns each stage?” This is not a semantic refinement. It turns a marketing metric into a leadership responsibility.

Where the value stream usually breaks

In growing companies, the causes rarely lie with one campaign or one individual. They arise at the handovers. The target customer profile remains too broad. An initial contact is passed to sales without context. A conversation is treated as an opportunity even though the problem, decision maker, timing or economic fit remains unclear. The CRM records activity but not the quality of the next decision. Reasons for lost business are fed back in categories that are too broad or not at all.

Buyers also do not behave in line with an organisation’s own structure. McKinsey surveyed nearly 4,000 B2B decision makers across 13 countries and 34 sectors. Respondents reported an average of ten channels in their buying journey and more than half wanted to move seamlessly between channels.1 This is not a universal conversion rule, nor a causal claim for every company. It is, however, a credible indication of a management responsibility: context must not be lost when the channel changes.

As a self-reported B2B sample, it does not establish an outcome for a particular company or industry. The increasing degree to which buyers inform themselves also changes the division of work. In Gartner’s survey of 646 B2B buyers, conducted in August and September 2025, 67% said they preferred a rep-free experience; 45% used AI in their most recent purchase. Gartner also stresses the importance of value clarity in the buyer’s specific context.2 This does not mean that complex selling works without people, nor does it indicate conversion in any particular funnel. On the contrary, when interested parties do speak with a company later in the process, the conversation must create orientation, clarify open risks and enable a relevant decision. A fast call back without prepared context is not a strength.

The common misjudgement is therefore to pay for more lead generation before existing enquiries are being guided through a sound process. This first increases workload and can later erode margin. Growth becomes less stable, not more predictable.

The CEO must decide the commercial definitions

Predictability does not begin with a dashboard. It begins with a small number of binding definitions. What is a target customer? What counts as a qualified contact? At what point does an opportunity exist? When is business considered won or lost? And which margin serves as the economic test?

These questions are not operational detail. They determine whether marketing, sales and finance are talking about the same reality. A definition need not be theoretically perfect. It must be usable in everyday work, documented and reviewable. The distinction between interest and verification is particularly important. A contact may fit the market segment well and still not be an opportunity. An opportunity exists only once the next step has been agreed by both sides and the key criteria around the problem, fit and decision path are sufficiently robust.

The following logic is not a universal funnel or a benchmark. It is a decision exhibit through which a leadership team can examine its own path from first contact to margin. The specific criteria should be adapted to the business model, sales cycle and risk profile.

StageWorking definitionMinimum evidencePrimary ownerReview question
Initial contactAn identifiable person or organisation responds to a relevant prompt.Source, consent where required and initial context.MarketingIs the contact real, attributable and recorded in a data-protection-compliant manner?
Verified leadThe contact broadly fits the defined target customer profile.Industry, role, company characteristic and exclusion criterion checked.Marketing with Sales OperationsWhat fit is evidenced and which assumption remains open?
Qualified conversationAn exchange about a specific commercial issue has been arranged or held.Trigger, conversation objective, contact and next step recorded in the CRM.SalesIs there a relevant problem or only general interest?
OpportunityThe prospect meets the agreed criteria for problem, fit, decision path and timing.Documented criteria, contact plan and accountability.Sales leadIs a real buying process visible or is activity being recorded as an opportunity?
Proposal and decisionA proposed solution is assessed against agreed requirements.Value assumption, scope of work, risks and decision date.Sales with functional leadIs the proposal economically sound and operationally deliverable?
Realised marginWon business is assessed for its economic quality.Revenue, direct effort, agreed delivery assumptions and variances.Finance with Sales LeadershipDoes the contribution margin match the assumption and what do we learn from it?

The table exposes a trade-off. Tighter verification may initially reduce the number of leads passed on. That is not failure, provided the conversations that follow are clearer and the organisation spends less time on contacts that are plainly unsuitable. Equally, a high threshold must not be used to exclude new market segments too early. The CEO is therefore not deciding for the highest or lowest rate, but for a definition that can learn and has understandable exceptions.

The baseline matters more than the next channel test

Before a company expands its lead sources, it needs a baseline. It will not answer everything, but it prevents arguments about symptoms. I recommend recording the data that actually exists across the six stages for a defined period. This is not a perfection project. It is an honest starting point.

The baseline includes the number and origin of initial contacts, the share of verified leads, the time between handovers, the number of qualified conversations, the opportunities that meet the organisation’s own definition, proposals, won and lost business and the realised contribution margin where it can be attributed to the business. Data gaps matter just as much. If source, decision path or reason for loss is not reliably documented, that is itself a finding, not an inconvenience to be removed from the report.

Each metric needs an owner and a fixed review. Marketing cannot be solely responsible for lead quality if sales never feeds back on the criteria. Sales cannot assess attribution alone if the contact’s origin is recorded inconsistently. Finance should not enter the process only after signature if economic quality is a management measure. In practice, a short shared commercial review works well, involving one person from marketing, sales and finance, together with a clearly named decision owner from the leadership team.

This review is not for the justification of individual functions. It tests four questions: where are we losing suitable contacts? Where are we investing time without verified progress? Where does expected economic quality diverge from the outcome? And which rule, handover or data requirement will we change before the next meeting? A review without a decision record remains reporting. A review with an owner, a date and a testable action becomes management.

AI scoring can prioritise, but it cannot assume responsibility

Once the baseline is in place, the question of AI-supported lead scoring becomes meaningful. Such models can identify patterns in available attributes and indicate which contacts should be assessed first. They do not, however, replace the definition of an opportunity or human judgement of a complex customer context.

The systematic literature review by Wu and co-authors finds no universally guaranteed improvement from a particular scoring model. It differentiates models and data sources according to their conditions of use.4 The review covers English-language publications from 2005 to 2022, selected from six databases and grey literature. Its management implication is clear: a model is only as useful as the attributes, outcomes and rules on which it is based. If historical data chiefly represents contacts that were easy to reach quickly in the past, a model can reinforce precisely those patterns. If reasons for loss are unclear, duplicates are common or sales activity is documented inconsistently, mathematical prioritisation does not automatically produce wiser economic decisions.

The technical connection also deserves attention. In Salesforce’s international survey of 4,050 sales professionals across 22 countries, conducted in August and September 2025, 51% of sales leaders using AI reported that disconnected systems were slowing their initiatives, while 74% of sales professionals named data cleansing as a focus.3 As a vendor survey based on self-reports, this is not neutral evidence of effectiveness or AI return and it cannot be generalised to a particular company. It nevertheless underlines a simple operational truth: placing another model over fragmented data does not solve the absence of a shared data foundation.

Good governance for scoring therefore consists of clear boundaries. First, the model receives a defined task, such as prioritising contacts for review, not automatically rejecting a strategically relevant contact. Second, the data sources, criteria and access rights used are documented. Third, a functionally accountable person reviews samples, exceptions and surprising results. Fourth, model performance is checked against real outcomes verified later. Fifth, it remains clear who may decide an exception and who bears the consequences of a false priority.

In my experience, trust does not emerge because a score looks precise. It emerges when the team can explain why a recommendation was made, when it will be disregarded and how feedback develops better judgement. Human decision making is not an inconvenient residual process. It is indispensable in new markets, large accounts, complex value propositions and situations that carry reputational significance.

A 30/60/90-day model for controlled clarity

The right sequence is not to digitise everything at once. First make the value stream visible, then make the rules binding, then test targeted improvements.

In the first 30 days, the common language is established. The leadership team appoints a commercial decision owner. Marketing, sales and finance define target customers, exclusion criteria, the stages in the CRM and the minimum evidence for each stage. In parallel, they create a first baseline from available data. Where data is absent, it is explicitly marked. The result is not a polished forecast model, but a shared picture of the starting point.

By day 60, the most critical handovers are simplified. This includes an unambiguous handover rule, a binding next step for qualified conversations and a small number of clear reasons for loss. A weekly review examines samples rather than totals alone. For example, the team can read selected cases marked as opportunities and assess whether they actually meet the definition. Only after this verification should it decide whether to adjust a lead source, an approach or a data field.

By day 90, the management rhythm is stabilised. The leadership team compares the baseline with new data collected under the same rules. Not every change is immediately a success or failure. The decisive questions are whether the quality of definitions improves, handovers become clearer, reasons for loss become usable and expected economic quality is more closely connected to realised outcomes. The company can then make an informed decision about which demand sources to scale, which processes to correct and whether to begin a limited AI scoring trial under governance.

This model requires discipline, not an oversized transformation. It protects against the common temptation to rebuild a CRM before knowing which decision it should help the organisation make better.

When independent clarification is useful

Some leadership teams can begin this work internally. Others find that the discussion stalls at the definitions: marketing and sales trust different data, economic quality is not visible or the causes of lost business are discussed politically rather than factually. In that situation, it would be premature to buy new technology or launch a large sales initiative.

An Executive AI Diagnostic can be the appropriate first step where clarity about the starting point, value potential, risks and the next 90 days is missing. It provides an independent management diagnostic with an Executive Summary, a baseline, an Opportunity & Risk Map and prioritised direction. In this context, the purpose is not to make AI the sales topic. It is to understand target customers, handovers, data, CRM discipline and commercial constraints well enough for the leadership team to make a well-founded next decision.

A sound outcome may also be that neither a scoring model nor a new channel is needed. That possibility belongs in a serious diagnostic. Predictable growth does not begin with more activity. It begins with the willingness to judge the quality of the commercial system after verification.

Frequently asked questions

Should we stop measuring the number of leads altogether?

No. Lead volume remains a useful input signal and can make reach or demand development visible. It must simply not stand in for quality, opportunity progress or economic outcome. Its meaning emerges only in connection with clear definitions and subsequent results.

How much data quality does a company need before it starts?

Enough to recognise the most important stages, sources and handovers honestly. Completeness is not an excuse for inaction. At the same time, missing data quality should not be concealed by elaborate modelling. The first baseline may be incomplete if its gaps remain visible and have an owner.

Who should lead the commercial review?

It should be led by a person who can make or escalate cross-functional decisions. Depending on the company, that may be the CEO, Chief Revenue Officer, sales leader or a clearly mandated executive role. Marketing, sales and finance provide the necessary perspective on process, market and economics.

When is AI scoring useful?

Only once the target customer profile, stages, outcome signals and data foundations are sufficiently clear. It should start with a limited prioritisation question, human review and a comparison with verified outcomes. A score is a prompt for attention, not a judgement on a customer’s value.

How will we recognise progress after 90 days?

Not through an isolated increase in lead volume. Progress is visible when the organisation defines cases more consistently, makes handovers traceable, records usable reasons for loss and feeds the economic quality of won business back into management more effectively. Only then does it become clear which demand truly fits the business model.

Related Insights

Strategic Advisory

From insight to a decision you can stand behind.

We translate AI potential into clear architecture, measurable priorities and an executable roadmap for your organization.