Why Most Lead Generation Problems Are Systems Problems
Most businesses with a lead generation problem do not have a sourcing problem. They have a systems problem. They are generating some leads — through referrals, inbound content, outbound campaigns, events — but inconsistently. Some leads convert well, others do not, and the sales team cannot reliably predict which will be which before investing time in them. The pipeline fills and empties in cycles that nobody fully controls.
The root cause is almost always the same: there is no defined system for who is a good lead, where to find them, how to qualify them when they arrive, and how to prioritise them against each other. Without that system, lead generation is an activity rather than a process. It produces effort and occasional results, but not a scalable, predictable pipeline.
An AI lead generation agent builds the system, not just the list. Harriet — the KissMySkills lead generation agent — covers every component of a working B2B lead generation system in one session: ICP definition, ranked outbound sources, inbound strategy with lead magnet recommendations, lead capture mechanics, qualification criteria, and a scoring model with defined MQL and SQL thresholds. The output is not a spreadsheet of prospects. It is the operating model for a pipeline.
ICP Definition: The Foundation Everything Else Depends On
Lead generation without a precise ICP definition produces volume without quality — and volume without quality is expensive. Every hour an SDR spends prospecting a lead that was never going to convert is an hour not spent on a lead that would have. Every meeting booked with an unqualified prospect is a meeting that displaced a qualified one.
Harriet asks about the best leads the business has ever closed — their industry, size, role, the situation that made them ready to buy, and what made them easy to work with once they became customers — and builds a lead profile from that pattern. The profile captures both firmographic signals (company size, industry, geography, revenue range, headcount) and the behavioural and situational signals that indicate a buyer is ready to engage now: trigger events like funding rounds, leadership changes, product launches, or compliance deadlines; technology stack indicators; growth stage signals.
Equally important: the disqualifiers. Knowing who not to pursue saves as much time as knowing who to target. The ICP profile includes explicit disqualification criteria — the signals that indicate a lead is unlikely to convert regardless of how well the outreach is executed. A company in the right industry at the wrong size, or the right size at the wrong growth stage, is not a good lead. Defining those boundaries before sourcing begins prevents wasted effort downstream.
Outbound Lead Sources Matched to the ICP
The right lead sources depend entirely on who you are targeting — and generic advice about lead generation tools ignores this. LinkedIn Sales Navigator is the right tool for finding senior decision-makers at mid-market technology companies. Industry associations and trade directories are the right source for decision-makers in manufacturing, professional services, or regulated industries where LinkedIn penetration is lower. Apollo covers broad B2B coverage across most segments. GitHub and Stack Overflow are the right sources for developer ICPs. Event attendee lists are underutilised for ICPs concentrated in specific verticals.
Harriet recommends lead sources ranked by expected volume, lead quality, and cost — with specific search parameters and filters for using each source effectively for the stated ICP. Not a generic list of tools with their logos, but a prioritised sourcing strategy with the specific approach for each source given the buyer profile. The difference between knowing Apollo exists and knowing which Apollo filters to apply for this ICP is the difference between a tool and a system.
Inbound Lead Generation and Lead Magnet Design
The most cost-effective leads are inbound — prospects who come to you having already demonstrated interest. An outbound-only lead generation strategy caps its efficiency at the cost per contact of cold outreach. An inbound component compounds: content and lead magnets created once continue attracting leads indefinitely.
Harriet designs an inbound strategy alongside the outbound approach: content topics that attract the ICP based on their actual search behaviour, SEO opportunities worth prioritising given the domain's current authority, and three specific lead magnet concepts designed around the stated buyer's operational problems.
Lead magnets that attract the right buyer are specific to their pain. A "free guide to B2B marketing" attracts everyone and converts few. A "pipeline coverage calculator for SaaS sales teams" attracts exactly the ICP who has a pipeline coverage problem and is actively looking for ways to address it — and that ICP is already pre-qualified before they fill in the form. Harriet recommends lead magnet formats and specific titles based on what the ICP is most likely to exchange their contact details to access.
Lead Scoring: Prioritising the Pipeline That Already Exists
Most businesses that lack a lead scoring model treat all leads equally — which means treating a prospect who has visited the pricing page three times the same as one who downloaded a top-of-funnel blog post six months ago. The result is a sales team that follows up on every lead with the same urgency, burns time on cold leads, and occasionally misses a hot one that went to a competitor while the SDR was working the wrong contact.
Harriet builds a scoring model that assigns point values to the signals that correlate with conversion for this specific business — firmographic fit against the ICP profile, behavioural signals like page visits and content downloads, engagement signals, and intent data indicators. The model includes defined MQL and SQL thresholds, routing logic for who receives which leads at what score, and early warning signals that indicate a lead is cooling before it goes completely cold.
The output maps to the fields available in standard CRM tools — Salesforce, HubSpot, Pipedrive — so the scoring model is implementable immediately rather than theoretical.
Who Uses an AI Lead Generation Agent
Sales and marketing teams at companies of 10–200 people where the pipeline is inconsistent and the cause is not fully understood. Founders who are responsible for their own pipeline and need to build a scalable lead generation system before hiring their first SDR. Revenue operations professionals who need to formalise the lead qualification and scoring process that is currently happening informally in people's heads. New sales hires who need to understand who to target and where to find them before they can start prospecting effectively.
How to Start a Lead Generation Session with Harriet
Load the Harriet skill file into Claude Projects. Paste the activation prompt. Harriet asks intake questions one at a time — the best customers you have closed, the customers you wish you had not, the channels currently generating leads, and what is known about why some leads convert and others do not. Answer specifically: the more precise the input, the more targeted the system. Receive the complete lead generation system. The full session takes 20–30 minutes for most businesses. Harriet works with Claude, ChatGPT, or any AI chat that accepts system prompts.
The agent behind this guide. Harriet builds a full B2B lead gen system — ICP profile with disqualifiers, ranked sources, lead magnets, qualification criteria, and a CRM-ready scoring model.