How LeadAgent works
A plain-language explanation of what the system builds, what each score means, and what is automated versus left to you.
Business profile
You describe what your company sells, who it serves, where it operates, its value proposition and obvious exclusions. This is the factual base layer.
Sales / Outreach profile
You define the specific offer, buyer outcome, problems solved, seller strengths, differentiators, proof points and preferred first step. These fields affect both discovery and message generation.
Customer discovery model
LeadAgent turns the business and sales profiles into an editable model: target customer profiles, priority segments, search queries, positive and negative fit signals, hard qualification rules and seller-fit logic.
Discovery and prechecks
Search queries find candidate company pages. Before paid analysis, deterministic checks remove obvious duplicates, excluded categories, unusable domains and candidates that clearly fail cheap prechecks.
Company research and qualification
For a candidate that survives prechecks, LeadAgent reads a limited set of public pages, extracts relevant evidence, checks the company against your current strategy and produces a reasoned fit assessment.
Personalized sales draft
The draft is based on your seller positioning plus evidence about that specific prospect. LeadAgent does not send it. You can edit, save and copy it into your own outreach channel.
Why a lead gets its score and status
The score is a relevance estimate for this specific Business and this specific strategy version. It is not a universal quality score for the company.
The available public evidence strongly matches the strategy and no hard rejection rule is triggered.
There is some plausible fit, but evidence is incomplete, mixed or not strong enough for automatic qualification.
The fit is weak based on the strategy, or the evidence points away from the target profile.
What evidence means
Evidence is the public information stored alongside the lead to support the qualification decision. It may include company website pages, text snippets and why each item matters. The evidence lets you audit the result instead of accepting a score blindly.
If evidence is weak or missing, treat the result as lower confidence and review the company yourself.
Unknown is not the same as failed
LeadAgent should not reject a company merely because its website does not prove a desirable trait such as team size. A must-have fails only when there is reasonable evidence that the condition is not met. Nice-to-have signals improve fit but are not mandatory.
Why strategy versions matter
Every saved strategy is kept as a read-only historical snapshot, and every research run records the exact strategy version it used. You can open Strategy history later to see what Search Queries, qualification rules, exclusions and seller-fit logic were active at that time. If you change the strategy, older leads are marked as needing re-evaluation instead of silently pretending their old score is current.
Why seller strengths affect search
Your strengths can change which prospects are valuable. For example, if your service is strongest where a small B2B team performs repeated manual prospect research, the strategy can prioritize companies showing signs of that sales motion rather than merely targeting an industry.
What the strategy fields actually do
The initial strategy is built from the Business profile and Sales / Outreach profile, but it remains editable. These fields are instructions to the discovery and qualification pipeline, not decorative notes.
Search queries
Literal web-search queries used to discover candidate company sites. They should describe the kinds of businesses and commercial signals worth finding, not broad informational topics.
Priority segments
Business categories worth searching first. A segment is a discovery priority, not an automatic guarantee that every company in it is a good lead.
Positive signals
Public evidence that increases fit: for example a matching customer type, relevant service model, commercial CTA, expansion signal or another seller-specific condition.
Negative signals
Evidence that lowers fit without necessarily rejecting the company. Mixed positive and negative evidence can place a lead in Review.
Must-have rules
Hard requirements. A company is rejected only when public evidence reasonably shows that a must-have is not met. Missing information alone should stay unknown.
Nice-to-have rules
Helpful traits that can strengthen the score. They are intentionally optional and should never become hidden hard gates.
Disqualifiers
Conditions that should reject a prospect regardless of score, such as a direct competitor or a business type the seller explicitly never wants.
Strategy exclusions vs discovery exclusion terms
Strategy exclusions describe unwanted company categories. Discovery exclusion terms are short conservative phrases used earlier to discard obvious search-result noise before paid analysis.
Seller-fit fields
Seller strengths, problems solved, best-fit conditions, high-value use cases and strength-derived signals connect what the seller is good at with observable conditions at a prospect.
Outreach angles
Supported ways to frame value later in a draft. They guide messaging but do not permit the writer to invent a prospect problem or unsupported seller result.
How LeadAgent avoids repeating the same search results
Discovery URLs come from live web-search results, not from the analysis model's memory. For each Business, LeadAgent keeps source-level history. If the exact source URL has already appeared in an earlier run, it is marked as previously seen and skipped as a new discovery source. Previously analyzed or intentionally rejected company domains are also skipped.
When you run the same strategy again, LeadAgent does not intentionally restart from the same first page forever. It prioritizes strategy queries that have not been used yet and continues previously used queries from later available search pages. If you create a new strategy version, search pagination starts fresh for that version, while Business-wide URL/domain deduplication still prevents known companies from being treated as new.
Each research run stores its strategy version, search queries, search pages and the actual URLs returned by search. Processes link directly to the found sources, and the full run view lets you audit every source and its outcome.
A previously failed website crawl may still be retried when the company appears through a genuinely new source, because temporary website failures should not permanently erase a valid prospect.
How contact confidence is shown
Email verified
The address has source/domain evidence strong enough for LeadAgent to mark it as verified. A verified address can still be unsuitable for sales outreach.
Outreach usable
LeadAgent separately checks whether the mailbox looks appropriate for a sales message. Operational addresses such as privacy, legal, security, press or support are blocked from automatic outreach usability even when the address itself is real.
Contact data comes from public pages and can be incomplete or stale. Users should review important contacts before outreach.
How a message is built
A good draft is not generated from the prospect alone. LeadAgent combines the seller's offer, value proposition, strengths, differentiators, proof points and CTA with the prospect's stored evidence and qualification context. The goal is to explain why the seller may be relevant to this specific company.
Use public facts about the prospect, supported seller claims, cautious pain hypotheses and a simple first-step CTA.
Invent internal processes, customer counts, revenue, ROI, tools, results or other private facts. Your “Claims the email must never invent” list adds business-specific restrictions.
How credits are charged
LeadAgent uses a credit balance for paid processing. Web-search requests and AI processing consume credits as they are successfully performed. Deterministic filtering and normal page crawling are not charged separately.
Processes and Credit history show the amount charged for completed actions, so you can see what was deducted from your balance. Duplicate active jobs are blocked to reduce accidental double spending.
There are currently no subscriptions or recurring plans. You only use your balance when you start paid processing.
What LeadAgent does not decide for you
LeadAgent helps discover, organize and prioritize public business information. Scores and generated drafts are decision support, not guarantees. You can edit strategy fields, review evidence, re-evaluate leads after strategy changes, override contact usability after manual review and edit every message before using it.