experience, not theory
Prospect research:
the limits of the platforms,
dirty data and what changes
when AI comes into play
In 2026, B2B prospect research relies on several tools that do different jobs: certified commercial databases such as Kompass, Cerved, Bureau van Dijk and Orbis for verified company data, LinkedIn Sales Navigator for precise mapping of roles, sales engagement platforms such as Apollo, Lusha and ZoomInfo for direct contacts and, in our case, AI-based sales intelligence for semantic profiling and context analysis. Each platform has its own strengths and limits, and what makes the difference is how the sources are orchestrated and varied by geography and sector, not the choice of a single platform. AI speeds up profiling and analysis, but it stops short of contact: the relationship stage remains human-led.
Prospect research today relies on four types of tool that do different jobs and that together, when used well, build something none of them can deliver alone. Certified commercial databases provide verified company data. LinkedIn Sales Navigator provides granular detail on company roles. Sales engagement platforms such as Apollo, Lusha and ZoomInfo provide access to direct contacts at scale. And AI platforms let you analyse and add context at a speed that manual work cannot match.
The problem is that each of these tools has its own distortions which, if ignored, produce useless prospect lists. And the most common mistake, one we see made every day, is to treat them as interchangeable sources: pick one and build the whole strategy on it.
Prospect research platforms all have one thing in common: they look wonderful when you buy them and turn into an obstacle course when you use them. It is worth understanding why before investing time and budget in tools that promise perfect lists and deliver noise.
Certified commercial databases: verified data as the starting point.
Certified commercial databases such as Kompass, Cerved, Bureau van Dijk and Orbis are the most reliable source of basic company data. They differ in geographical coverage and depth of data, and none of them is enough on its own.
They aggregate verified data: turnover, headcount, sector, corporate structure, organisation charts, shareholdings. This is declared and checked information, with a methodology behind it. When you build a list starting from these platforms, the filters you apply correspond to something real.
Each one has a different specialism. Cerved is particularly strong on the Italian economy, with granular data on domestic SMEs. Kompass has good European coverage and a detailed sector classification. Bureau van Dijk, with Orbis, covers the international level, with depth on multinational groups and complex corporate structures.
The question is not “which is best” but “which one do we need now”. Prospect research on manufacturing SMEs in Lombardy does not call for the same tools as research on industrial groups with a presence in Asian markets. The limit of these platforms, even the best of them, is that they stop at basic company data. They tell you what a company is, not what stage it is at, not what is going on inside it, not who really makes the decisions. For that you need other sources.
Sales Navigator excels at job-title searches but suffers from four structural distortions: profile optimisation, delays in recording job changes, unverified company data and uneven geographical coverage.
Sales Navigator is probably the most widely used tool in B2B prospect research.
It is also the one most often misjudged, not because it is a poor tool, but because it is used with expectations that do not match what it actually does.
Its strength is real and specific: searching by job title. No other platform comes close when it comes to mapping company roles. If you are looking for an “Export Manager” in manufacturing companies in northern Italy with 50 to 200 employees, Sales Navigator gives you a list that no traditional database could build with the same granularity.
The trouble starts when you forget that the data is self-reported.
Every LinkedIn profile, whether a person’s or a company’s, is filled in by the user. There is no verification process and no external check. This creates four distortions that anyone doing prospect research needs to know well.
Apollo, Lusha, ZoomInfo: the promise of direct contact and its limits.
Sales engagement platforms such as Apollo, Lusha and ZoomInfo offer access to millions of direct contacts, including business email addresses, phone numbers and roles… but they raise specific questions about data quality, regulatory compliance and responsible use.
In recent years a new category of tools has changed the way many sales teams think about prospect research. Apollo, Lusha and ZoomInfo, along with Cognism, Lead411 and Seamless.ai, promise something that traditional platforms do not offer: direct contacts at scale. Business email addresses, phone numbers and roles, all in a single interface, with instant filters and the option to export lists in a few clicks.
The potential is real. For anyone doing outbound, having a direct mobile number rather than a company switchboard can make the difference between a conversation and silence. Sales engagement platforms have undoubtedly increased the speed at which a sales team can build a pipeline.
But beneath the surface there are three limits you need to know about before building a strategy on these tools.
Apollo, Lusha, ZoomInfo and the like have real value when used as one source among many, with data verified before use and outreach processes that stand out from the background noise. They become a problem when they are treated as the complete answer to the challenge of prospect research.
AI platforms for prospect research
ChatGPT, Gemini, Copilot and Claude are useful tools for speeding up the analysis and qualification of prospects, with differing capabilities in a field that is evolving rapidly.
Right now Claude handles complex prompts and long contexts more consistently than the others, but the field moves fast and the gap narrows every quarter. Here too, the point is not which tool you choose; it is understanding what you can get it to do and where it will lead you astray.
Building a prospect list: why getting the basics wrong is costly
A prospect list built by applying standard filters on a platform reflects the quality of the platform’s data, not the quality of the real target. The result is commercial noise that burns through contacts, time and reputation.
Building a prospect list from scratch looks simple. You take a platform, apply the filters (sector, size, geography) and you have your contacts. What you get reflects the quality of the platform’s data, not the quality of your real target. Companies that fit the formal parameters but not the real profile. Sectors wrongly classified. The wrong decision-makers. Sending a list built this way into outreach is just making noise. And in sales, noise has a specific cost: it burns through contacts, wastes time and damages the reputation of whoever generates it.
For years the answer was manual. One person opening one website at a time, checking, cross-referencing, deciding. Slow, expensive, not scalable and, in some markets, simply impossible. In many Asian countries company websites are inaccessible because of firewalls, public information is scarce, and traditional profiling grinds to a halt before it has even begun.
Manual work also has a limit that tends to be underestimated: it is subjective. Two people profiling the same database against the same criteria reach different results, because human judgement introduces variables that cannot be controlled. The quality of the profiling depends on experience, concentration and the time of day. That is not a system.
How our sales intelligence works: from semantics to the hook
Mallei’s sales intelligence is based on semantic search: our data analysts describe the ideal client in natural language, and the system interprets the description and returns a reasoned assessment of each candidate company. It identifies decision-makers, relevant news and contextual signals that are useful for the first contact.
What we have built starts from a simple idea: profiling should be an act of understanding, not of filtering. The difference is substantial.
A filter tells you whether a company meets certain formal parameters: sector, size, geography. Understanding tells you whether that company has the right characteristics, timing and context to be a real prospect.
Our method starts with orchestrating the sources. The platforms described so far (certified commercial databases, Sales Navigator, sales engagement tools and general-purpose AI) are not alternatives. They are complementary layers that need to be integrated and varied according to the specific context of each search. Research on agri-food SMEs in southern Italy calls for a different combination from research on industrial groups with a presence in North America or Asia. Geography and sector determine which sources are primary, which are secondary and which should be excluded because they are unreliable in that context.
The heart of the system is semantic search. Instead of applying rigid filters to predefined categories, our data analysts describe the ideal client in natural language: who they are, what they do, what stage they are at, what problems they have and what goals they are pursuing. We build a prompt that describes a concrete business: “a company that makes industrial components, has an in-house sales team and is trying to expand into foreign markets, probably with an export manager or sales manager who handles negotiations directly and looks for companies with a particular appetite for cutting-edge technology.”
Semantic search interprets this description and compares it with publicly available information about companies such as their website, press releases, news, job adverts and tone of communication. The system returns a reasoned assessment: why this company is on target, what makes it relevant and which specific signals show it, and why another is not and what does not add up.
This reasoning is valuable for two reasons. The first is operational: it lets the contact team grasp the context straight away without further research. The second is systemic: over time, by analysing the reasoning, our data analysts learn which qualification criteria really work and which generate false positives. So the system sharpens with use.
News, decision-makers, LinkedIn: the hook that changes everything
Once the target companies have been identified, our system looks for recent news that points to specific moments in the life of the company: management changes, acquisitions, new markets, significant hires. It also analyses decision-makers’ public communication on LinkedIn to understand how they think and which topics they focus on.
Profiling on basic company data is only the first layer. What really makes the difference at the contact stage is knowing what is happening in that company right now and how the person the team is about to speak to thinks.
Once the target companies have been identified, our system goes looking for recent news: new offices, acquisitions, management changes, new markets entered, significant hires, trade show appearances. These are signals that point to a specific moment in the life of the company, and moments matter enormously in sales. A company that has just changed its sales director is redefining itself. A company that is hiring export staff is building something new. A company that has just closed a funding round is accelerating. These are contexts that completely change how it makes sense to open a conversation.
Then there are the decision-makers. The system identifies the relevant people in the organisation chart – title, career history, how long they have been in the role, how they communicate publicly. And this is where LinkedIn comes in: used this way, it becomes a completely different tool from the one used to look up company data.
Combing through a decision-maker’s communication on LinkedIn means understanding how they think, what matters to them and what language resonates with them. What they post, what they comment on, what they react to, which conversations they join. A manager who posts constantly about internationalisation and new markets has a clear frame of mind. A manager who comments on content about managing the sales team has something specific on their mind. These are openings, and a contact team that knows about them before the first approach starts from a radically different position from one that turns up with a standard message.
Where AI stops (and why it is right that it stops there)
AI speeds up profiling, analysis and contextualisation. It stops before the contact stage, which remains human-led: in B2B sales, people recognise immediately when they are talking to an automated process, and that recognition closes the door.
All of this work (the semantic profiling, the news search, the analysis of decision-makers) stops before contact is made. And that is a choice, not a limitation.
People can tell when they are talking to an automated process, often in subtle ways: a sense that something is off in the tone, the timing or the relevance of the reply. In B2B sales, where trust is the precondition for any deal, that recognition closes the door before it has even opened.
AI analyses, cleans, qualifies and contextualises at a speed and on a scale that no human team could match. Building a real commercial relationship takes presence, listening, the ability to adapt in real time and to read what is left unsaid. Those skills remain human.
A contact team that arrives at the first approach with precise profiling, the company’s news, the decision-maker’s real profile and a contextual reason why the conversation makes sense is a team that respects the time of the person in front of it. And in the long run, that is what builds a company’s commercial reputation.
Frequently asked questions about B2B prospect research
There is no single best platform. Certified commercial databases such as Kompass, Cerved, Bureau van Dijk and Orbis are the most reliable source of basic company data, each with a different geographical specialism. LinkedIn Sales Navigator is irreplaceable for job-title searches and for analysing how decision-makers communicate. Sales engagement platforms such as Apollo, Lusha and ZoomInfo offer access to direct contacts at scale, with limits on data quality and regulatory compliance. AI platforms speed up analysis and semantic profiling. An effective prospect research strategy combines several sources, knowing the specific limits of each and varying the mix by geography and sector.
Only in part. The data on company roles is accurate as a category, but it suffers from four structural distortions: profile optimisation (inflated or generic titles), delays in recording job changes (weeks or months), unverified company data (self-reported size and sector) and uneven geographical coverage (low penetration in some international markets). It should always be cross-checked against other sources.
They have real value, but they need to be used with care. Contact quality varies because the data comes from mixed sources (scraping, user contributions, algorithmic inference), and real bounce rates are often higher than those quoted. In Italy there is also the constraint of the Registro Pubblico delle Opposizioni (RPO): anyone registered with it has exercised their right not to receive unsolicited marketing communications, and these international platforms have no built-in mechanism for filtering out those contacts. They should be treated as one source among many, not as the complete answer.
AI is effective at the profiling and analysis stage: semantic search to identify target companies, gathering relevant news and analysing decision-makers’ public communication. It stops before the contact stage, which remains human-led. The value of AI in sales lies in speeding up preparation, not in replacing the relationship.
A qualified prospect list combines three layers: a match with the formal parameters (sector, size, geography), alignment with the real profile of the ideal client (business model, stage the company is at, organisational complexity) and the presence of contextual signals (recent news, intent signals, decision-makers’ public communication). A list that meets only the first layer produces noise. Sometimes the profiling can only be completed through the first direct contact with the prospect, because some information simply cannot be filtered from a desk.
Manual profiling is slow, expensive, not scalable and subjective. Two people profiling the same database against the same criteria reach different results. In markets with information barriers, such as many Asian countries (though not only there), traditional profiling is effectively impossible. AI-based sales intelligence makes it possible to process thousands of companies in parallel while keeping the assessment criteria consistent.