Every client has a different audience.
Something we’ve noticed over the last few years is that many AI-powered prospecting tools start from the same assumption. Businesses operating in the same sector must all be targeting broadly the same audience. On paper that sounds perfectly reasonable. In reality, it’s rarely true.
Take the construction sector as an example. You’ve got principal contractors, construction consultants, specialist trades, subcontractors, facilities management companies, property maintenance companies and plenty more specialist areas. Although they all operate within construction, they don’t all sell the same thing or to the same types of businesses and organisations. Some focus on commercial clients, others target councils, housing associations or the wider public sector, whilst many sell into each other. Even two businesses offering almost identical services can require completely different prospect lists depending on what they actually offer, the size of projects they take on, where they work geographically, the time of year and how their business is structured. Then there’s the question of who you should actually contact within each type of business or organisation.
In a smaller business, speaking directly with the owner, CEO, or Managing Director often makes perfect sense. In a larger organisation, you’ll usually get much further speaking to somebody responsible for a specific function. Even then, it’s not always as straightforward as searching for a particular job title.
Take a Project Manager, for example. A Project Manager working for a principal contractor could be exactly the right person to approach. But, a Project Manager working for a local authority could be responsible for anything from waste and recycling collections to traffic management, with little or no involvement in property regeneration projects. Same job title. Completely different suitability.
The above is just an example, and yes, we know you can train an agent, but so far to get to this level of nuance we’ve identified two caveats with that approach:
- Things start off well, then the agent starts going off the rails, not doing what you’ve asked and categorising unsuitable prospects as being suitable.
- It’s actually slower at making a decision at this point than a human.
This is where prospect research starts becoming about much more than databases and AI prompts. It becomes about understanding businesses, understanding buying journeys and applying commercial judgement. AI can identify companies, people and job titles incredibly quickly. Deciding whether they’re actually a suitable prospect for a specific business still requires human sales expertise.
Finding suitable prospects… and finding all of them.
Once you’ve established exactly who your audience is, the next challenge is finding them. More importantly, finding all of them.
It’s surprisingly easy to build what looks like a healthy prospect list and assume you’ve captured your market. Select a handful of filters, export a few thousand companies and away you go. The problem is, businesses don’t always sit where you’d expect them to.
Take a retail fit-out company. Most people would naturally look for them under one of many construction industry filters. In reality, there’s every chance they’ll be categorised under retail because that’s the sector they serve rather than the work they actually carry out.
The same applies to people. Someone may still be associated with an industry they worked in several years ago. Someone else may never have updated their LinkedIn profile after changing jobs. Another person may simply describe their responsibilities differently to everyone else doing exactly the same role.
The point is, if your prospect research only follows the obvious route, you’ll almost certainly miss a sizable part of your audience.
Good prospect research isn’t about finding enough prospects. It’s about designing searches that not only identify the obvious, but also uncover the businesses, organisations and people sitting outside the anticipated search criteria. The companies that don’t quite fit neatly into predefined industries. The people whose profiles don’t follow the norm. The prospects your competitors never find because they’re relying on the same filters as everyone else, because a lot of the time – that’s where the best opportunities are hiding.
Finding prospects is one thing. Accessing them is another.
Even if you’ve identified exactly who you want to target, there’s another hurdle that often gets overlooked. Can you actually access them?
Take LinkedIn Sales Navigator as an example. You might build what appears to be the perfect search and discover there are well over 100,000 matching prospects. Brilliant. Except you can’t actually view all 100,000 of them. Instead, you’re limited to the first portion of the results, the first 2,500. So what do you do?
You could start removing filters and work through the list manually. You could split your searches by county and hope the numbers become manageable. But in our experience, many counties still return more results than you can realistically review, meaning you’ll still miss large parts of your audience, or you’re forever tweaking filters trying to squeeze them out gradually – it’s messy, time-consuming, and easy to lose track of where you’re up to.
At Neptik, this is actually something we spent a lot of time solving. Rather than searching county by county, we developed our own software to combine searches with criteria individualised for each client with individual postcode searches. The software incorporates every postcode search (which contains every city, town, village and hamlet for each individual postcode). Our software then merges an individual prospect criteria search with an individual postcode search. When we create the individual criteria searches for our clients, we design them so whatever the postcode, the volume of prospects generated sits just below LinkedIn’s Sales Navigator viewing limits, so we can view the full audience being generated.
Did AI solve that problem? No. It required us to understand the limitation first and then build the right software to overcome it.
So yes, AI is incredibly powerful and will continue to get even more powerful. But not every problem is an AI problem. Sometimes the best solution is simply understanding the challenge well enough to build something better.
Prospect research isn’t something you do once
Even if you’ve identified the right audience, found every potential prospect and managed to access them all, there’s another challenge that often gets overlooked. Keeping that information up to date and valid.
Businesses start trading, go bust, merge, get acquired, rebrand and change direction all the time. People move jobs, get promoted, move departments, retire or simply leave and aren’t working. The reality is, B2B data loses validity surprisingly quickly. Industry estimates suggest around 20% every quarter, which means a prospect list that’s only a few months old could already be missing opportunities whilst including people you no longer want to contact.
That creates a number of problems. You could be sending exactly the right message to somebody who left the business six months ago. You could be contacting somebody who’s changed roles internally and is no longer responsible for buying what you sell. Or, you could be emailing an address that no longer exists.
Aside from wasting time, there’s another consequence that’s become increasingly important over the last couple of years – deliverability.
Email providers have become far less tolerant of poor-quality data. Microsoft’s latest guidance, for example, has reduced recommended bounce thresholds significantly. That means outdated prospect data doesn’t just affect the people you’re trying to reach, it can literally destroy an email outreach campaign. Even a smidgen of invalid email addresses can damage your sender reputation, reducing inbox placement for perfectly good prospects that should have received your message.
This is one of the reasons we research prospects every single day for every client. We don’t build a database once, dust it off every few months and hope it’s still accurate. Businesses evolve too quickly for that. Keeping prospect lists current is every bit as important as finding the prospects in the first place.
The smallest details are often the ones AI misses
Another thing we’ve learnt is that prospect research isn’t just about finding the right companies, organisations and the right people. It’s also about presenting that information naturally in the content you send.
Take company names as an example. Imagine LinkedIn lists a business as “Industrial Engineering Distributors Limited”. The obvious thing to do would be to strip off the “Limited” and reference them as “Industrial Engineering Distributors”. Job done, right? Except… that’s not necessarily what the business calls itself.
Their website may refer to them as IED. Their customers may know them as IED. They may answer the phone as IED. In reality, hardly anybody uses the full company name.
It’s a small detail, but these are exactly the sorts of things people notice. If your email refers to a company in a way nobody inside that business actually does, it immediately feels less personal and like they’re part of a mass outreach campaign.
It’s one of the reasons our researchers manually cross-reference company names rather than relying entirely on automation. Sometimes it’s removing a legal suffix. Sometimes it’s shortening a name. Sometimes it’s replacing the full legal name with the one everyone actually uses. Individually they’re tiny edits. Collectively they make campaigns feel considerably more natural.
Personalisation isn’t the same as relevance
This is probably where we see AI being used most lazily.
We’ve all received those emails.
- “I loved the work you did with…”
- “I just read your case study about…”
- “I was really impressed with…”
Nine times out of ten it’s obvious what’s happened. AI has pulled something it found from a website or LinkedIn profile and dropped it into the opening sentence.
Technically, that’s personalised. Psychologically, it’s often the opposite. Think about it another way. If you went on a first date, would you open the conversation with a list of facts you’d found about them? Probably not. At best it feels forced. At worst it feels a little creepy.
Sales conversations aren’t really any different. People don’t respond because you’ve proved you looked at their website. They respond because you’ve given them a reason to care about what you’re saying.
We’ve always found it’s far more effective to establish the connection first. Let people know you understand what they do, explain what you do and introduce something that might genuinely be relevant and helpful to them. Once you’ve got their attention, you can gradually reinforce that relevance in a way that feels natural rather than manufactured.
There’s a big difference between research and personalisation. Research helps you understand who you should prospect. Personalisation is simply one way of opening the conversation. But, too many businesses have started treating them as the same thing.
Research still needs to be… research
Perhaps the biggest misconception of all is that AI-generated prospecting somehow equals researched prospecting. It doesn’t. Research still needs to be research.
Every day I receive emails telling me they can generate another forty leads a month for Neptik. In the very same email they’ll tell me they’re impressed with how Neptik helps B2B companies generate leads.
Think about that for a second. They’re pitching lead generation… to a lead generation agency. The irony is almost impressive, and people using these kinds of agencies are literally being robbed by cowboy freelancers and agencies that don’t truly care about the clients they have.
But it also demonstrates exactly what’s happening across the industry. Businesses claim they’re running highly personalised campaigns when, in reality, they’ve simply asked AI to scrape a website, pull out a couple of talking points and stitch them together into an email. There’s personalisation delivered by AI. But very little, if any, actual research has taken place.
Not only does that mean they’re approaching the wrong prospects, it also means the messaging never really lands with the right ones either. The conversation starts from the wrong place, feels generic despite trying not to be and, if enough of those emails bounce or get ignored, the performance of the entire campaign suffers.
So where does Neptik actually use AI?
Reading this far, you could be forgiven for thinking we’re anti-AI. Far from it. We’re not and actually use AI every single day. The difference is how we use it.
Every client’s prospect search is designed specifically for them, ensuring we’re capturing the widest possible audience before we even think about outreach. From there, AI helps us make client-specific exclusions, identify companies operating within particular sectors, create merge fields such as our {{company types}} to make messaging more relevant, compare campaign performances against previous results and suggest potential audience refinements.
We also use AI where it genuinely saves researchers time. For example, if a client only wants to target surveyors or cost consultants involved in historic buildings, AI can search beyond the company website to identify projects mentioned elsewhere online that might not feature on their website.
Alongside that, we use our own software and scripts to organise data, validate email addresses, split campaigns intelligently and help prevent businesses being contacted too frequently.
What we don’t do is ask AI to make the final decision on whether a prospect is suitable. Our researchers still review every company, every website, every job title and every prospect we contact. They still apply the commercial judgement that AI can’t consistently. They still make the small edits that make messaging feel natural rather than automated. Because that’s where the value really is.
AI is an incredibly powerful tool, and we’ll continue investing in it, testing it and incorporating it into our processes wherever it genuinely improves what we do. But we don’t believe the future of B2B prospect research is AI replacing researchers.
We believe it’s AI supporting experienced researchers, removing repetitive tasks, speeding up analysis and allowing them to spend more time applying the judgement that ultimately determines whether a campaign succeeds or fails.
That’s where we think AI delivers its greatest value. Not replacing people, but helping them apply their expertise to greater effect, so we can continue to improve the results we deliver to our customers.
Where AI Helps and Where Human Expertise Matters.
If you’ve spent any time on social media recently, you’d be forgiven for thinking AI has already revolutionised how effective B2B prospect research is done. Lots of posts seem to be about replacing traditional research methods, building prospect lists in seconds or generating perfectly personalised outreach with a few prompts or an agent that can do it all for you far better and quicker than a human. The impression being given is that prospect research has become little more than finding companies, identifying decision makers and letting AI do the rest.
As metioned earlier, we’re all for AI at Neptik. We’ve invested heavily in it ourselves, continue testing new ways to use it and have developed our own software that supports different stages of our prospect research process. But somewhere along the way, the conversation seems to have shifted from AI being a really useful tool to AI being a replacement for sales judgement. From our experience, that’s where things start to unravel in a way that’s contrary to all the hype.
The reality is, prospect research isn’t simply about finding companies or job titles. Good prospect research is about understanding businesses (yours and your target audiences’), understanding buyers’ needs, challenges, buying journeys and, more importantly, understanding who is actually worth approaching. Those aren’t always the same thing.
AI is brilliant at processing information. What it isn’t brilliant at (at least not yet) is understanding the commercial nuances that experienced sales researchers often spot almost instinctively. So, rather than talking about AI versus humans, we thought we’d explain where AI genuinely helps, where it still falls short, and why we believe the best results come from combining the two.
