AI in Sales: Your New, Expensive Way to Alienate Customers

KI im Vertrieb · 5. März 2026 · Anthony Filipiak

AI in sales promises miracles but often delivers only digital junk. Learn why 90% of industrial companies fail and how to do it right.

Last week, an email landed in my inbox. Sender: A young salesperson from a SaaS company that – ironically – sells AI sales software. The salutation? 'Dear Mr. Müller, I hope this email finds you well.' A phrase so hackneyed it's almost iconic. Then came the kicker: In the second paragraph, I was offered a solution for 'challenges in the media industry.' I've been an industrial journalist for over 18 years. My focus is manufacturing. The thing is, this email wasn't just bad; it was an insult to my intelligence and a perfect example of what's going colossally wrong in B2B sales right now.

We've opened Pandora's Box called 'AI in Sales,' and instead of targeted precision, all we're getting is faster, louder, and, above all, dumber spam. Most companies aren't using these powerful tools to better understand their customers, but to bombard them with irrelevant junk across more channels simultaneously. That's not a sales strategy. That's digital harassment.

Why AI in Sales is just expensive spam for most

Let's be honest: The promise sounds fantastic. An AI that automatically identifies leads, writes personalized emails, interacts with potential customers on LinkedIn, and even initiates initial phone calls. A dream for any sales manager struggling with shrinking margins and a shortage of skilled workers in mechanical engineering. Platforms like Outreach.io, Salesloft, or Apollo.io promise exactly that. They boast 3-5x higher response rates and an ROI that brings tears to your eyes. And what happens in reality? German SMEs – the backbone of our industry – buy these tools for a lot of money and then do exactly what they did before: bad sales. Only now it's automated and on a scale that was previously impossible manually.

The problem isn't with the technology. It's impressive, no question. During a visit to the Siemens plant in Erlangen, I was shown how they process procurement leads for their automation technology with such systems. It's clean, data-driven, professional. But Siemens is not the typical hidden champion from the Sauerland region. The average sales employee at a medium-sized supplier is Hans-Peter, 54, who has known his customers for 20 years and prefers to win new leads at Hannover Messe over a cup of coffee. Now Hans-Peter is suddenly supposed to build 'sequences' in Salesloft and evaluate 'engagement signals' on LinkedIn. This is putting the cart before the horse. People are given a Formula 1 car without being taught how to drive it. The result is a crash – or at best, extremely expensive parking in the pit lane.

The uncomfortable truth behind the glossy figures

Providers and analysts are falling over themselves with success stories. HubSpot, in its 2025 benchmarks, talks about 30-40% engagement in orchestrated multi-channel campaigns. John Deere is said to have generated a $15 million pipeline value with it. Sounds great. But let's take a closer look. The average response rate to a poorly crafted cold email is less than 2%. Combine that with an equally poor LinkedIn request, and the chance of being ignored only increases exponentially. According to a VDMA survey from late 2024, 67% of mechanical engineering companies are experimenting with new digital sales channels, but less than 15% have a dedicated, data-driven strategy for it. The rest? 'Spray and Pray' – only now with AI acceleration.

What does that mean specifically? A salesperson who used to call 50 hand-picked contacts a week now sends 500 AI-generated emails, the content of which is as personalized as a horoscope from the daily newspaper. The result: The response rate might drop to 1%, but because the volume has increased tenfold, he still gets 'more' responses in the end. The fact that he has alienated 495 decision-makers forever and catapulted his company's domain onto various spam lists along the way doesn't appear in any statistics. That's the dirty truth behind the hype. You optimize for short-term, measurable metrics and destroy long-term trust. There's no getting around it.

Multi-channel without AI is 'spray-and-pray'; with AI, it's sniper-like precision. Our response rates have quadrupled since we let AI choose the channel based on buyer signals.

— David Cummings, VP of Sales at Kong Inc. (in SaaStr Podcast 2025)

Cummings hits the nail on the head. The difference lies in the 'how.' But most people just hear 'AI' and 'quadrupled' and run with it.

But... what if it does work? The argument for AI-powered offense

Now I have to be fair. Of course, there are success stories. The companies that not only buy the tools but also do their homework. Last week, I spoke with the sales manager of a precision tool manufacturer from Baden-Württemberg. A classic SME, around 300 employees. Two years ago, they set up a holistic process – and in the right order.

First: They spent months refining their Ideal Customer Profile (ICP). Not just 'Company XY from Industry Z,' but in-depth profiles with technological, organizational, and even psychographic characteristics. Second: They cleaned up their content and created relevant information for each phase of the customer journey. Third: They intensively trained their team – not just in using the software (in their case, it was Outreach), but above all in modern B2B sales craftsmanship: How do I write an email that offers value? How do I use LinkedIn for social selling, not for flat pitches? Only then, at the very end, did they activate AI automation. The result? Their cold outreach response rate climbed from a dismal 4% to a stable 22%. The conversion rate from initial contact to qualified meeting is now over 10%. This isn't magic; it's just damn good work.

What I really see in practice: A battlefield of missed opportunities

When I drive through German industrial areas and talk to sales managers, I mainly see three things. First: A rampant belief in tools. 'We now have Salesforce Einstein and Apollo, so sales must increase.' Wrong. That's like giving a hobby chef a professional kitchen and expecting them to earn a Michelin star the next day. The software only amplifies what's already there. If the process is good, it becomes excellent. If the process is bad, it becomes a catastrophic money-burning machine.

Second: A criminal neglect of the foundation. The Ideal Customer Profile – the ICP – is a foreign word for most or a superficial PowerPoint slide. But AI can only work as well as the data it's fed. If you tell the AI to target 'all mechanical engineers in North Rhine-Westphalia,' you'll get generic mass emails. But if you can tell it, 'Find me all production managers of automotive suppliers with more than 250 employees who have posted about 'efficiency improvement in CNC manufacturing' on LinkedIn in the last 30 days and whose company is currently advertising new positions for mechatronics engineers' – then, and only then, does it make sense. Then the AI can play to its strengths and formulate a truly relevant, almost personal approach.

The Amplifa ICP Playbook: The Foundation for Successful AI Sales Before investing in expensive tools: Define your ideal customer with precision. Our free playbook shows you step-by-step how to develop data-driven ICPs that truly make your AI sales strategy effective.

And third: The fear of humans. We automate like champions, but what ultimately makes the difference – especially with complex, expensive industrial products – is human interaction. Gong.io data is clear: A short, value-adding email, followed by a call two or three days later, is the most effective cadence. What do most salespeople do? They hide behind their email sequences. The telephone receiver seems to have mutated into a toxic object. AI should create space for exactly these valuable, human interactions for the salesperson, instead of replacing them completely. But in many companies, it is misused as an alibi to avoid picking up the phone.

The comparison: What really works in AI-powered sales

Let's look at this soberly. Not every approach is the same. The difference between success and failure – or rather: between a 10x ROI and reputational damage – is enormous.

The table doesn't lie. The leap to AI orchestration is massive – but it's also conditional. Without a clean process and good data, you're just buying the bottom rows of the table in a faster execution for a lot of money.

Sales Automation Services by Amplifa You have the right mindset but not the resources? Our team implements powerful AI sales processes for you – from ICP definition to the go-live of your Outreach or Apollo sequences.

The most pressing questions about AI in sales (and honest answers)

1. Is cold emailing still legal in Germany under GDPR?

Yes, but. That's the classic lawyer's answer, but it's true. The GDPR is not a blanket ban. You can indeed initiate B2B first contacts via email under the legal basis of 'legitimate interest' (Art. 6 Para. 1 lit. f GDPR). However, the requirements are strict: Your company's interest must be weighed against the interests of the contacted person, the content must be highly relevant to the recipient's professional function, and there must be an easy unsubscribe option. Mass spam to info@ addresses is taboo. Targeted, relevant approaches to decision-makers are a gray area, but usually tolerated in practice. Important: Document your balancing of interests! Tools like OneTrust can help keep the process clean. Violations are not minor offenses – penalties can be up to 4% of global annual turnover.

2. Which tools are best suited for a medium-sized mechanical engineering company?

There is no 'best' solution, only the one that best fits your process and budget. The 'Big Three' are Outreach, Salesloft, and Apollo.io. Outreach and Salesloft are the enterprise solutions – extremely powerful, but also expensive and complex. They are the Mercedes S-Class among sales engagement platforms. For many SMEs, this can be overkill. Apollo.io is often the more pragmatic entry point: It combines a huge B2B contact database with solid sequencing functions at a fraction of the price. My advice: Start smaller. Perhaps even with a tool like Instantly.ai or Lemlist for email automation and combine it with manual LinkedIn outreach. Learn the craft first before buying the most expensive tool. Is it really that simple? No, but it's the only sustainable way.

3. Will AI soon replace my sales staff?

No. At least not the good ones. AI replaces repetitive, mundane tasks: data research, sending follow-up emails, logging calls. It does not replace empathy, strategic thinking, negotiation skills, or building genuine human relationships. I bet that in three years, we will see that the best sales teams are not those with the most AI, but those with the best symbiosis of man and machine. AI qualifies leads and books appointments; humans close the deal. Those who understand this will win. Those who believe they can completely outsource sales to a machine – especially in the highly complex industrial environment – will fail.

B2B Intent Data: Know What Your Customers Want Before They Say It The key to precision lies in data. Learn how to use Intent Data to recognize the buying signals of your target customers and elevate your AI-powered acquisition to a new level.

What needs to happen now: An appeal to German SMEs

1. Stop the tool-buying frenzy: Don't buy software until you have your process and strategy on paper. A bad process that is automated remains a bad process.
2. Obsession with ICP: Make defining your ideal customer a top priority. Talk to your best existing customers. Find out what truly drives them. This is the fuel for your AI.
3. Education before automation: Train your team. Not in operating buttons, but in modern sales. How do I research a lead? How do I formulate a value-adding hook? How do I conduct a conversation that isn't an interrogation?
4. Start small and manually: Before throwing 5,000 contacts into an AI sequence, try to process 50 contacts manually, but according to the new principles. Measure the results. Learn. And only then, when the process is solid, scale it with technology.
5. Celebrate the phone: Establish a culture where picking up the phone after a positive email interaction is the rule, not the exception. Humans always win the crucial deals in the end.

We in German industry have always been at our best when we didn't blindly adopt technology, but intelligently integrated it into our proven processes. The same applies to AI in sales. It's a powerful tool, no question. But a hammer has never built a house on its own. It needs an architect, a blueprint, and skilled craftsmen. And that's exactly what's missing everywhere right now.

What do you think? Are we shooting sparrows with AI cannons in sales, or are we missing out on the future if we don't join in? I look forward to your opinion in the comments. Let's discuss!