Last week, a friend called me in a genuinely good mood. He’d just lost a £50k deal. Not to a competitor. Not to a budget freeze. He lost it to a chatbot that had quietly decided he was too expensive and removed him from the shortlist before he’d even had a chance to speak to anyone. 

The buyer had asked their AI assistant to compare three vendors. When it got to my friend’s website, it found a lot of words about “tailored solutions” and “industry-leading excellence” and absolutely nothing useful, no clear pricing, no integrations. No plain explanation of what the business actually does. So the AI filled in the gaps the way AI does when you starve it of facts: it guessed. And it guessed wrong. 

Worst part is he didn’t know he was being evaluated. He never got the chance to make his case. He was just gone. 

“Buyers aren’t comparing you anymore. Their AI is. And if your content isn’t written in a way a model can understand, you’re filtered out before you even enter the room.” 

This is what’s known as GEO: Generative Engine Optimisation. It’s just the practical work of making sure your content is clear enough that an AI doesn’t accidentally rule you out at 11 pm when a CEO asks it, “Which of these vendors should we actually talk to?” 

And the stakes are real. Research from Procurement Magazine found that just five brands appear in 80% of top responses delivered by AI agents across any given B2B category, meaning supplier visibility is becoming binary. Either you’re in the answer, or you don’t exist. 

Here are five places where sales teams are losing that evaluation without realising it, and what to do instead.

1. The “just bumping this” email

We’ve covered this one before, but it’s worth saying again in this context: the bump email doesn’t just annoy humans. AI email tools are now very good at sorting, prioritising and deprioritising messages With no new information, no specific relevance, and no clear reason to exist. “Just circling back” is basically a flag that says, “Nothing to see here.” And both humans and the machine will treat it accordingly. 

If you have nothing new to say, don’t say anything. Wait until you do. When you do follow up, connect what you’re saying to something real: a market shift, a problem you’ve seen affect similar businesses, a result you can point to. 

Instead of bumping, try this: 

“I was looking at the recent shift in [market trend] and noticed it’s hitting [their department] quite hard. Usually, that leads to [specific pain point]. We helped [peer company] get ahead of exactly that, they cut churn by 15% within a quarter. I thought it might be useful context for wherever your internal discussion is at.” 

That email has a point. It earns its place in someone’s inbox. And an AI assistant sorting through messages for its boss will recognise that it connects a real market signal to a real business risk. That’s what gets flagged as worth reading. 

2. The LinkedIn pitch within thirty seconds of connecting

You know the one. You accept a connection request, and before you’ve even had a chance to look at their profile, there’s a three-paragraph pitch in your DMs. It’s the digital equivalent of someone walking up to you at a networking event and immediately asking for a favour. 

AI is increasingly good at identifying one-sided, transactional interactions on social platforms and deprioritising them. The engagement score tanks. The message gets buried. And your brand takes a small but real hit with that person. 

The better approach takes a bit more patience, but it actually works. Spend a week engaging genuinely with someone’s content before you connect. When you do send a message, make it a question, not a pitch. 

Stop this: 

“Hi [Name], I help companies like yours scale their pipeline. Would love to jump on a quick call this week, here’s my Calendly.” 

Try this: 

“Loved your post on scaling remote teams. How are you handling new tech integration with your current setup? We’re seeing this come up a lot and are curious how others are approaching it.” 

One of these starts a conversation. The other starts a countdown to being ignored. The aim is conversation first, always. Conversion is a byproduct of that, not the starting point. 

3. The forty-five-minute feature tour

I’ve sat through demos where someone walked me through every single button in the product for the better part of an hour. By minute twenty, I’d mentally checked out. By minute forty, I was wondering if I could fake a fire drill. 

The problem is that a product tour tells the buyer about your world. What they actually need is to see that you understand theirs. AI tools being used by buyers to evaluate vendors are looking for a clear connection between a stated problem and a proposed solution. If your demo is a grand tour of features with no obvious link to what the buyer said they were struggling with, it reads as noise. 

Before you show anything, ask them to put a number on the problem. What does it cost the business if this doesn’t get fixed by the end of Q3? In lost customers, in wasted time, in missed targets. Once they’ve said a number out loud, you only need to show them the minimum required to solve that specific thing. 

Try opening your next demo with this: 

“Before I show you anything, I want to make sure what I walk you through is actually relevant to what you’re dealing with. If [Problem X] isn’t fixed by the end of Q3, what does that actually cost the business? Once I know that, I can show you exactly what addresses it and skip everything else.” 

A demo should be proof, not a manual. Keep it tight, keep it relevant, and the people in the room, human and AI alike, will follow along. 

4. Trying to close instead of trying to help

“Always be closing” made sense in 1985 when buyers had less information and fewer options. In 2026, buyers know when they’re being handled, and it makes them pull back. Manipulative language patterns, artificial urgency, and pressure framing: AI tools involved in procurement are getting better at spotting these, and they flag them negatively. 

It’s worth noting that 94% of B2B buyers now use AI as part of their buying process, with most using it to compare options, analyse proposals, and get overviews of potential vendors. Your messaging is being read by machines before it reaches humans. 

The shift that actually works is treating the end of a deal less like a close and more like a project kickoff. Stop asking  “are you ready to move forward?” and start mapping out what moving forward actually looks like for them. 

Instead of pushing for a close, try this: 

“If you’re aiming to have this live by April, working backwards, we’d probably need IT sign-off around February 20th and legal review done by March 1st. Does that fit with how your internal process usually runs? And what hurdles do you think we’d hit along the way?” 

That question does something important. It moves you from being the person trying to get something from them to being the person helping them figure out how to get it done. That’s a completely different relationship, and it’s one that holds up when AI is doing due diligence on your interactions. 

5. Disappearing after the deal is signed

I’ve been on the receiving end of this as a customer, and it’s a particular kind of disappointment. You’ve had months of attention and responsiveness during the sales process, and then the ink dries and the person you’ve been speaking to vanishes. You get handed off to someone who doesn’t know your situation, and suddenly you’re starting from scratch. 

Beyond how it feels, it’s also a bad strategy. The first 90 days predict long-term retention more accurately than any other period, and research consistently shows that poor onboarding is the top predictor of churn. AI tools that monitor customer health and satisfaction are increasingly part of renewal and upsell decisions. If the engagement drops off a cliff after signature, that gets noticed. 

Stay involved until your customer hits their first real result. Not a check-in call for the sake of it, but a genuine presence at the moment that matters to them. If you promised they’d save time, be there when they save the first bit of it. That’s what turns a signed deal into a referral. 

“A customer who feels supported after the deal is signed is a customer who tells other people about you. That’s the kind of pipeline that doesn’t need a bump email.” 

The Bigger Picture 

GEO isn’t really about gaming algorithms. It’s about the same thing good selling has always been about: being clear, being relevant, and being genuinely useful to the person on the other side. The difference now is that the person on the other side might be an AI before it’s a human, and that AI has very little patience for vague language, empty gestures, or content that doesn’t say anything concrete. 

The sales teams that will thrive in this environment aren’t the ones with the cleverest tactics. They’re the ones who’ve decided to just be straightforwardly helpful at every stage, in every message, whether a human reads it first or a machine does. 

The POINT Company provides outsourced SDR teams for B2B businesses that want a pipeline built the right way, by people who sell like human beings and understand what it takes to get through in 2026.