Buyers are skeptical, AI-generated content is everywhere, and the old playbook isn’t cutting through anymore. In other words, B2B SaaS marketers are in a trust recession.
In this episode of SaaS Half Full, Lindsey Groepper sat down with Chris Silvestri, founder of Conversion Alchemy, to talk about what actually moves the needle: real customer understanding, a strong point of view, and using AI to listen, not just to produce.
The bar for trust has never been lower
Buyers don’t trust polished content. Some marketers are now deliberately adding typos to prove they’re human. Chris has noticed it on LinkedIn, too, where people are writing in dense paragraph blocks just to avoid looking like a bot. These are all reactions to the same underlying problem: AI has made content so uniformly polished that it all blurs together.
“AI can write all about your value and your features, but if you don’t have that particular insight into the market — what customers are doing wrong, how you solve for it — customers won’t really pay attention.”
— Chris Silvestri
The antidote isn’t to write worse. That won’t get you anywhere. You’ll get much better results by having a stronger point of view rooted in something only your team has seen.
Weak positioning shows up in your churn numbers
Most teams use demo conversion rates to gauge whether messaging is working. Chris says churn is the more telling signal. When messaging overpromises, and the product underdelivers, customers leave. That’s worth examining as a potential positioning problem masquerading as a retention problem.
Chris’s process for finding message-market fit starts long before anyone writes a word. His work is roughly 70% research (think internal interviews, customer conversations, competitor messaging audits, and review mining). He’ll even test early messaging through sales collateral before anything goes live because call recordings tell you things that metrics never will. Only once he has a clear picture of how customers actually speak does the copywriting begin.
“First, join the conversation they’re already having in their head. Then you can start explaining who you are.”
— Chris Silvestri
Most B2B SaaS homepages get this backwards by opening with what the company is and does, rather than the problem the visitor is already trying to solve.
AI as an input machine, not an output machine
Chris isn’t anti-AI, but he is anti-lazy AI. Rather than prompting a model to generate content, he feeds real customer data and call transcripts into purpose-built synthetic persona platforms to stress-test messaging assumptions. The result of this approach is directional insight without weeks of customer recruitment.
The output is only as good as the inputs. For B2B especially, you need a real foundation of customer data before synthetic personas add genuine value. And if every persona starts giving you the same answers, word for word, that’s a sign they’re reflecting your own assumptions back at you, not real customer insight.
Check out all SaaS Half Full episodes
Transcript
Generated by YouTube and cleaned up with ChatGPT.
VIDEO INTRO
[0:00] Chris Silvestri
There’s no excuse anymore, right, to not know about your customers when you have so much data at your disposal. Even if you don’t have real customers, you can actually create fake ones. I think it’s a good thing, because companies can actually start putting themselves in their customers’ shoes more — and especially with the context. All these companies didn’t have any research, or it wasn’t organized. Now they’re being forced, in a way, to bring everything together, organize it, and make sense of it, which is a good thing if they use it correctly.
EPISODE
[0:34] Lindsey Groepper
Hi, and welcome back to SaaS Half Full, the only show serving B2B SaaS marketers. I’m Lindsey Groepper, EVP at PANBlast, and I will be both your host and bartender today. I had a great conversation today with Chris Silvestri, who is the founder of Conversion Alchemy. Chris and I are talking about the ways that AI has impacted copy and messaging — and not just how it screwed it all up and there’s AI slop, but also how you can leverage AI to actually create more trust and empathy with your customers. I know, not something we talk about too much. So if you care to grab a drink and join me as I speak with Chris. Hey, Chris, welcome to SaaS Half Full.
[1:12] Chris Silvestri
Hey, Lindsey. Thank you so much for having me.
[1:16] Lindsey Groepper
Absolutely. I know that my team connected with you just a week ago, so I appreciate you getting on with us here just a week after. And where are you based?
[1:24] Chris Silvestri
I’m in the UK, in the south — Winchester — which not a lot of people know was the capital of England in the Middle Ages. So if you look for Vikings and everything, Winchester is the place to go.
[1:36] Lindsey Groepper
It’s only midday for me and I’ve already learned something new. And it makes sense now why our team was not able to get you a cocktail kit to share with me on the show. Did you bring anything at least to share with me to drink today?
[1:49] Chris Silvestri
I have my good sleepy tea, because this is my last thing of the day. So after this I’m going to start my shutdown routine.
[1:58] Lindsey Groepper
There you go. Well, as our listeners know, I always stick through the process no matter what time of day, no matter what day of the week it is. So I am cracking open at 12:30 on a Thursday — it’s a Good Boy vodka, which is basically a non-carbonated vodka water in a can. Cheers, nice to meet you.
[2:20] Chris Silvestri
Cheers. Cheers.
[2:23] Lindsey Groepper
Okay, we are going to dive into our topic here in just a bit, which is all things messaging and copy, specifically as it relates to AI — both how it’s destroyed it and made it bad, but more importantly, what we’re talking about today is how smart brands are able to actually leverage it to create more trust with buyers. So I’m glad that we’re going to be talking about more of the positives versus how AI slop is wrecking us all. But before we do that, I want to just real quickly, Chris, give our listeners a little bit of background on you. You are today the founder of Conversion Alchemy. So certainly want to hear what Conversion Alchemy is, why the company exists, and if you could also just give us a quick little path into how you founded this company, that would be great.
[3:08] Chris Silvestri
Yeah. So Conversion Alchemy is a small consultancy where we help B2B SaaS companies — typically growth-stage with product-market fit — find their message-market fit, which means understanding who they are speaking to, what language they should speak, and also how to convert them into more loyal customers down the road. And we do that through a mix of positioning, messaging strategy, conversion copywriting, user experience design, and the buyer psychology behind all of it. How I got to Conversion Alchemy, it’s a bit of a winding road. I started my work life as a software engineer in industrial automation back in Italy, where I’m from. Did that for 10 years, and in retrospect I learned a lot about putting myself in other people’s shoes, about systems thinking and UX. But I kind of got fed up of all of that and I wanted to play drums with my band. So I decided, why not start looking for how to make money online? And I stumbled on — after a lot of things I tried — copywriting, and I was instantly hooked by the psychology behind it, the decision-making aspect, understanding what makes people tick. So yeah, I definitely stuck to it. I studied on evenings and weekends until I was able to make it full-time, and through other different journeys I worked as a UX lead at a startup, then went back to copywriting with the messaging and positioning strategy, and founded Conversion Alchemy in 2021.
[4:45] Lindsey Groepper
All right. Wow. Both sides of your brain are definitely working on overload, which is unique. Are you still drumming in a band?
[4:49] Chris Silvestri
I’m not drumming in a band, but I’m playing a couple of shows here and there whenever I can. It’s more of a logistical thing right now, just because having a drum set is a bit of a nightmare. But maybe at some point I will have my full-time band as a hobby. We’ll see.
[5:06] Lindsey Groepper
Yeah. What type of music? What’s the way to your heart?
[5:09] Chris Silvestri
It’s typically punk rock. I’ve always been playing stuff like Blink-182, but then we’ve evolved. I used to have my band in Italy, we used to tour Europe. We did Russia twice — lots of places, lots of fun moments. And also some metal here and there. I had another post-hardcore band with screams and all that. Fun stuff.
[5:30] Lindsey Groepper
Love that. I have seen both Blink-182 and Metallica in concert. Neither of them are really on brand for me, but it was like, you know, you just go for the good of the team. And so I went and actually got an appreciation and enjoyed them both.
[5:45] Chris Silvestri
Good.
[5:47] Lindsey Groepper
Awesome. Well thank you for that background. I came across something really interesting on LinkedIn two days ago — it was a discussion around the idea of this trust recession that we’re in right now and how it is just creating chaos for marketers. What was interesting, though, is in the comment section they were answering the question that was posed, which is: what are the least fakeable things that we can do as marketers now? And there was a lot of going back to basics. There were things around events, in-person speaking, getting on the phone. Someone even said, ‘I purposely put typos in my copy now so people know it’s human.’ I’m like, oh my gosh, is this what we’ve come to — is the little red underlined squiggle now a signal of trust versus error? And as a copywriter, I’m sure that’s crossed your mind. So just kind of level set for us where you see things, and from your perspective, where does this leave marketers? What have you been seeing over the last two years that has certainly changed your business?
[7:00] Chris Silvestri
Yeah. Speaking of LinkedIn content, did you see that people started using longer paragraphs now — blocks of paragraphs — just so they don’t look like AI? Which is crazy. I mean, the whole gist of copywriting is to make copy easier to read, and now people are going backwards. But anyway, what I’m actually seeing with my clients and what I’m actually trying to do is go back to the basics in the sense that, first, in order to be in touch with your customers, you actually need to understand your customers. So to me, 70% of my work is typically always research — deep research. And that means understanding the internal perspective of my client’s team: product, support, marketing, the founder. Having that perspective as much as possible. Then the external side of things, which means understanding prospects and customers and how to create the transformation that needs to happen. And also the market side: the competitors, looking at their website messaging, looking at reviews. Having a crystal-clear picture of the landscape — how customers speak, what their motivations are. Only once you have that can you start working on your positioning, messaging, and copy for your website. But I think the clearest path to being more authentic and more human — now and going forward — is having a strong point of view, which is the starting point of your messaging strategy. Until you have that, even if you have good value messaging and it’s not all features like SaaS companies have typically done, if you don’t have a strong point of view now it’s even harder to stand out. Because AI can write all about your value and your features, but if you don’t have that particular insight into the market — what customers are doing wrong, how you solve for it, the typical lies that are spread around — customers won’t really pay attention. You need to grab them with something unique that only you have seen: the specific reason why you’re building what you’re doing. So I think point of view and insight is definitely something that companies will need to invest more into and understand better.
[9:21] Lindsey Groepper
What I love — for people like you and I, and for context: I’ve worked at a PR agency for B2B SaaS and emerging AI brands and been doing this for 20 years. So for people like you and I, part of it is just like, told you. We’ve been saying this for so long. And now all of this — point of view, differentiated thought leadership — there’s even this move towards unpolished points of view, because AI has made every point of view so polished and so buttoned up that it all fades together. So having a little bit of a rougher but authentic point of view — this whole ‘authentic storytelling’ thing is all of a sudden treated like it’s new, and it’s like, come on, we’ve been talking about this.
[10:12] Lindsey Groepper
What are some of the less obvious signs that you’ve seen that a company’s value positioning or point of view is off? There are obvious metrics — like if you have X amount of demos and they’re not converting, or there’s all this traffic but no one’s taking the next step. Are there some less obvious signals that something’s off?
[10:38] Chris Silvestri
Churn is another one. A lot of that stems from the misalignment of expectation between what you message and what comes after — the product. A lot of customers might stumble on your messaging, maybe done with AI, they see something, then they enter the product and find something completely different. That’s again a misalignment. Something else — from sales conversations you typically see the early signs. I really like, whenever I work on messaging, to test it with sales teams before launching any live messaging, especially if I’m working with a sales-led organization. We might do some MVP messaging through sales collateral and test that out in the market first to see the responses we get, and then adjust from those responses to see how much it resonates. And the great thing about that is that with a website, emails, you get a lot of metrics and numbers, but you don’t really get a feel for the resonance — which you only get when you look at those recordings of people talking to other humans, right? So I love to dive into those.
[11:51] Lindsey Groepper
Where do you most often see the disconnect between what SaaS brands are saying and what customers are hearing? I’ve seen a couple of your recent posts that talk about there often being a big disconnect and misunderstanding there.
[12:07] Chris Silvestri
Yeah. Whenever I conduct customer interviews — which I typically do after I speak with teams internally — I definitely see different pictures and different vibes that I then need to connect the dots on and match. But also when it comes to website messaging, the clear telltale sign whenever I do an audit is when the homepage value proposition — literally the first thing on the page — is just introducing the company: what it does, what it is, without actually matching the motivation that people have when they land on the page. The conversation they’re having in their head — that’s how you actually grab them, hook them, and move them down the page. Rather than throwing everything at them, first join the conversation they’re having, and then you can start explaining what you do and who you are. That doesn’t mean you need to be fluffy or airy-fairy. It just means using the right job-to-be-done framing and then moving them down the page or to specific use case pages on the website.
[13:15] Lindsey Groepper
Right. I would imagine if folks are hitting your website, they already have you in a bucket of what you do, and they’re looking at a bunch of companies A through D that do the same thing. So they want to know not what you do, but how you do it differently and what problems you solve. That makes sense. The other area I wanted to lean into — that you talk about a lot — is how AI can actually be used to build empathy. Getting away from the efficiency conversation, which we all agree on — yes, it’s more efficient — but unpack this idea for me of how you can use AI to create empathy, because those two things definitely don’t go together typically.
[13:53] Chris Silvestri
Yeah. The way I see it, I really started appreciating using AI more as an input machine rather than an output machine, which is what a lot of people use it for. A lot of people just prompt it to produce some kind of output. What I like to do is feed it with the right context and then use it to improve and augment the input I have as part of the research. A big thing I do is creating synthetic personas, for example — which a lot of companies I’m seeing have started doing with custom GPTs, like using those commercial models. You prompt it and say, ‘okay, you are a CMO of this company or that one.’ But the problem is that first, a lot of companies don’t feed the right context, or even if they do, those commercial models don’t really work effectively — because those models are pre-trained, they have their own biases and influences. There’s always going to be that sycophantic behavior that tends to want to please you, all of those inherent biases. And so what I started experimenting with — there are a couple of platforms that have popped up that actually do synthetic research as a productized service, which is pretty interesting, because they already include different algorithms and systems that prevent those typical biases from influencing the results. So they’re more reliable. From, I don’t know, 50–60% accuracy that you’d get otherwise, you can get to 70–80%. Which I would say isn’t nothing revolutionary and doesn’t replace message testing or real conversations with customers — but still, when you’re at the early stages, or if you want to do research or testing regularly, even just getting directional feedback on your messaging, your copy, or your ideas, it’s worth a lot. It saves you a lot of time, a lot of hassle, and risk later down the line. And the main thing is the context — organizing your context so that your personas are more realistic because they’re based on real data, rather than just using the whole knowledge of the internet that AIs have access to. Which is interesting, because just in the last couple of days I saw the HubSpot CEO talking about an agentic customer platform. They’re basically about to launch their own agentic CRM, which is in three layers. The first layer is context — constantly collecting data from customers and internal teams. The second layer is collaboration. And the third layer is action — agents taking action. I think the future is going to be oriented in that direction more and more: constantly gathering context, building your synthetic personas or your systems to validate and test internally, and then taking some kind of action that’s informed by data.
[17:13] Lindsey Groepper
Yeah, it’s interesting — all these new terms are popping up. My last interview was with a founder of Heatseeker, and they do synthetic personas and market research. I don’t know if that’s one of the ones you’ve come across. And then I had a conversation with a company I’m looking to bring on as a client this week, and they’re fitting into this new category — trying to create this new category of context management. So it’s no longer just the data management side. It’s: as you’re creating either AI systems or agents, how do you put context around all of this data for all the different people trying to touch it, use it, and safeguard it? All of these terms are starting to bubble up, and the old categories are no longer going to serve the new problem. We’re going to see the demise of some pretty historical categories, at least for us in the SaaS space, and have these new ones emerge. It’s hard to get your head wrapped around all of it. Even in my conversation with the market research and synthetic persona company, it was all Greek to me right away — but I have to start wrapping my head around it. You mentioned HubSpot, which obviously has massive data sets. Do you work more with the emerging SaaS side of the house — companies that clearly don’t have much data? If so, how does your process work with them, or what suggestions do you have for them?
[18:41] Chris Silvestri
Yeah. I mostly work with companies that either have very sparse research data or have never really done customer research or market research. If they have customers, obviously we do the whole process — we try to interview at least 10 to 15 customers. If they don’t, one thing I’ve been doing lately — yesterday actually, with another company — we’re trying to go to market with a new segment of their personas. The problem is they don’t have a lot of conversations, and it’s also hard because these are bigger enterprise clients, so it’s hard to recruit them or do interviews. What we’re doing: they have a couple of recordings, so I’ve been able to watch two, three, four recordings. I’ve created my assumptions and ICP sheet and profile based on that. But to augment and fill in the gaps to create an early customer persona, I basically created an interview guide asking: what are some questions we can ask to fill in the gaps in this persona? Then we created mirrors of those personas — 20 synthetic profiles — with a platform called Synthetic Users. And then we basically conducted those interviews with these synthetic personas, which were also created from transcripts of the actual calls. So in theory, what we did was: we have these assumptions plus the actual voice of the customer, and in theory these personas were able to reconstruct them but also expand and transform these assumptions through different lenses — different locations, different types of companies — so we could get a more accurate idea. And we used that literally to refresh our positioning, our messaging, and the copy we’re going to write. I think this is a pretty good way to empathize with your customers. That’s why I say: use AI to empathize, because at least it forces you to put yourself in your customers’ shoes. Especially now — there’s no excuse anymore to not know about your customers when you have so much data at your disposal. Even if you don’t have real customers, you can create synthetic ones. I think it’s a good thing, because companies can actually start putting themselves in their customers’ shoes more.
[21:30] Lindsey Groepper
Can you distill that use case down for us further — without mentioning the company name — but give us an assumption, the theory, what buyer persona you created, and the outcome?
[21:43] Chris Silvestri
So it was a nonprofit — not strictly a SaaS — that I’m working with. They had the partner side of things. They have three different types of partners, and with those partners they’re still trying to validate their offer and their pricing. Some of the assumptions were around their buying committee. We needed to know: we know this is the main buyer, but we’re not entirely sure about these two roles — do they need to be in the decision-making process, or are they not really someone we need to speak to? So we used those synthetic personas to actually flesh out those insights and understand: do they enter the picture? And we got good insights because we got a pretty good idea of the different layers of when these different characters come into play in the decision-making process — so we actually know, okay, we need to flesh this persona out a bit more, we need to describe how they collaborate and how they discuss the deal with this other persona. I think it’s a pretty good use case because it doesn’t need to be super specific, but at least it gives you good insights into how you can structure your messaging and your copy, maybe what pages to create to speak to these buyers. It’s pretty valuable. Even if you’re not 100% sure, it’s still a pretty good shot when you’re just creating your beachhead segment.
[23:20] Lindsey Groepper
Okay, that was a great example. Thank you. It just helps people get their heads around it — especially these synthetic personas. I think the first thought behind it is: okay, so we’re trusting fake people to give us real-world answers and solutions. But I suppose the synthetic persona is only as good as the data inputs you’ve used to create it. You can’t create it out of nothing.
[23:44] Chris Silvestri
Exactly. Yeah. You have to have some basis of customer conversations — real conversations. Especially in B2B. B2C you might be able to get around it just because there’s a lot more data around, but B2B, yes, I would definitely always start with some kind of real data — whether it’s reviews you find online, whether it’s case studies you scrape from a competitor’s website. At least something to start with. But also, a good example of something that happened in that exercise: from my persona profile, there were a lot of assumptions when we created these cloned personas, and we fed all of the insights into our strategist AI to analyze. It clearly pointed out the assumptions that these personas basically repeated almost verbatim. There were a lot of things these personas said in exactly the same way. And we realized: if they all said the exact same thing, maybe this is actually just an assumption we have — it’s not something we can claim, because in reality a lot of people would have different opinions on that specific point. So that’s an example of: okay, maybe let’s not trust this thing entirely, because it’s way too absolute.
[25:08] Lindsey Groepper
Okay, that makes sense. Well Chris, is there anything that we haven’t tackled that you want to make sure we cover in this conversation today?
[25:17] Chris Silvestri
No, it’s been great. I think lots of good stuff. I really like talking about these innovations, even though a lot of people are kind of skeptical.
[25:23] Lindsey Groepper
Yeah.
[25:27] Chris Silvestri
But this is the way. This is the way forward. And again, I’m personally just having more and more conversations that are all in this sphere. So this is where we’re going.
[25:37] Lindsey Groepper
Well, Chris, I end every episode the same way, which is I ask my guests if they have a signature or favorite toast to send us out.
[25:46] Chris Silvestri
Chin chin. I would say — I’m Italian, I can say chin chin.
[25:50] Lindsey Groepper
We’ve not had a chin chin in five years. That’s nice. Appreciate it. Great to talk with you.
[25:57] Chris Silvestri
Thank you.
[25:59] Lindsey Groepper
Thanks again to Chris for joining me on SaaS Half Full. I hope you took a couple of things away from that — maybe how you’re looking at your homepage copy, or thinking about synthetic personas, because now we’ve had a couple of episodes here in the last month talking about how to leverage those for better customer insights. We always appreciate the listen, and until next time — bottoms up.