Stop Buying AI Tools and Start Building a Sales System

(upbeat music) - I want you to picture a conference room in your own building on a Thursday afternoon. A whiteboard on the wall is filled with lively drawings. Someone dedicated nearly an hour to sketching various boxes. Each box represents an important system, your revenue team depends on. While the arrows illustrate how information flows seamlessly between them. By the sessions end, there are about 11, maybe 14 boxes, and the arrows have become quite tangled, making it difficult to trace a clear path from a buyer's initial interest to your salesperson making that vital call. Eventually, somebody says, the thing everyone in the room is already thinking, how did we end up with this? Here's the honest answer, nobody designed it. It just evolved. A few years ago, someone put together a spreadsheet because a report was missing. Then someone added three important fields to the CRM because a manager wanted a clearer view of a particular quarter. Later on, someone purchased a point solution after seeing a very impressive demonstration, especially since a competitor had something similar. Another time, someone created an approval process because a single, unusual deal didn't go as planned. Eventually, the person who built the spreadsheet left the company and the one who requested the extra fields moved to a different division. Over time, nothing was ever removed because removing something means knowing why it was there in the first place and unfortunately, no one remembers anymore. And eventually, we look at that accumulation and call it our sales process. I'd like to clarify something important. This isn't about incompetence. Every decision made was reasonable at the time, each solving a genuine problem for someone under real pressure. The issue isn't with individual decisions, rather it's that there were only decisions without a larger guiding plan behind them. Now, let me show you what that costs because the bill arrives every single morning. Your best salesperson starts their day at 8 o'clock, savoring a full cup of coffee and looking over a promising pipeline. By 8.30, they're often hopping between different systems, not because the task is complex, but because the answer to one simple question is scattered across multiple places. Now, what actually happened with this account? Some details are in the CRM, others are buried in emails, some are in call recordings, no one has listened to, and some exist only in the memory of a colleague who's currently on an airplane. So your salesperson works hard to reconstruct the story. They carefully piece together fragments, make educated estimates, and fill in the gaps with thoughtful guesses. Then they craft an email based on those insights. By 10 o'clock in the morning, two hours into the workday that talented and valuable professional hasn't yet spoken with a customer, identified a business challenge, changed anyone's opinion, or built any trust. Instead, they've been serving as a data clerk, laying the groundwork for meaningful connections to come. This isn't just a story, it's backed by solid analysis. Did you know that salespeople nowadays spend only about a third of their time on activities that truly bring in revenue? A good chunk of their week, almost two thirds, is taken up with internal processes, administrative tasks, and document management. Interestingly, the average salesperson juggles around 10 different tools just to close one deal. And here's a surprising thought. Almost 30% of these professionals feel that fewer tools could actually help them work more efficiently. Take a moment to think about that. Nearly a third believe that the very technology meant to support them might be the biggest obstacle. So what do we do about it? Almost always, we buy something else. That's just how it goes. When revenue slows, we like to showcase what we have through demonstrations, which are always quite impressive because they highlight the product's strengths. We then move forward with negotiations rollout and announcing it to the whole company. For about six weeks, there's genuine excitement and enthusiasm. But unfortunately, that excitement often fades, usage quietly declines, and suddenly the new system feels like just another item on the whiteboard. Industry research shows that about 70% of AI implementation projects don't meet their initial expectations. The surprising and important part is that most of these failures could have been easily avoided. It's not that the technology was lacking, rather the real issue is that many decisions about what the system should do were not defined before the purchase. There's another important factor to consider. The direct tool cost accounts for just about 20% of the total implementation expense. The remaining 80% goes toward integration, training, change management, and ongoing optimization. This means that the purchase order is actually the smallest decision you'll make. And it's often the only one really up for discussion. I've often heard owners and sales leaders ask me, what AI should we buy? I totally get why. It's urgent and feels like the right move for a decisive leader. But honestly, that's not the best question. Asking it can lead to 14 boxes on a whiteboard and no real engine powering them. The right question sounds almost too simple. Where is human time being wasted? Because our revenue system isn't doing what it should. That question does not begin with a vendor. It begins with a design. I'm Sean O'Shaughnessy. And my passion is helping company owners unlock the full potential of their businesses by boosting their revenue. I work closely with owners to strengthen their sales management, refine strategies, optimize processes, develop their teams, and craft compelling messaging. Together on this podcast, we will address the often overlooked revenue structures and the hidden costs of not intentionally designing them so that business owners can achieve greater success in clarity. Right now, this podcast operates without accepting advertising revenue and sponsorships or paid placements. I'm not compensated by software vendors to mention their products. When I discuss tools, categories, or use cases, it's entirely based on my personal experience. My goal is to share insights on how sales organizations can generate revenue responsibly and effectively, all while reducing unnecessary selling efforts. I will occasionally share my thoughts on the B2B Sales Lab, a wonderful community I co-founded where sales professionals and leaders come together to exchange ideas, overcome challenges, and support one another in creating better sales systems. It's important to note that the lab isn't a sponsor of this podcast. Rather, it's a space dedicated to helping sales people and sales leaders improve and excel. It's a fantastic platform to help good salespeople become truly great at what they do. If you're interested in exploring any of the ideas I share on this podcast, I'd love to connect with you in the B2B Sales Lab where you can tag me in a post or comment or even send me a direct message. Also, feel free to reach out to me through my website and book a time that works for you. Just visit newsales.expert/calenders. Yes, newsales is one word.expert/calenders with an S. I understand that's quite a few characters, so I've made sure the link is right there in the show notes for your convenience. The philosophy I want to put before you in the next few minutes is easy to state, but genuinely difficult to practice. Architecture before acquisition. The old model emphasizes acquisition first. When you feel a pain point, you identify a category, purchase a product, and quietly hope it will lead to an improved process. Often, it doesn't. A product represents an opinion on how work should flow crafted by a vendor who may not have a full understanding of your buyers, your sellers, or your margin structure. When you prioritize buying before designing, you're essentially outsourcing the design of your revenue organization to the account executive who delivered the most polished demonstration that quarter. The new model completely changes the order. First, you start by designing the operating model, deciding what should happen and what sequence who will perform each step, what information they'll need, the appropriate speed, and the evidence required. Only after this do you choose the technology selecting tools that support a decision you've already made. In this approach, technology acts as an amplifier for your decisions rather than a substitute for where you might be hesitant to decide. I call the resulting architecture the cognitive revenue engine and it rests on one operating principle. Automate the input, humanize the output. Many organizations often misunderstand how to allocate their resources. They rely on their most valuable team members to handle the most tedious and time-consuming tasks, such as manual research, record maintenance, CRM logging, and account history reconstruction. Only after that do they ask artificial intelligence to create the customer-facing content. Unfortunately, this approach leads to tired sales people and generic messages that savvy buyers quickly recognize and dismiss. By shifting focus and leveraging AI earlier in the process, companies can empower their teams and deliver more personalized, impactful communication. Think of it this way, machines are truly remarkable at their tasks, watching, assembling, and repeating with incredible precision. They can keep an eye on a thousand accounts for leadership changes, monitor your website at two in the morning on a Sunday, or notice when a former champion switches jobs, sometimes even within the same week. Four people from one organization begin exploring your content. They effortlessly compile CRM histories, call transcripts, firmographics, past proposals, support issues, and buying signals faster than any human could switch between browser tabs. And the best part? They do all of this on a quiet Friday afternoon and December, never bored, never forgetful, and always ready to research every account, no matter how small. Humans have a valuable place in different contexts. We understand that ambiguity is part of the process. We notice when the person speaking the most in a meeting may not actually hold any real power. We catch when the economic buyer mentions that something sounds interesting, even if they don't seem genuinely interested. We're aware of political tensions within a buying committee. We're comfortable asking unexpected follow-up questions, gently challenging a customer about a decision that might not be in their best interest, negotiating without giving up too much, and helping a nervous group of people feel confident enough to take the next step. The goal has never been to replace salespeople with artificial intelligence. Instead, it's about easing their workload by letting machines handle tasks they're naturally better at. So salespeople can focus on what they do best. When you design the engine in this way, it transforms from a simple, one-time workflow into a continuous learning loop that keeps evolving. I'll share eight important verbs to capture this process, and if there's one thing to remember from this episode, let it be these. Sense, resolve, understand, reason, orchestrate, act, engage, learn. Pay attention to what's happening in your market, determine who and what your signals relate to, grasp the commercial context surrounding them, reflect on what the evidence suggests, coordinate the best response, take action automatically when machine execution makes sense, involve a human when judgment is needed, learn from the results, and then repeat the whole cycle with even more insight. Now, I want to address something directly, because I think the provenance of this framework matters more than the framework itself. You might think I came up with 12 neat categories for 12 podcast episodes, but that's not quite the case. The real timeline is actually more helpful to you than the diagram. In episode eight, the AI sales process map, I shared that the true strength of artificial intelligence and sales lies not in individual tools working alone. Instead, it comes from seamlessly connecting every part of the sales process, which enhances human abilities at every step. This was my initial way of explaining it, even before terms like "agentic" or "common in meetings," the core idea remains the same. Stand-alone automation offers limited benefits, but a well-integrated system creates much greater value. In my episode 30, titled "Why B2B Sales Teams Miss Targets" an AI operating model to eliminate admin drag, I finally had a clear name for the idea. I contrasted what I called "artisan sales." Where the salesperson personally conducts research, writes logs, and follows up with the cognitive revenue engine. In this tech-driven human-centered model, machines handle data mining, formatting, and repetitive tasks, leaving humans to focus on empathy, judgment, and relationship building. That episode also laid out the operating principle I mentioned earlier, automating data input while making the output feel natural and human. It also introduced an exciting new perspective of viewing the salesperson as editor-in-chief, rather than just the author. In episode 36, the 12-part AI revenue stack that reclaims selling time and drives revenue growth, I outlined the architecture into 12 categories and addressed some pushback I received from listeners. The goal isn't just to own 12 separate pieces of software, rather it's to weave these 12 capabilities into a seamless, intelligent revenue engine that relies on trustworthy, well-governed data. That episode also highlights a key insight that reshapes the entire conversation for owners. Teams with embedded artificial intelligence see a 77% increase in revenue per representative. This isn't just a productivity stat, it reveals a fundamental performance gap. Let's take a brief look at the 12 capabilities. I'll move through them quickly, as the real value isn't in the categories themselves. We have category one as autonomous customer relationship management, and with sales engagement and revenue action orchestration, sales intelligence and identity resolution as the next two categories. The fourth category is conversation intelligence and coaching, which I have talked about many times on this podcast, including episode 24 titled Call Analysis and Coaching with AI, part one core coaching skills, and episode 25 call analysis and coaching with AI, part two, building a feedback loop. The fifth category is revenue intelligence and predictive forecasting. The sixth and seventh categories cover similar concepts with inbound orchestration and automated scheduling and lead routing and lead to account matching. Category eight is a big one covering account-based marketing and predictive intent. I also cover category nine repeatedly in this podcast, and it is workflow automation and enrichment waterfalls. Sales enablement and digital sales rooms is the tenth, and then we wrap it all up with categories 11 and 12 covering sales, performance management and incentive compensation, and finally, autonomous prospecting agents. The true value lies in the strong connections among the parts. Conversation intelligence seamlessly updates the CRM, which in turn enriches the buyer workspace. How we behave in the workspace influences revenue intelligence and guides the next best action. That action then feeds into the orchestration engine, creating a smooth interconnected system, remove any one of these links, and it stops being an engine, turning into just inventory. Everything works together to create a dynamic and effective process. Let me share the specifics of the technologies involved because I believe you'll find them interesting. At the memory layer, Salesforce and HubSpot serve as the foundation with pipe drive noting that CRM software can deliver an impressive average return of $8.71 for every dollar invested. Moving up to the orchestration layer, outreach and sales loft have transformed from simple email sequencers into comprehensive revenue action hubs with sales loft boasting an exceptional 398% ROI and doubling win rates. At the same time, outreach reports revenue increases of 13 to 15%. At the identity layer, Zoom info, cognizm, and data axle help identify who someone truly is. The intelligence layer features gong and clary, which analyze conversations and buyer behaviors to gather valuable evidence. The intent layer with 6 cents and Bambora keeps a close watch on market trends. For automation, clay, make.com, Zapier, and N8N work together to connect and streamline enrichment processes. In terms of enablement, seismic and high spot effectively support sales teams. At the compensation layer, exactly and captivate IQ make managing rewards straightforward. Lastly, at the prospecting layer, regg.ai, Apollo, and qualified help find and engage potential customers. Here's an important insight about that list. The particular vendor isn't as crucial as the underlying architecture. There's a benchmark called the AI Native Score that assesses how thoroughly artificial intelligence is integrated into a platform rather than just added on. Examples with a score above 80 tend to deliver about 2.8 times higher returns around 241%. Compared to 87% for traditional tools. The key difference lies in the architecture, not just the surface features. When the architecture is well designed, the outcomes are truly impressive. For example, Cap Gemini used AppTivio's buyer intent platform and experienced a 40% increase in high intent leads, along with nearly five times as many marketing qualified leads. Companies using demo desks assistant saved a total of 6,700 hours on call prep, follow-up and CRM updates, freeing up 20 to 30 hours per person each month and doubling their close rates. Inside sales saw a 30% increase in productivity after implementing an intelligent dialer, resulting in 45% more contacts, 35% less idle time, and a 240% return in just one year. On the forecasting front, 40 net hit 97% forecast accuracy with Clary. Uipath enjoyed a -- 448 percent return over three years, and Honeywell generated $150 million in revenue with Aviso, saving $1 million on CRM licenses. Jumio doubled its proposal responses without extra staff and improved its win rate by 20 percent. Additionally, Salesforce reports that users of generative AI save about 11.4 hours each week. Every one of those is a story about a system, not a purchase. Notice how the biggest gains often come from unexpected places. Usually, they aren't from pushing your top performers to work more on the accounts they already handle. Instead, they come from areas of your market that get virtually no human attention, simply because there aren't enough hours in the day. For example, ConversaGa created a compelling case study with the LA Film School in which their assistant was directed to leads that human reps had already deemed unresponsive and had almost abandoned. This effort brought back an extra 1 percent of those previously written off leads. While 1 percent might seem small, it translates into a significant impact between $4.2 million and $7 million in recovered revenue. Similarly, CallMiner took an innovative approach by integrating customer health scores into their account platform. Accounts flagged as healthy that were targeted with this approach saw a 10 percent increase in their annual contract value compared to controls. Moreover, 48 percent of at-risk accounts improved to healthy status. And struggling accounts were three times more likely to improve than those managed solely by humans. In both examples, the key takeaway isn't AI versus your top salesperson. It's AI versus doing nothing at all. The engine also transforms what your sales managers focus on. And I believe this is often overlooked. A huge part of sales management today revolves around gathering information. Did you reach out to them? What was the outcome? When is the upcoming meeting? Why is this still forecasted? These questions tend to have little value and are typically asked because managers lack clear visibility. When designed effectively, the system can automatically track whether calls were made, what was discussed, if the buyer was engaged, and if there's a confirmed next step. This allows managers to shift their focus to questions that truly influence results. Like, what's holding back the economic buyer from engaging? Why do we often lose momentum after technical validation? Or with three deals featuring strong user champions, but no executive sponsorship, what's our strategy? The first set of questions is all about gathering data. The second is about coaching. The purpose of the engine is to handle the first. So your managers can dedicate their time to the second. If you're exploring where your architecture might need some extra attention or wondering which layer to focus on first, that's a really lively topic within the B2B sales lab community. Members openly share their real world maturity assessments, how they decide on sequencing, and quite helpfully what they wish they'd fixed before moving forward. You're warmly invited to join in the conversation and enjoy a free trial at b2b-sales-lab.com. Just remember it's the letter B, the number two, the letter B, dash sales-lab.com. Now, let me be honest about the risks, because I would be doing you a disservice if I made this sound inevitable. The first challenge is pilot fatigue. When experimenting with new technology, it's easy to see the difference between a small test and full-scale implementation. A pilot might involve a small team working with clean data in a controlled setting. However, taking that into production requires much more. Infrastructure, integration, security checks, compliance, monitoring, and ongoing maintenance. Tasks that may take three months often end up taking 18, as one healthcare AI leader shared, without a clear strategy, you're just chasing the latest novelty. Running numerous pilots can lead to disappointing results and missed opportunities for real value. The second risk to consider is governance where I would encourage you to pay extra attention. Only about 21% of companies report having a well-developed model for overseeing autonomous agents. The leading concern for 73% is data privacy and security. This is followed by legal and regulatory compliance at 50%. And governance capability and oversight at 46%. If you're planning to allocate any part of your revenue process to a system that operates independently, it's important to set clear boundaries on what it can decide independently, implement monitoring to detect any unusual activity, and maintain a detailed record of all actions taken. The third risk is the one I care about most. Automation acts like a powerful amplifier. When you automate a good process, it gives you greater leverage and efficiency. But if you automate a bad process, it can lead to widespread mistakes. Artificial intelligence makes this especially risky, as it can make a faulty system appear impressive. For example, an email might be perfectly crafted, the routing rule works flawlessly, and the dashboard looks great. But the underlying answer could still be wrong. In fact, a poorly designed routing rule that is automated flawlessly can be more harmful than one done manually, because it speeds up the wrong decision across many cases. I want to emphasize an important point. Artificial intelligence doesn't eliminate the need for management decisions. Rather, it makes them more explicit. For example, what qualifies as a high-value inquiry? Who owns a subsidiary of an existing customer? What should we do when two rules clash? How do we handle things after hours? What if the assigned salesperson hasn't taken any action in three days? In the past, these questions were answered informally by whoever cared the most. But now, an engine can't operate on informal decisions. It needs clear choices from us. For owners, especially, this becomes more than just about boosting productivity. It shifts towards enhancing overall enterprise value. Relying on heroic efforts in a revenue-focused team can be risky. It's the people who understand key relationships, the nuances of forecasting, and the reasons behind why certain accounts choose us that truly add value. Savvy investors, acquires, and lenders all recognize this. When a company's revenue hinges on just a few irreplaceable individuals, its value can be significantly less than that of a business with a reliable, repeatable system in place. One that empowers talented team members to excel. Remember, the goal isn't to diminish individual talent, but to create an environment where knowledge remains, even when someone moves on. Now, that difference between designing an engine and gathering tools is the most costly strategic choice. Most sales organizations will face in the next couple of years. It's exactly the kind of decision where learning from others who have already navigated it can be incredibly helpful. They can share what it costs them and what they learned along the way. That's exactly why the B2B sales lab is here. Join us to connect with others building similar solutions and start with a free trial at b2b-sales-lab.com. So here is your new tactic, and I'm going to name it so you can actually use it on Monday morning. It is called the architecture audit and it has exactly three parts, bloat, silence, friction. Bloat occurs when multiple systems redundantly perform the same task. To identify this, ask yourself a simple but slightly uncomfortable question. How many different places does a salesperson need to visit to understand a single account? If the answer is seven, what you have is an extreme line technology stack. It's more like a scavenger hunt. Take a moment to jot down all the places a seller must go to find an answer to just one basic question, and you'll quickly see where the redundancies are. Recognizing this can help you make your processes more efficient and less frustrating. Silence often indicates areas where your system isn't paying attention. Are your buyers visiting your key pages without your sales team realizing? Do closed lost opportunities vanish without a trace when marked as closed? Do former champions leave for new companies unnoticed? Are there vital details in your customer conversations that never get organized into data? Take a moment to reflect on where your organization might be missing these signals because that's where revenue could be slipping away silently without any alert. Friction is the human cost. Instead of asking your struggling salesperson, turn to your best one for insight. Simply ask, what manual task stops you from spending more time with customers? If your top seller shares that they lose an hour each day on preparation, research, and updates, remember it's not about coaching or effort. It's a systems issue. And it's up to you to fix it. Blow silence friction. Three lenses, one afternoon, no purchase orders. Now here's your first step. And I want to be specific about how small it should be. Your first step is not to schedule 12 demonstrations. Your first step is to prove the architecture manually on exactly one process. Identify the main area where skilled people are doing repetitive, low value tasks. It could be managing meetings, researching prospects, routing inbound inquiries, or putting together proposals. Focus on just one. Be honest in measuring it. How many hours per week are spent? How many people are involved? What's the lag time between an event and when you're aware of it? And finally, consider what that delay might be costing you and missed opportunities. Before making any purchases, take a moment to sketch out that process on paper and ask a simple yet powerful question. If we were creating this process today with all the automation and AI tools available, would we design it exactly this way? Most people find themselves saying no, and the moment you do, you begin to take control. Starting from just inheriting your sales process to truly designing it to fit your needs. Take a hands-on approach by running the redesigned process manually for two weeks, perform the enrichment, routing, and recap writing yourself. It might seem tedious, but that's part of the learning process, as it helps you understand the philosophy before investing in automation. When you see improvements with the manual method, you'll have earned the right to automate knowing exactly what you want the technology to do. Right now, inside the B2B sales lab, we're discussing the sequencing. What to build first and how different parts depend on each other. You can join this exciting conversation with a free trial at b2b-sales-lab.com. That's the letter B, the number two letter B, dash sales-lab.com. Let's focus on simplifying bureaucracy first, removing unnecessary hurdles, and then automate what's left behind. It's important to build the right foundation before jumping into tools. Remember, salespeople aren't hired just to hunt down email addresses, transfer data between systems, or right meeting summaries on a Sunday night. They're hired because what truly drives revenue is the ability to connect with people understand their needs, challenge incorrect assumptions, build consensus, and earn trust. That vital work isn't going away. In fact, it's becoming even more valuable. Forward thinking companies won't use artificial intelligence simply to replace sales teams. Instead, they'll harness it to eliminate barriers to selling effectively. The future of B2B sales isn't about choosing between humans and AI. It's about humans amplified by AI. Let's build that future together. [MUSIC]