Eliminating the discovery call with AI
Still a work in progress but wanted to get this one out there to iterate on
Discovery calls are crucial for understanding new users and prospects. It lets us learn their most pressing problems in their own words.
Most companies need an extra call to qualify and route you before reaching an Account Executive for a demo. This mostly benefits the vendor, not the buyer. As I optimize our sales process, I can see why. The main reason is that an AE’s time is precious when scaling. AEs can only handle a limited number of calls per day. From my experience doing sales for the past few years, doing more than four calls in a row is tiring, and doing too many may lead to burnout.
The Limitations of Traditional Sales Demos for prospects and sales teams in the early qualification stage
In the world of sales, time is the critical factor. Effective time allocation for AEs on qualified leads can boost conversion rates. However, the traditional sales demo model places significant hurdles on this path. There's a lot of steps between a prospect being interested in a demo and getting one:
Scheduling Hassles: Imagine a customer that's curious about a product over the weekend. They are motivated, want a demo, but scheduling one requires time - sometimes even a week. This delay can frustrate the prospect, adding an unneeded layer of anxiety to their buying journey.
Inadequate Lead Filtering: Traditional methods often fail to filter leads effectively. No one wants to fill out a long form before getting a demo, so there's a balance between what you can ask now vs. later. This means prospects who are a poor fit for your product, may still make it to the demo/discovery call leading to a drain on resources and time.
Qualification Challenges: Not every interested party is a qualified lead. Sales teams often spend precious time on unqualified leads, thereby wasting the potential of their best performers. This traditionally calls for a strong separation between Sales Development Representatives (SDRs) and AEs - SDRs manage the less nuanced aspects of sales, qualifying customers first, while AEs conduct the detailed, tailored demos.
Limited AEs: AEs take time to ramp up, knowing how to tailor demos to close prospects. They also can only do so many calls in a day. Assuming you have infinite leads coming in, this ultimately becomes the biggest bottleneck (besides support) since AEs need to be ramped up and trained, and need to physically take the call. If the call including prep time and cool down time takes an hour, an AE can do about 4 calls a day without burning out. This leads them to being able to manage only 80 opportunities per month.
A new approach to discovery
Imagine a user lands on your webpage and decides to request a demo. The first step should be friction-less. Asking for an email address is an effective method to capture potential leads, asking for the smallest amount of information, while simultaneously gathering information for enrichment to make our communication more personal.
Booking a time and qualifying questions
Once the user requests a demo, it's crucial not to bombard them with a lengthy form. A better approach is to show them a calendar to choose a convenient time. By doing this, you can create a sense of commitment, making them more likely to complete the subsequent steps. Big challenges are around incentivizing the buyer with enough personalized feedback that they feel the experience is tailored to them, that their answers do matter, and have an impact on what’s to come with the demo.
Asking 2-3 questions here that are most critical for filtering can be helpful in case they drop off, such as how many leads do you manage each month, or how many clients do you do reporting for? Ultimately we want to make them feel that we are giving them the best chance at success of solving their problem and as quickly as possible.
Data Enrichment
The magic happens when we begin to further enrich and use the data they provided us with. Enrichment can empower our LLM to make the questions feel more personalized, if it's able to understand the person's job title/role, company size, and responsibilities, we can map that to past calls of people with a similar role. If their email doesn't provide sufficient data, you can always prompt for additional info, such as their name, in a non-intrusive way.
Needing incentive to continue
At this stage, the user may feel more fatigued. The good news is we've confirmed the demo with all the relevant info needed. The bad news is we still haven't eliminated the need for a discovery call. To keep them engaged, we need to first make a promise/incentive. Something that encourages them to keep answering relevant questions with more detail to the extent that we need them to eliminate a discovery call.
The Voice Connection
I think this might be the hardest part. Having the user agree to allow their microphone access can be tricky, as it introduces a new user behavior. In general I lean towards avoiding new user behaviors because it just adds friction to adoption. In the near term, this interaction is quite foreign to users, but in the future (when AI is more prevalent) this will be easier as it becomes a much more common way to interact with computers. So in the mean time we need to figure out the incentive to get users to want to do this in the first place:
- The promise of a live demo with a salesperson could be one way to entice users. The problem with that is then it requires routing that knows that the salesperson is online. Approaching things in this way too soon may be a step too far for to get sales teams to want to adopt this product, since it isn't progressive and addititive to the way they do things currently.
- The promise of skipping two calls and just doing one is very enticing (I think this might be the best/better incentive)
- Finding ways to convey that their answers will get them closer to understanding "How do I as the buyer get the most out of this tool to help me achieve my goals and my problems?" or "What am I not aware of that I should be to get the most out of this part of the business/problem?"
Overall having a buyer speak about their needs allows us to have a richer understanding of the user's unique needs and paves the way for a more personalized demo using their own words. I need to be mindful to request consent, provide clear reasons for asking, and offer an incentive. Another worry I have is that their answers will be too short. Take this for example, the screen prompts:
SaaS: "how did you hear about us?"
Prospect: "Youtube"
Without the feedback of a sales person verbally saying "can you tell me more about how you heard about us on YouTube, for example were you searching for something in particular?" I think this is the hardest challenge of it all, because without this, the discovery call cannot be replaced.
SaaS: "how did you hear about us?"
Prospect: "YouTube"
SaaS: "Could you tell me a bit more context around how you stumbled upon the YouTube video?" - something just feels quite cold about this via text vs. a conversation. Which makes me think that you'd have to use some visual hacks to give live feedback to the user to help increase the length of their response but not too much so as to taint it or cut it short (like a greenlight when it's enough). There's also something a bit strange when it comes to speaking with a screen but getting a response as text. Which then makes me feel if we have to go as far as the LLM NPC approach where it is more interactive with what you're saying to get the desired result. Then latency starts to matter and a whole other can of worms.
Overall, a key part is making sure the person feels heard and that this experience is personalized. This makes them feel like what they're sharing is actually being processed and used to improve the situation, so I think it will be doable with text as long as the responses contain information that feels unique to them (this is where leveraging data from enrichment is critical).
This is the main area I would focus and iterate on, although worst case the other portions of simply booking the demo and enriching the lead to get more relevant information to properly route them is a viable small step change to the process.
Routing
Once the user shares their thoughts with voice, it feels like a discovery call. The challenge here is two fold, one in the routing – ensuring that the right salesperson is available to handle the interaction in real-time if necessary. Otherwise, saving the recording and having them ready for the AE with a summarized call works well. The second is making sure the LLM understands what the prospect is saying. Most calls feel quite similar, it's the one's that stray from normal that typically aren't the best fit, but will still be problematic to make it feel seamless for the buyer. The end result of an interaction like that will likely require a review by a SDR to conclude if a further demo would be helpful, and what next action should be taken.
[Screenshot here]
Perfecting the User Journey
By the end of this process, we want the user to feel heard and understood. Inform them that their responses are being evaluated to match them with customers we've helped in the past. In the case of routing to a pre-recorded demo, this will help give a sense of credibility and might nudge them towards a purchase that is completely hands-free.
[Screenshot here]
Balancing effort in the qualification and discovery process
It requires a lot of effort to do this, so it’s probably best taken step by step to see how things convert, but due to how involved this is you can see why every company wouldn’t go beyond just a form/calendar link. It’s just too much effort. Also to continually improve the system takes away too much from the team's core product. So it’s likely that there will be a few players that do this for everyone else, rather than each company on their own. Overall it's interesting to think about since this fundamentally is trying to solve the "Intake" portion of a sales process. The intake process can be found in numerous industries that require personalization at scale, such as Health, Education, and many others beyond just SaaS. The opportunities are endless and it's only possible now because AI has reached a certain level of reasoning and is only going to get better.