Company Profile Info

• Company name: ScaleX, Inc.
• Years in business: Three.
• Location(s:) Virtual.
• #empolyees: Five.
• Principal management: Chad Burmeister, CEO, Rich Blakeman, CRO.
• Revenue % growth over last year: 85%.
• Primary business successes to date: 100+ clients
• Primary sales model (direct, indirect. telemarketing, etc):Direct and indirect model (15 resellers growing to 50 by June 20, 2021)
• Average size of sale. $6,000 to $50,000
• Primary marketing methods.  ScaleX uses our own outreach technologies – data + digital (email, social, paid-ads, warm introductions) and dials (using agent-assisted dials).

Contact Info

• Website:  www.scalex.ai 
• Email : [email protected]
• Twitter: chad_burmeister

Introduction

(OK, yes, we admit it. The reason that title sounds familiar is that I’m paraphrasing Kyle Reese, who’s who’s been sent back in time to stop Ahhnald from killing Sarah Connor and preventing the end of the world, as well as anymore Terminator sequels. As we all know, he failed.)

AI is a hot, hot, hot topic these days. Everywhere you look companies are boasting about the quality and quantity of AI in their products. as opposed to the Brand X used by the competition.

First, a quick primer on the entire topic. Broadly defined, there are two types of AI, “strong AI” and “weak AI.” Strong AI does not exist. No robot, industrial machine, or computer program exists that is self aware and capable of independent moral decisions. These are the base level qualifications for strong AI. In my novel, Rule-Set: A Novel of a Quantum Future, I discuss the AI winter of the late 80s and early 90s and why HAL 9000, scheduled for an appearance in 2001, was a no show.

Still, “weak AI” is getting a lot done today. Weak AI is based on neural networks, structures in code that conceptually resemble how the human brain assembles data into manipulated patterns. The network’s “neurons” are assembled into layers and stacked, with each layer able to pass the results of its computations to the layer “above” it.

Over time, this net becomes increasingly powerful as the amount of data it can access increases. That’s being provided by Big Data and SaaS, and is one of the principal reasons for the rapid acceleration of the powerful but highly constrained AI being incorporated into software today. Another factor is that as their complexity grows, the net is able to use the techniques it’s learned from its previous training (such as recognizing a cat in a photo and repurposing them to learn how spot a dog.)

Of course, there’s a lot more that “weak AI” can accomplish. If fact, many people believe the term “weak” is misleading and that phrases such as “single purpose” are accurate. We’ll let you make up your own mind about that, but we decided to spend some time with Chad Burmeister, CEO of startup ScaleX. ScaleX’s business model is building virtual human assistants (VHAs) powered by deep learning AIs to increase close rates and boost sales. We want to know how this all works.

Softletter: Chad, what is a VHA in a sales/prospecting context?

Chad: It’s pretty much as it sounds. Our VHAs automate the tasks of contacting prospects in a targeted database. In our case, our product is optimized to run against LinkedIn. It’s one of the largest, most popular and well known B2B databases.

The tasks a ScaleX VHA automates are context sensitive and include:

For email campaigns:

• Checking the calendar of the sales rep for meeting openings.
• Scheduling meetings.
• Tracking when a prospect is out of office and pausing communications until the prospect is back.

For social selling:

• Sending connection request to new connections.
• Executing sequences to new or existing connections.
• Following prospects on LinkedIn.
• Liking a prospect’s tweet.
• Sending multiple InMails per day. (An InMail is an internal communication a LinkedIn member can send to other members).
• Viewing a prospect’s profile.

There’s nothing here that a human can’t do.

You’re right. But we’ve found out something very interesting about the sales process. Over time, humans drift away from what works. One reason is very understandable; much of the work involved in prospecting is tedious and boring. People will start to do almost anything to break the monotony.

Another reason is more subtle. We all carry our own mythology and we impose it on our work habits and working processes. For example, the ScaleX VHA may determine the best time to send an InMail is X period of time, but the human will insist on doing at Y, perhaps because it’s more convenient for them or in the belief the AI is wrong. But it normally isn’t. We’re constantly testing conversion and response rates via hundreds of thousand of AB tests and the AI almost always wins. Of course, that’s all it has to do. Human’s have a wider range of interests.

Finally, there’s the issue of time. The common thread I constantly focus on is that AI-backed systems collapse hundreds of hours of analysis into a few minutes. Let me provide you with a recent case study of a sales conversion program we ran for a client. The goal of this campaign was to ultimately schedule phone calls with new prospects.

We started with a dataset of 1492 contacts extracted from Zoom Info, a list broker service many people have heard of. We set up a VHA to:

• Locate likely prospects in the client’s LinkedIn connection base, then identify paths to reach them. The pathway could be membership in a group, a first, second, or third connections, requests to connect, email address and so on. The system found 8681 likely contacts. After further analysis, this number was reduced to 5418.
• Each contact was then sent an email via ScaleX. This email requested that the contact provide an introduction to a client. We followed this strategy in this case because an introduction to a prospect from someone they already know is a powerful way of opening a closed door.
• The mailing generated 2623 opens from potential introducers. This is a 50% open rate; the norm is 18%.
• From the response to request to make an introduction 115 introductions were generated about 7%, with another 106, 6%, listed as “pending.”
• The final number of “in person” meeting schedules was 40, 2.7%.

Those are excellent numbers for this type of campaign. It took 10 days to go live. We first tested it via a pilot program that ran for 90 days to 750 prospects. The program was administered by two reps who adjusted the parameters of the VHA based on feedback and results.  The cost was $5K. We ran an analysis of what this would cost if done via manual systems and the number came at around $180K.

The live results I provided you were generated within several days. Again, if you attempted to execute this type of program manually, it would take weeks. This is why AI and sales are merging. Instead of assigning a single salesrep to a few major accounts, a single VHA can automate the sales process cycle to the point where many accounts can be run by a single salesperson. The gains in productivity are remarkable.

Who generated the emails?

Humans. However, we’re moving to an AI for future campaigns. The technology for automatically generating basic copy is rapidly improving.

Softletter has used these programs. While you’re right that they’ve improved, my take on them is that the prose they create is lifeless and sounds robotic. I suggest you consider using resources that “punch up” the copy in the appropriate places.

I’ll keep that in mind.

Did the emails that you sent to the LinkedIn contacts violate CAN SPAM?

No. These were first and second level connections. I have a previous relationship with these people. The pieces are compliant.

Since the primary dataset you run against is LinkedIn, what are some issues you should be aware of?

We recommend you have at least 1K connections on LinkedIn. And your projects will be far more effective if your network knows who you are. In other words, if you haven’t connected with your connections before, start before using AI for lead generation.