Will AI result in net gain or loss for GTM teams?

Every major technology humans have invented has produced efficiency gains along with undesirable side effects. With every new technology revolution, the adoption cycle gets shorter. The elapsed time between society reaping the benefits and paying the cost to remediate the undesirable side effects—let’s call this the “payback gap”—also gets shorter. Here are a few examples to illustrate the point.

The shortening of the payback gap

The agricultural revolution drastically improved society’s efficiency in food production, which freed people to pursue work outside of farming, such as the arts, politics, and sciences. Scaling agriculture gave us the agricultural industrial complex, which contributes to everything from destruction of ecosystems and climate warming to unhealthy diets dominated by processed food. We are now paying for these side effects in the form of more frequent and severe natural disasters, plus higher healthcare costs. The payback gap is on the order of millennia.

The industrial revolution drastically improved society's ability to make stuff cheaper and faster, elevating our standards of living in the process. It also gave us undesirable side effects that include unhealthy air, microplastics in our water, and a warming planet that is turning into an existential crisis. The payback gap has shortened to centuries.

The internet made it possible to share information and interact at unprecedented speed. The internet also gave us social media and its negative impact on our youth and politics. The internet has proven to be as efficient at distributing information as it is at misinformation for people with bad intentions or driven by greed. The payback gap has shortened to just a few decades.

Mobile put the internet on steroids and, while it has made us even more connected, at the same time it has made us more isolated and has degraded human interactions. The payback gap is now less than a decade.

With AI, the payback gap is now less than five years.

How the payback gap is relevant to RevOps

When the payback period is long, businesses can focus on reaping the benefits and kick the payback cost down the road for future generations to pay, which makes new technology artificially cheap and accelerates its adoption.

What may be different now is that five years is short enough to be within many businesses’ planning horizons, so the payback cost needs to be considered today.?

So what does this have to do with RevOps? Well, the potential benefit of AI has been written about ad nauseam, so I won’t spend more time parroting that gospel. Plenty of people have also written about the apocalyptic potential of AI. While that makes good musing, it’s not very relevant to RevOps specifically. Instead, I want to discuss the realistic negative side effects of AI on RevOps and what payback may look like.

Data volume explosion

If you think we have Big Data now, AI says: “hold my beer.” One of the biggest impacts AI is already having is the generation of data at unprecedented speed and scale. AI is not just super-efficient at generating new data, but also at summarizing existing data, extracting structured data from unstructured data, and creating permutations of data. AI-generated data has already become so prevalent that subsequent generations of AI may be subject to “poisoning” by AI-generated training data. Here is how Scientific American explained this: https://www.scientificamerican.com/article/ai-generated-data-can-poison-future-ai-models/

Most companies already struggle with how to make vast amounts of data usable to its employees and technologies so they can extract value from the data. Exponential data growth makes these efforts even more difficult and expensive. For example, in addition to human visitors and web crawlers, your website traffic will likely drastically increase in the next few years once AI agents gain widespread adoption. While RevOps teams are busy creating AI agents to gather more data to power their GTM team’s efficiency and effectiveness, these AI agents are simultaneously generating more data for the RevOps teams that own the website that the agent is interacting with. So for every piece of data gathered off the page by an agent, the website’s visitor log data also increases correspondingly. For every action, there is a reaction. Ha! You didn’t know that Newtonian physics applies to GTM data, did you? :)

Signal-to-noise ratio degradation

The sheer increase in data volume alone—and we are talking orders of magnitude here—already presents a significant technical challenge in terms of storage, transfer, access, and processing. The fact that much of this data is generated by AI automation creates a bigger, undesirable side effect of signal-to-noise ratio degradation. In the last 15 years, all GTM disciplines have strived to be more science than art. GTM teams use ever-increasing quantities of data to target the most likely buyers, identify the most profitable customers, figure out what to upsell, orchestrate the best next step in the buyer’s journey, and predict churn risk. All this data-driven GTM science is based mostly on human behavior data and our digital footprints. Even pre-AI, during a few parts of the customer’s lifecycle, machine-generated data has already made it more difficult for the GTM team to separate the signal (human-generated data) from the noise (machine-generated data). Two areas that marketers know well are ad-clicks from click farms and email clicks from corporate firewalls.

AI-based automation will increase machine-generated data, which in turn reduces the signal-to-noise ratio by orders of magnitude, making it exponentially more difficult to find meaningful signals that can power data-driven GTM strategies. If you can’t separate out the signals from the noise, then instead of being data-driven, you’re simply noise-driven. For example, where’s the meaning in your A/B test if most of the visitors to your website are bots that click on every button regardless what color that button is?

What does payback look like?

Faced with these AI-induced data challenges, I doubt that GTM pros and RevOps data geeks will just give up on data-driven methods and revert back to brand-driven methods. We will try to overcome these challenges, which means:

• We will invest in data science skills and data technologies to separate out the signals from the noise and handle ever larger and more complex datasets.
• New technologies will need to be created to separate AI-generated engagements and signals from actual human activities.
• Just like what has happened in the cybersecurity industry, this will evolve into an endless technology one-upmanship. In one corner, you have AI that tries to behave more like humans to avoid detection and blocking. In the other corner, you have technologies that try to sniff out and potentially block non-human activities. If you ever dealt with CAPTCHA or looked into what it takes to scrape LinkedIn data, you know exactly how this cat-and-mouse dynamic works and escalates.

All this takes new money and resources. So while on one hand we are saving money using AI to accelerate research, write personalized content, and even qualify prospects, on the other hand we are hiring data experts and buying more technologies to overcome these AI-induced data challenges. AI may produce no, or even net-negative, efficiency gains for the GTM team because the payback period for new technologies is now near zero.

I’m sure some of you disagree with my outlook here. Would love to hear your feedback.