The state of sales tech

Of all the categories in the purview of the Revenue Brew journalist, none quite divides opinion among readers, leaders, and industry insiders like the business of sales tech. It is perhaps the noisiest bucket that our work goes into, and the one (honorable mention to “strategy and leadership”) that is changing the fastest. Our readers want to know where that change is happening, how it applies to their organization, and where best to deploy capital and resources to harness its capabilities. In short, AI is reshaping the future of sales and go-to-market tactics, and it’s our job to capture that change.

AI tools offer great rewards, but they often pose a sizable risk. From generating phantom work, to rogue agents deleting your company’s database (yes, that actually happened) there is a lot to be aware of.

Revenue Brew conducted a survey looking at how organizations are using tech platforms, tools, and AI today, where implementations are succeeding (or not), and how teams are preparing for the next stage of progression. We have plucked out some of the key responses, and along with the testimony of several experts in our ecosystem, we will do our damnedest to make sense of it for you. This report offers no grand conclusions, but simply places markers in the ground for where we are now and pointers to where we may be heading.

According to a Revenue Brew poll, 39% percent of respondents cited “confusing or overly complex tools” as a reason for AI tool implementation failure, while 36% pointed toward “bad or incomplete data.”

This seems like as good a place as any to start. How do you—as a seller, marketer, customer success agent, or revenue leader of any kind—best mitigate risk from the ground up and avoid ending up in this category?

Data and the human touch

AI implementation is a balance. Yes, it can create efficiencies that previously seemed impossible, but it can also undo “40 years of IT security” in moments; that’s according to Chris Willis, chief design officer at AI software firm Domo.

“I think OpenClaw is a great example of [how it] got very exciting and very scary, very fast,” Willis said, referring to the open-source autonomous AI agent that grabbed headlines in February.

He suggested that agentic automation without real visibility is a risky strategy. “You have to anchor it to governance, because without it, you’re really just creating a huge amount of potential for chaos.”

Willis said human rationality will be a protective measure when AI makes mistakes or goes rogue.

“The biggest wins happen when our customers are pairing AI with real-time governed data, and human judgment is in the loop to interpret and act,” Willis said. “Judgment is the missing piece; models generate [and] humans pass judgment on those things.”

For this reason, Willis warned against companies slimming down workforces in favor of AI (using Klarna as an example), as less human governance can often mean worse outcomes, even if efficiency is increased.

Find effective use cases

Determining the reasoning for your tech deployment in the first instance is key and might help avoid AI “workslop,” another problem driving leaders in circles. Emily Nash-Walker, senior director of product strategy at workflow automation firm Tungsten Automation, says the phenomenon—where AI produces work that lacks substance—is the biggest risk to AI deployment.

“I had to tell my team: If you do not review the outcomes of the engines that you’re using, I will remove you,” Nash-Walker said. “What I was starting to get was these outputs, and there were hallucinations.”

Nash-Walker echoed Willis’s emphasis on the importance of human intentionality and said an emerging hiring trend on her teams is to look for individuals with liberal arts, or creative, backgrounds, rather than technical expertise. She says these individuals are often more adept at prompting AI in more human ways.

“That human in the loop is so important,” Nash-Walker said. “The other risk, in my view, is the tendency for companies to over-rely on automation.”

Combatting the slop

Nash-Walker’s comments on AI workslop track as a key theme for us. It has emerged as a potentially major downside of integrating large language models into sales motions. Revenue Brew data shows that 70% of respondents have integrated LLM tools, but there is still a sustained push from some of the revenue leaders we’ve spoken with to ensure that the undeniably human element of the sales process isn’t obscured.

Run your own race

If there’s one thing our work has told us over the last few months, it’s that you’re probably best off not trying to keep up with the Joneses. You might want to jump right in, but take the time to find your footing. Dawn McGrath, marketing director at oil and lubricants company Keller-Heartt Oil, said her team is just dipping its toes in the AI waters, but so far they are liking the temperature.

“What we’re doing is using the people who are really excited. We’re doing demo videos… specific to where we’re using these new technologies, and then showing people who are maybe a little bit more adverse [that] this is to help you,” said McGrath.

Her team started to use Anthropic’s Claude to ideate and produce visuals. McGrath said that part of the reason her team chose Claude was because of Anthropic’s messaging around data safety—a step that she said mitigated risk. Easing into this allowed her team to create standards around what data is acceptable to put into these tools.

So there’s plenty of food for thought if you’re looking to avoid some of the common missteps in tech stack deployment. But let’s say you’re up and running and hitting some inevitable roadblocks. Then what?

Pain management

Tech stack pain is part of the territory for revenue organizations in the age of AI hyperscalers and agentic AI overload. With a plethora of tools available and AI integration at an all-time high, issues are bound to arise. No revenue organization is immune but there are ways to minimize the impact: Identifying the pain points early is one.

According to Revenue Brew’s survey on the state of sales tech, 23% of respondents said if they were able to turn back time and onboard their tech tools again, they would spend more time testing integrations before any rollout. Additionally, 37% said they would also simplify their workflows.

The perils of misalignment

The proliferation of SaaS tools means an entire suite of agentic verticals now exists to address hyperspecific issues, but getting different tools to mesh seamlessly with each other is another challenge altogether.

Miller says a problem that arises is not understanding how these platforms connect with each other and serve the overall goals of a business. She’s hitting on something that crops up regularly in our coverage of the tech stack space: Tools can sometimes create friction rather than working with existing functions. This is something that Irina Soriano, vice president of strategic enablement services at sales engagement platform Seismic sees, especially when AI tools are added into the mix. “If things are positioned as a standalone, particularly for the sales teams, that is then starting to create some of the scenarios where we have the change fatigue, the overwhelm.”

Soriano says it’s a different story when there is alignment with existing workflows.

The right balance

Integration is a necessity, but avoiding bottlenecks is hard when the technology is moving at such a rapid pace. Abhijit Mitra, CEO of revenue intelligence platform Outreach, says what is needed right now are go-to-market architects.

“They are essentially architecting the end-to-end business processes, the data flow, and as part of that, the technology that will be needed to solve the use cases,” said Mitra.

He said the company recently posted some job posts for goto-market architects, as there are specific use cases it needs to figure out.

“How does Outreach work with a Snowflake? How does Outreach work with my marketing tools? How does Outreach work with Salesforce?” he said.

Mitra said customers need to figure out how the different tools under a business umbrella work “homogeneously” with each other to orchestrate efficient business processes. Part of this formula will be a delicate balance between human and AI to not only ensure tools are being fed accurate information to produce accurate outcomes, but also to provide direction in best practices.

[Data highlight: Revenue Brew data shows 52% said onboarding of new technology didn’t succeed because of insufficient training.]

The secret sauce

Coaching undoubtedly plays a crucial role in the successful adoption of new tools. Most survey respondents (52%) said onboarding didn’t succeed because of insufficient training. As we’ve reported over the past year, achieving some form of consensus on the tools chosen, and why they’ve been selected, is a critical first step for revenue leaders. Without this the training process can be more difficult than it needs to be. A breakdown at this stage can lead to bigger problems later on.

We get it: Taking a shine to a SaaS platform or an AI tool is one thing, and successfully implementing it is another. Tech stacks can give revenue teams an edge, but deployment requires thought and precision to get the best results—not an easy feat. In a worst-case scenario, businesses splash the cash to onboard revenue tech that is ultimately ignored by confused employees or causes longer-term problems.

Revenue teams are feeling, well, meh, about AI implementations on the whole. Across most job functions—from sales engagement tools (52%) to automated data entry (56%)—companies are describing their AI tech deployment as only “somewhat successful.”

So how do organizations overcome these challenges before they become insurmountable? One way of doing it is getting your team on the same page. Thinking about those that use the technology as a potential solution, not a problem, might be the best way forward at this stage.

Keep it simple, stupid

If it has all become way too complicated, you probably need to take a step back. Seismic’s Irina Soriano said inflated tech stacks can often be a barrier to AI adoption because sellers are constantly having to move between technologies.

“If I’m a seller on the receiving end, I now have to use 15 point solutions to do my job. That means I have to log in 15 times,” Soriano said. “There’s likely redundant workflows I have to go through because some of those tools are overlapping.”

Soriano advises selecting additions to an existing tech stack with centralization in mind, so that data works in orchestration across various tools. In short, helping your sellers will help everyone in the end.

The right tool for the job

On that note, the level of orchestration you’ll need will vary depending on your particular requirements: Some organizations will use a CRM from one provider, integrate it with an analytics system from another, and a marketing automation tool from yet another. Is this burning money, causing complicated workflows, and leading to the dreaded siloed data? For some businesses, yes. For others, it’s the magic potion: It depends on the necessities of your organization.

Sell your sellers on AI

Sure it’s better to do this in the beginning, but any time you can get your team singing from the same hymn sheet, it’s a win. Edward Gorbis is a sales lead at Amazon Web Services and writer of the Morning Sales B2B newsletter. He said where he has seen AI successfully adopted, teams worked backward to identify where tools can be useful. Moreover, Gorbis explained that nothing is more important to getting buy-in than proof of concept.

“What are you solving for? What are you trying to improve? Because if you’re just giving people a tool and telling them to go use it, some people will figure it out, and other people won’t, and other people get frustrated with it,” Gorbis said.

In instances where there is dissonance between the expected use case of an AI tool, the results revenue leaders expect, and how sellers use those tools, Gorbis said that there is often a “frustration loop.”

“If you just release the genie, most people don’t know how to tame the genie or use the genie the right way. You need to give them instruction,” Gorbis said. “If you’re the leader, and you’re not demonstrating that there’s high ROI with these tools or demystifying it for your individual reps, then you’re going to struggle with team adoption,” Gorbis said.

Adoption is pulled, not pushed

Jamie Cleghorn, senior partner at Bain, said nothing is more persuasive than results—especially with sellers who benefit directly from increased performance.

“Adoption works when it’s pulled, not pushed. So the single most effective way to drive adoption in a revenue organization is to demonstrate proof points with small groups and then hold those up as hero examples,” Cleghorn said.

It’s clear there is no exact formula for designing a winning tech stack. It varies by organization, by go-to-market strategy, and by leadership preference. There are, however, some broad tactics emerging around tech stack deployment that could deliver positive outcomes for global business leaders.

Best practices

Firstly, most CROs and revenue enablement leaders will likely need to work with the technology they have at first and build upon it in time (throwing out the baby with the bathwater won’t be an option for all, especially the biggest companies).

Secondly, the tech integrations that companies curate will depend on the specific needs of the business. The key is determining those needs early on, avoiding complexity, and ensuring everyone is on the same page.

Finally, it has become clear that no modern-day tech stack can reach its full potential without some form of AI integration. This could come in the form of automated workflows that are more deterministic in nature, or even fully-fledged agentic systems. The winning strategy could be a mixture of both, but silos are not an option (read more about our multiplayer future). Ensuring the applications work seamlessly with existing data and other tools is also a top priority.

Follow the data

Our data shows 36% of those surveyed said new tech or AI tools didn’t succeed in their revenue organizations because of bad or incomplete data (known more colloquially in the industry as the dreaded “garbage in, garbage out”). So-called clean data is the foundation for all accurate insight, and without it, the foundations of your business are faulty.

Additionally, 34% of those surveyed said new tech or AI tools didn’t succeed in their business because of poor integrations. ZoomInfo’s CRO James Roth says within the myriad of tools that offer to solve specific problems, there isn’t much success without integration into an overall data repository.

“One of the things that we focus on aggressively is anything that we put into the tech stack has to be set on the right data foundation,” said Roth.

No more throwing spaghetti at the wall

The key to successful integration could be as simple as CROs being strategic rather than reactive when onboarding new AI tools. At a time where revenue leaders are inundated with new platforms, this will be harder than it sounds. The Revenue Brew team has more than once referred to this as the “spaghetti at the wall” syndrome (seeing what sticks isn’t exactly a scientific methodology).

[Data highlight: Revenue Brew data shows 25% of respondents say they are reactive in adopting new tech in response to pressure or trends.]

While many of those adopting new tech are reactive (25%), the majority of respondents said they are strategic in adopting AI and new tech to align with clear needs (47%).

“Make sure that this is worth a product or even a company to have as a vendor,” said Jason Ambrose, CEO at revenue intelligence and operations platform People.ai. “How far are you on the AI maturity curve? Can you evaluate these things to be able to say: Is this something that we really need to pay somebody to go do, or can we figure this out on our own?”

Top-down vs. bottom-up

Determining the analytical maturity of your business is not easy, but it is a worthwhile endeavor if time allows. As Ambrose suggests, an audit of this kind can allow revenue leaders to gauge whether deploying capital is really necessary (ROI concerns are rife in this tricky economic environment).

People.ai has transitioned from a “top-down” approach—where sellers would “create this sameness across selling experiences [and] selling behavior” with the “same marketing email to 20,000 people,”—to a flipped script where AI is helping create original content specific to each prospect.

“That very granular bottom-up view is a totally different way to approach your revenue stack,” said Ambrose.

Ambrose said changes in the company’s tech stack resulted in email click-through rates jumping from 2%–3% to over 50%.

“We’re seeing a big lift in organic traffic as a result of that,” he said. “We’re focused on the top of funnel with these things.”

The balance between human and AI capabilities is crucial to a winning tech stack. Sellers might not always see AI as a silver bullet (probably best not to), but as a way to make their day-to-day functions stickier with potential customers. As with any new technology, it doesn’t always work out as planned. At ZoomInfo, Roth said his team learned AI best practices the hard way, when it implemented an AI chatbot that started creating duplicate agents that were a “less good version with a less good prompt.”

“If you don’t have a strong process for what AI you are using, you end up with 4,000 similar agents that are solving for typical account management use cases,” said Roth.

Roth said companies need to make sure they’re building the best version of that agent.

“I don’t necessarily need account managers to be great agent builders,” he said. “I need them to be a great account manager that is leveraging agents to do the things that, frankly, the agent can do better.”

Final thought: measure twice, cut once

While agentic systems might be the glitziest AI use case of the moment, it’s still important for revenue leaders to take a 20,000-foot view of their tools to know when to get rid of outdated functions. From there they can build, iterate, and find the perfect harmony for their business and prospects.

“I don’t think there is life without a technology stack,” said Riju Vashisht, chief growth officer at technology solutions company Genpact. “The days of relationship based showing up to a client to drive sales are long over, because every buyer in the market is looking for, ‘How do you create value? How do you add value?’ It is difficult for any salesperson to be effective in driving value for their client if they don’t look at all the possibilities.”