More than 90% of commercial organizations are experimenting with AI. Most are not yet seeing meaningful productivity gains. In Bain’s survey of 1,125 sales and marketing leaders across 18 industries and 40 countries, AI experimentation is now the norm. Value realization remains uneven.

Some companies are focused on single use cases that are too narrow; others layer AI onto broken processes that deliver only micro productivity.

Winners—companies that are pulling ahead in the top quartile of revenue growth for their sector and region and exceeding their margin targets—are achieving twice as much AI-attributed revenue growth and 1.8x greater cost efficiency than laggards (see Figure 1).

These successful companies are not just experimenting more; they take a different approach to redesigning and embedding AI into how work gets done.

The question for executives today is not if they should invest in AI but rather whether their AI approach is set up to change how they go to market. Is it tied to real value? Set up for scale? Is there clear commercial ownership?

Figure 1

Today, companies tend to fall into one of three stages of maturity.

• Exploring: deploying discrete tools, such as email templates, call summaries, or customer relationship management (CRM) automation.
• Enhancing: building a constellation of use cases focused on bottlenecks, such as marketing content creation, automated account research, or sales support agent-assist tools.
• Transforming: reinventing high-value, cross-functional workflows, such as end-to-end marketing and collateral creation, outbound prospecting, and sales enablement.

While many companies are moving beyond piloting AI, few have used it to automate sales and marketing workflows end to end.

How winners deploy AI differently to achieve measurable goals

In our experience working with companies across industries, AI impact accelerates only when organizations change how the work is done.

Winners treat AI as a lever to reengineer how they go to market. That shift rests on four building blocks: a redesigned workflow, a solid data foundation, dedicated ownership, and a focus on leading and lagging key performance indicators (KPIs).

Redesign critical workflows end to end. A sales workflow is the path from commercial trigger to measurable outcome. It defines who does what, which systems support them, how decisions are made, and how handoffs occur.

AI that is simply tech layered onto existing processes without a deeper end-to-end rethinking of those processes will fail to provide the step-change value for which many companies have invested.

Winners identify high-value, cross-functional workflows—such as sales enablement, outbound prospecting, pricing and discounting, lead-to-deal conversion, or marketing content creation—and redesign them with AI from trigger to outcome.

That’s what one financial operations platform did when redesigning its outbound engine. Historically, sales development reps constructed lists manually, digging through LinkedIn, drafting cold emails from scratch, and updating CRM fields by hand. All accounts were treated similarly regardless of value. Significant human effort went toward prospects that rarely converted.

Today, the company has rebuilt its early workflow around AI. For lower-tier accounts, automated pipelines identify targets, generate personalized sequences, and log activity in the CRM. Human reps engage only when signals merit attention. Sales development reps focus on high-value accounts. They still customize outreach to these top-tier accounts, but now they start from AI-generated insights pulled from emails, transcripts, and past interactions.

The transformation changes not only who does the work but how it gets done. More human capacity now flows to complex, high-value accounts while the long tail still gets timely, tailored coverage.

Build pragmatic fit-for-purpose data foundations. Technology and data maturity are fundamental to scaling AI, but 60% of respondents stated that their data foundation was not robust, their technology was not ready, or both. Thirty-six percent of laggards reported having the right technology and tools in place but that their data was either incomplete or low quality. Slightly fewer winners, 29%, are in the same situation.

Launching AI with data as is risks inaccuracy. On the other hand, perfectly preparing all data before launch is expensive and time consuming. A middle path that cleans the minimum viable data set needed to run one workflow and then expand as use cases grow is the more pragmatic path. One by one, these uses create data foundations that enable AI to scale into real value.

For go-to-market systems, cleaning up select CRM data is often the right place to start, using AI to accelerate data cleanup and enrichment.

Recently, a global logistics company faced a familiar problem: Unlinked customer data spread across multiple CRM and enterprise resource planning (ERP) systems, filled with duplicate records and inconsistent hierarchies, offering limited visibility into sales by customer. Rather than attempt a full overhaul, the company focused on the elements that matter most for its near-term sales priorities and that have the greatest potential for revenue gains. It matched and cleaned millions of transaction records, reconciled customer hierarchies, and clarified what had been sold across product lines. It then enriched the data by pulling in prospect information and built an account scoring model to prioritize cross-selling opportunities and new logos.

The company did not treat every product or industry the same. It went deep on two core product categories and key sectors, such as retail and manufacturing, while taking a lighter approach to lower-priority areas. This targeted cleanup created a reliable foundation for AI-driven targeting and lead allocation. The cleanup was worth the effort: Qualification rates more than doubled, first-meeting rates doubled, and reps saved time they otherwise would have spent figuring out who to target. Additionally, management can now make a better business case for cleaning up the rest of their data.

Treat AI like a product with a clear owner and dedicated teams. Survey results highlight the primary obstacles organizations face in leveraging AI for growth—they include limited cross-functional input (38%), inflated leadership expectations (35%), no clear ownership (29%), and insufficient training and adoption support (27%).

Winners address these barriers by treating AI solutions like products, not side projects.

They are 40% more likely to have cross-functional AI teams with dedicated and persistent product management and ownership. In many cases, that includes a product manager, data engineers, subject matter experts, and two to three frontline users. When building new technology, teams also require a technical lead, AI/machine learning (ML) engineers, and software engineers. This work is not an extra 10% added to existing roles; it is a dedicated mandate.

They pair that structure with leadership that actively engages by learning, experimenting, and modeling usage rather than simply sponsoring from a distance.

Most importantly, they assign clear commercial ownership. Fifteen percent of laggards report shared or undefined ownership of AI initiatives compared with just 1% of winners. High-performing companies in our survey most often place responsibility with a clearly accountable commercial or technology leader, ensuring that outcomes are owned end to end.

Track leading and lagging KPIs, not just adoption and use. Leading companies avoid the common mistake of scaling pilots based on good use numbers only to be surprised when real business value fails to follow up. Instead, they decide to scale pilots based on proven leading indicators of value such as a larger, high-quality pipeline or better win rates.

The right KPIs vary by what a company is working on.

If building agentic lead discovery and prioritization based on intent to purchase:

• leading KPIs include email open rates, conversion rates of marketing or sales qualified leads, and lead response times;
• lagging KPIs include incremental revenue from the prioritized approach (vs. baseline), costs per qualified sales lead, and revenue per lead.

If generating demand by optimizing campaign mix and channel allocation:

• leading KPIs include campaign engagement rates, A/B test outcomes, and content click-through rates;
• lagging KPIs include customer acquisition costs, campaign and channel returns on investment, and contributions from marketing campaigns.

If supporting sellers with relevant content-augmented guided selling:

• leading KPIs include the speed of deal progression across the funnel, the number of concurrent deals managed by reps, and the number of products pitched per rep;
• lagging KPIs include tracking win rate improvements for reps using tools (vs. those not using them) and incremental sales per rep.

Where AI is gaining traction in B2B sales

Many of the most promising AI use cases can boost both revenue and efficiency when embedded in workflows. Here are some of the results we’ve seen.

• Targeting: 50% more high-quality leads.
• Marketing and collateral generation: 40% more top-of-funnel leads, with 40% to 70% reductions in human time.
• Personalized, automated lead generation: boosts leads about 20% and revenue approximately 1%.
• Customer support and success agent tools: 25% to 35% efficiency improvement on average.
• Pricing (contract optimization and automated pricing engines): 4% to 8% additional revenue growth.
• Knowledge assistants (such as smart search): 25% to 35% increase in effective selling time.
• Contract review: 1% to 3% increase in revenue.
• Request for proposal (RFP) response generation: 50% of RFP tasks automated.
• Data preparation: 80% of data enrichment and qualification automated.
• Training and coaching: 20% to 30% time saved.

Emerging applications show promising results that include sales enablement, lead-to-conversion automation, marketing and sales collateral, contract review, RFP response generation, data preparation (such as account intelligence or automated data cleansing), and training and coaching.

Three high-yield questions for executives

Three questions can help CEOs, chief revenue officers, and boards cut through the noise as they work to build a strong commercial engine:

• What is a single workflow in which a 10% improvement would materially change performance this quarter? If you can’t name it, you’re not ready to scale AI.
• Do we have one trusted view of the business outcomes for updating that workflow? To show demonstrable returns, the focus must be on pipeline, revenue, and cost, not adoption and use.
• Who owns outcomes end to end, and what is our release rhythm? If ownership is shared and updates are ad hoc, pilots will never compound into transformation.

The advantage will not go to those with the most pilots; it will go to those that redesign their commercial workflows around AI and hold themselves accountable for measurable outcomes.