Episode 4
How to Scale AI in B2B Marketing Without Losing Control with Aby Varma
Aby Varma
He is the founder of Spark Novus, an agency that drives marketing transformation through strategic and responsible AI adoption. He also serves as CMO at Shapiro Negotiations and founded the Marketing AI Pulse Community, which has grown to nearly 600 members focused on learning, connecting, and growing in AI marketing.
How AI Is Actually Changing B2B Marketing Operations With Aby Varma
Introduction
Most companies adopting AI in marketing start in the wrong place. They start with the tool.
Aby Varma is the founder of Spark Novus, an agency focused on strategic AI adoption for marketing leaders, and he also serves as CMO at Shapiro Negotiations. He founded the Marketing AI Pulse Community a year ago with 30 members, which has since grown to nearly 600. On this episode of Ops in Motion, he joined Shiv and Kawal to talk through where AI is genuinely changing marketing operations, and where teams are getting the adoption process backwards.
Who should read this: marketing and RevOps leaders evaluating AI tools, vendors, or use cases for their team.
Key takeaways
- Starting with the technology instead of the business goal is the most common mistake in AI adoption. Tie every AI initiative to something that actually moves the business first.
- AI-driven personalization is only as good as the data behind it. A segment of one is possible now, but only with clean, enriched CRM data feeding it.
- Vendor evaluation needs to go past the pitch. Pilot before you scale, check how the vendor actually handles your data, and keep a human able to override the output.
- AI is already removing real busywork in campaign operations, from first drafts to QA to reporting, but it works best as a way to free up strategic time, not as a replacement for judgment.
Most AI adoption starts backwards
Aby’s advice to marketing leaders is counterintuitive on purpose: don’t start with the technology. In workshops with clients, his team routinely surfaces 40 to 50 possible AI use cases in half a day. That’s the easy part. The harder part is prioritizing which ones actually move the business.
He gave a real example. One client considered two use cases: re-tagging millions of assets in a digital asset management system, and using AI for pipeline development and cold outreach while entering a new market. Both were technically sound use cases. Only the second got funded internally, because it tied directly to revenue impact. The first wasn’t unimportant, it just mattered less relative to what the business needed right then.
The fix is simple to state and easy to skip: attach every AI initiative to a clear business goal before evaluating any tool against it.
“Teams often just dive in headfirst and start with technology, and I recommend not to do that, as counterintuitive as it may seem.”
— Aby Varma
Personalization at scale only works with clean data underneath it
Aby’s central point on personalization is that AI has removed the resourcing problem that used to cap it. Where personalization once meant producing content for a handful of predefined segments across a few channels, AI can now identify a segment of one, based on actual behavior in your database rather than categories a person guessed at in advance.
But he was clear that this only works if the data quality is there to support it. Shiv connected that directly to what AI is doing inside CRM databases right now: correcting incorrect emails and outdated phone numbers, filling in missing fields like job titles or LinkedIn profiles, and standardizing formats across the database. Historically, roughly 70 to 80% of CRM data problems fell into exactly those three buckets. With data going bad at a rate of 20 to 50% annually from job changes, closures, and mergers, keeping that process automated and continuous matters more than doing it once a year.
If your personalization strategy is more ambitious than your CRM data hygiene can support, the data is the place to fix first.
Most AI vendor evaluations skip the parts that actually matter
Aby’s framework for vetting an AI vendor goes well past the sales pitch. Pilot before you commit the whole team, and treat a failed pilot as useful information rather than a setback, fail fast and move on. Check vendor transparency on how your data is actually used and trained on, which usually isn’t in the marketing material but buried in the MSA, and involve legal and IT in that review rather than evaluating it alone.
Keep a human able to see how the AI reached a decision and override it if needed. A tool that outputs a result with no visibility into how it got there, what Aby calls a black box, is a real red flag, especially for decisions like shifting ad spend between platforms. And measure impact, not activity: more AI-generated content isn’t a KPI on its own. What matters is whether it’s driving more qualified traffic or better conversion, not just more volume.
“The buck stops with human beings. Make sure there’s an understanding of how AI is making decisions, and that you have control to change the output.”
— Aby Varma
AI is already removing real busywork from campaign operations
Kawal, co-founder of Digital DI Consultants and one of the episode’s hosts, walked through where AI is currently reducing bottlenecks day to day: drafting initial email and landing page copy, cleaning and validating audience lists before a campaign launches, generating content variations for different personas without rebuilding a campaign from scratch, flagging broken workflow logic before it goes live, catching integration sync errors between tools, and summarizing campaign performance instead of manually stitching together spreadsheets.
Her framing matters as much as the list itself: AI isn’t replacing the people doing this work. It’s freeing up the time that used to go into cleanup and manual QA, so that time can go into actual strategy instead.
Action items
Before evaluating any AI tool, write down the specific business goal it needs to move. If you can’t name one, that’s a signal to wait.
Audit whether your CRM data is clean enough to support the personalization you’re trying to build. Fix the data before scaling the campaign.
Run every new AI vendor through a short pilot with a small team before rolling it out company-wide.
Ask vendors directly how your data is used and trained on, and loop in legal or IT to review the fine print.
Track the actual impact of AI-generated content or campaigns, not just the volume being produced.
FAQs
Where should a marketing team start with AI adoption?
With the business goal, not the tool. Prioritize use cases based on which ones move revenue or a clear operational metric, rather than which ones sound impressive.
How does AI improve personalization in B2B marketing?
It can identify behavior-based segments of one instead of relying on predefined categories, and generate tailored messaging for each. This only works reliably with clean, enriched CRM data behind it.
What should you check before adopting an AI vendor?
Run a small pilot before scaling company-wide, ask directly how they handle and train on your data, and make sure a human can see how the AI reached its output and override it if needed.
Talk to Digital DI Consultants
If your CRM data isn’t clean enough to support the personalization or AI initiatives you’re planning, see how we approach CRM data hygiene, or talk to our team about your setup.
