Most ABM strategies aren't failing because of bad targeting. They're failing because teams try to run them at full scale before proving the model works. I sat down with Andrei on the Demand podcast to talk through the account tiering framework and pilot structure that fixes this, and why lead scoring was never built to answer the question teams are actually asking of it.
Most ABM strategies fail before they get a fair test. A marketing leader gets buy-in, sales hands over a list of target accounts, and six weeks later the program is quietly running as disguised lead gen with a fancier name.
I sat down with Andrei from FullFunnel, on the Demand podcast to talk about why this happens so consistently. Andrei has built his entire practice around ABM, and what stood out in our conversation is that the failure point is rarely the target or the budget. It's sequencing. Teams try to run at full scale before they've proven the model works at all. Listen in here.
This post is my summary of that conversation, with a few of Andrei's own lines pulled in directly. If you're planning an ABM strategy and want a framework for sizing, segmenting, and measuring it before you scale, this is the guide.
Andrei described a pattern he sees repeated across organizations of every size. A marketing leader wants to launch ABM. They build a deck, get buy-in, and ask sales for a list of accounts. Sales hands one over, usually pulled straight from territory planning.
The problem is that territory planning is often opportunistic. Reps pick the biggest logos in their patch, regardless of whether there's any real relationship or evidence of need. Nobody has agreed on how many accounts the team can realistically work, or how many buyers per account they need to reach.
Once marketing has that oversized list, the only executable option is broad coverage: programmatic ads, LinkedIn campaigns, and general air cover across every account on it. Andrei put it plainly: "literally what happens to ABM, it just becomes one tiny thing, which is kinda, honestly, it's the same lead gen, but under a new source."
The bottom line: if your ABM strategy launches against every account sales can name, you don't have an ABM strategy. You have lead gen with better targeting.
The instinct to fix this is usually to get more precise: better lead scoring, tighter thresholds, more signals. Andrei's take is that scoring was never built to answer the question teams are asking of it.
Point-based lead scoring grew out of the MQL-to-SQL waterfall model, which he traces back to a specific organizational problem: marketing needed a way to demonstrate its influence on revenue. The model gave marketing a seat at the table, but it also created an arbitrary system where an email open might be worth five points and a webinar signup might be worth twenty, with no real basis for either number.
That arbitrariness is the issue. A score built to justify marketing's budget was never designed to reflect actual buyer readiness, and treating it as a readiness signal is why sales keeps complaining about lead quality, no matter how the scoring model gets tuned.
The bottom line: if your ABM strategy relies on a lead score to decide which accounts get attention, you're optimizing a proxy that was never built to measure the thing you care about.
Instead of scoring, Andrei segments accounts into three tiers using criteria agreed on by the marketing and sales team, not gut feel.
The specificity matters. Andrei was clear that vague criteria like "good relationship" aren't useful until the team defines exactly what that means in observable terms, whether that's a role tenure threshold, a technographic signal, or a specific behavior.
Once accounts are tiered, the next mistake is trying to run the whole thing at once. Andrei's recommendation is to scope the first version of an ABM strategy down to one use case, not a full vertical, because accounts in the same vertical often have different problems and buying triggers.
Pick one sales rep to partner with, and pick the most collaborative one on the team, not the top performer. As Andrei explained, "the person who is open to collaborate with marketing... who's willing to try new approaches" is a better pilot partner than someone who already has a system that works for them and doesn't want to change it.
Define real capacity together. If the rep has five hours a week for this, that determines how many accounts and buyers per account the pilot can support. From there, marketing and sales build a joint plan: shared content for the use case, one-to-one account research for active focus accounts, and a point-of-view document that shows the account you understand their specific situation, built from research rather than a sales pitch.
Note: this document isn't meant to be fully accurate. Its value lies in showing that someone took the time to understand them, which builds trust before a sales conversation starts.
Long B2B sales cycles make raw conversion counts a poor early signal for an ABM strategy. A pilot running for a single quarter may not produce a closed deal in that window, and judging it on that basis kills good programs before they've had time to work.
Andrei uses the account-to-pipeline ratio instead: out of the accounts selected for the pilot, how many produced a real discovery call? If 30 accounts produce 5 discovery calls, that ratio is directly comparable to other motions the team runs, like cold outbound, where it often takes far more volume to generate the same number of qualified conversations.
This is the same principle behind the way we think about measurement at Omni Lab. Platform metrics and lead volume are leading indicators, not proof of value. Pipeline-level signals, even early ones like a booked discovery call from a tiered pilot, tell you more about whether the model works than any score or click-through rate.
Toward the end of our conversation, Andrei made a point about AI that applies directly to how teams execute the tiering and research work above. His framing: "AI is an advanced automation." It accelerates a process you already run well. It does not create a good process on its own.
He gave an example of an AI-generated account research report that looked polished on the surface, until a sales rep clicked through and found the cited strategic initiative was two years old. The output had the right shape but the wrong substance because there was no defined, human-verified process in place before AI got involved.
The same caution applies to outreach. Andrei was direct that relationship-building touchpoints, particularly personal outreach on LinkedIn, are the last thing to automate. People recognize AI-written messages, and using them damages trust with the exact accounts an ABM strategy is trying to win.
The bottom line: build and verify the process by hand first. Only automate the parts that are already producing consistent, correct output.
The fastest way to kill an ABM strategy is to scale it before proving it works. Lead scoring won't fix a program built on an oversized account list, because scoring answers a different question than the one that actually matters: does this specific account show real evidence of need?
Tier accounts using criteria your team agrees on. Pilot with one use case, one collaborative rep, and a realistic capacity limit. Measure the account-to-pipeline ratio instead of lead volume, and bring AI into the process only once the underlying work is already good without it.
Start smaller than feels comfortable. A tight pilot that proves the model is worth more than a full rollout that gets shelved in six weeks.
At Omni Lab, we run 1:1 and 1:few ABM campaigns across a number of paid digital channels like LinkedIn Ads. Reach out if you are thinking about running a better ABM motion.
"Demand gen" has sadly become synonymous with lead generation and performance marketing. But the best agencies still understand that brand preference and demand capture are complementary, not competing. This guide breaks down how the top demand gen agencies work and why it's so important to be philosophically aligned before getting sold on the word "demand gen".
ChatGPT has 900 million weekly active users, but ads only reach the 19% on the free tier. Our own poll found that 74% of SaaS marketers are already on paid plans, meaning most of your ICP has already opted out before you spend a dollar.
Google Ads isn’t getting worse. It’s getting more selective. As AI Overviews and LLMs absorb low-intent searches, CTRs are falling while CPCs continue to rise, leaving advertisers competing for a smaller pool of higher-intent buyers. The marketers winning in this environment are shifting from a volume mindset to a precision strategy built on tighter campaign structure, stronger measurement, and a focus on commercial-intent keywords.