Every founder I talk to says pipeline is harder to build than it’s ever been. In my last piece, “Your pipeline isn’t broken, your strategy is,” I argued that the fix starts with deciding which accounts you go after. But, once you’ve picked the accounts that matter, how do you get one of those buyers to pay attention to you?
David Politis has spent years testing and iterating on this very question. I’ve known him for more than 20 years and he’s still the guy I call to argue about ABM and pipeline building. After the “Your pipeline isn’t broken…” post mentioned above, he sent me a four-page email listing all the plays he’s seen work and had me on his podcast to dig into them. Here’s what I took away.
Welcome to the attention economy
The challenge of pipeline building really comes down to a fundamental change in the buying environment. Three things are hitting buyers at once. AI made software cheap to build, so they’re looking at hundreds of thousands of options that are hard to tell apart. They’re scared of getting fired for buying tech that doesn’t work, and their CFO is nervous about the economy. And we destroyed the email inbox with AI slop.
Five years ago an SDR could send 120 emails a day and book 50 meetings a month. Be a goldfish and forget those numbers. Outbound still has a place, but the spam cannon is done unless you’re a mature company with a big installed base or you’re one of the handful of companies everyone wants to talk to right now.
Take the micro-segment trust fall
Politis calls micro-segmentation a trust fall, and he’s right. Think about how counter-intuitive it sounds. With how hard it is right now to get people’s attention, you would think more is better. The reality is that because it is so noisy and crowded, you have to find a smaller group that you are most likely to be successful with, and then deliver personalized, memorable experiences to get their attention. It is scary, but it is step one towards effective attention-grabbing.
Here are some micro-segmentation ideas we have seen work::
Look for a homogeneous use case. I got that phrase from David Boskovic, the CEO of Flatfile. Find the use case and microsegment the accounts that likely fit. His rule: if a micro-segment has more than 500 accounts, try again
Start from accounts you’ve won that didn’t churn. Then find lookalikes
Backtest against your best rep. Politis did this with a company that had no ICP. They defined one, then checked it against the top rep’s deals: 87% fit. The worst rep’s deals: 32% fit. The top rep had found the profile on his own without even knowing it
No win data yet, or moving into a new segment? Get on the phone. Talk to 10 to 20 people a week who fit the profile you think is right, and pitch them instead of just interviewing them. Until you ask someone for money or a next step, you don’t know how they really feel. Start somewhere and then start tracking data. Everyone refines segments over time.
Unique gets you in the door
Remember when I mentioned Politis’ email he sent after my Pipeline Playbook article? He gave offered up some great ideas which I incorporated below. These plays have one thing in common: they give the buyer something before asking for anything in return.
Offer value before asking for the meeting.
HubSpot’s Website Grader generated thousands of leads a week by telling people how to fix their SEO in exchange for a URL and an email. You can do a heavier version by hand. At BetterCloud, Politis’s team had prospects install the product before paying, pulled data from all their SaaS apps, and walked them through a report on risky apps, elevated permissions and exposed data that they couldn’t have produced themselves.
Today AI makes this effort faster and cheaper. Run deep research on one account, merge it with your case studies into a custom whitepaper, and turn that into a short audio briefing with NotebookLM that the buyer can play on a commute. Or mock up your product with their logo and data and send a Loom walkthrough. Politis has seen that land best when it’s sent through LinkedIn. Boskovic’s team at Flatfile uses AI to write market reports for a segment and sends them to a few executives. When someone writes back asking who wrote it, they know the offer works.
Give your buyer the spotlight.
Most podcasts are built around the host. An effective B2B podcast is built around the buyer you want to reach. Telling a buyer you want to feature them because of what they’ve built engages them far more than a demo request, and the prep call and interview offers legitimate opportunities for discovery. A “state of” survey is a similar vehicle. Assemble 15 to 25 questions, offer a $25 to $50 gift card per response, send three to five emails direct from the CEO, inviting buyers to participate. In six to eight weeks, you can have proprietary data and perspectives that build credibility in your market.
Create peer-learning opportunities.
Host off-the-record peer-learning groups to discuss a topic they care about. Invite people from different companies with no competitors in the room. The small group helps create exclusivity. Buyers come to meet peers and hear what’s working elsewhere.
We’ve done countless dinners like these and they are straight up magic. Make 40 to 50% of the room customers so they do the selling. When Politis was at BetterCloud, his team hosted 15 people (ten prospects, five customers) at Eleven Madison Park. It cost $20,000 and drove roughly $3 million in revenue. Nobody presented. A customer stood up and toasted them unprompted.
If you don’t know what people want to do, ask: send a list of four or five options (whiskey tasting, a ballgame, axe throwing) with dates and build the events people pick.
Also, reach out to buyers when you’re traveling, let them know you’ll be in town visiting someone else and ask if they’re free. Politis finds the hit rate is higher than asking to fly in just for them because there’s less pressure.
Connect with buyers on a personal level.
One rep learned their prospect was a UGA football fan and a self-described girl dad, so they sent a custom UGA jersey with “Girl Dad” on the back. The prospect reached out to set up a call when it arrived.
Someone selling to Politis framed a New York Times article he’d been featured in and sent it with a handwritten note. He hung it on his office wall and thought of them every time he saw it.
Gift cards to a good local restaurant beat DoorDash. For big enterprise targets, David say great ROI on mailing iPads preloaded with a video: send 10, get five calls, close one, and it pays for itself.
Create high-energy webinars with social-like engagement.
Webinars aren’t dead, but bad ones won’t fly with your prospect anymore. The ones that drive greater buyer engagement are those with two or three polls, a high-energy presenter, and active chat conversations. Politis also points out that telling registrants a session won’t be recorded raises show rates. I run mine to a list of about 60 and get 20 to 30 people. That works once you’ve earned credibility with a community.
Note: Politis shared more than 20 different plays with me, but many require meaningful effort and meaningful cost. If your average contract value is under $50,000, most of these don’t make sense.
Be unmistakably human
Kristina McMillan, who works with me at Scale, has a phrase I’ve stolen: unmistakably human. In the months since Politis and I recorded our podcast episode, AI has flooded every channel with good enough messaging. Every buyer I talk to now gets a daily stack of AI-written emails that all sound the same.
With the right segmentation and signals in place, your differentiator is now creating interactions only a person could have designed. People can tell when real care went into something. At a dinner, that looks like Uber codes so nobody has to drive, a name tag and a gift at every seat, and a guest list curated so well that nobody needs a conversation starter. Buyers know an AI didn’t do that.
Everyone who reads this will agree, yet very few will do it because it takes a lot of work. That’s also why it works. And it’s why micro-segmentation comes first. You can only throw the book at accounts you’ve already decided you’re going to reach through hell or high water.
What to do this week
Backtest your ICP with AI. Pull the call recordings and CRM notes from last year’s won and lost deals and run them through your conversation intelligence tool or an LLM. Have it cross-reference them against your ICP: which attributes show up in wins, which show up in losses and churn, and what your top rep asks that others don’t. Fold what it finds back into your ICP definition.
Have AI draw the segment. Export your own accounts that didn’t churn and have AI cluster them by use case, then build a lookalike list for the strongest cluster. If it comes back with more than 500 accounts, have it split the cluster again until every account shares one use case.
Put an agent on your “when” signal. Ask AI to list the events that would make a buyer in your segment need you now, like a compliance certification about to expire. Pick the one you can track, set up an agent to watch for it across websites, filings, job posts and news, and route each hit to the account owner in your CRM.
Scale a give-first offer, then check every one. Have AI run deep research on 20 accounts in your segment and draft a custom report or mocked-up demo for each. Have a person review every one before it goes out from your CEO. If responses are low, test the same offer through a different channel (the AE, LinkedIn) before you change the offer.
Let AI prep the dinner and keep the dinner human. Invite 10 to 12 people, at least 40% of them customers. Use AI to research each guest so you can seat them next to the right peer and pick a gift that’s actually about them. Then leave the pitch deck at the office.
Now, go pick your accounts and get to work.
Listen to my full conversation with David Politis on the Not Another CEO podcast, and read his “8 account-based marketing examples that cut through the noise”.
Craig Rosenberg is the Chief Platform Officer at Scale Venture Partners, where he works with portfolio companies on GTM strategy and execution. He was the co-founder of TOPO, which was acquired by Gartner, and the host of The Transaction podcast.
'Unmistakably human' is going to get more valuable as the AI emails pile up. A short video of an actual person talking to one account is very hard to mistake for automation.