How AdRoll Eliminated a 3-Day Signal Lag With a Scalable Signal Engine

"I easily saved six, seven hours every single time I want to create a new table."

— Kelly Hogan, AI Operations, AdRoll

METRICS

• 10+ signals built on a shared, reusable architecture

• 6–7 hours saved per new signal table

• 3-day signal lag eliminated; alerts now real-time

BACKGROUND

AdRoll, part of the NextRoll family of products, helps B2B and D2C marketers run cross-channel campaigns across display, social, CTV, and retargeting. With AdRoll, AdRoll ABM, and RollWorks under one roof, the marketing team is responsible for surfacing the highest-intent accounts to their SDR organization at the right moment, at scale.

THE CHALLENGE

A year before working with The Kiln, AdRoll's revenue marketing team had attempted to build a signal-based selling motion using MadKudu. The system was cumbersome. Signals lagged three days between when they fired and when they reached an SDR. Most of the signals were commoditized: promotions, generic firmographic triggers, the same plays every other team in the space was running. Worse, building a single signal table internally took hours of manual column-by-column setup and didn't scale beyond a handful of plays.

"We were using MadKudu, and the system was very cumbersome," Kelly Hogan, AdRoll's AI Operations lead, said. "It had a lag time of three days from when a signal was shown to when it would actually get into an SDR's hands. A lot of the signals we were going after were really run-of-the-mill. We needed something faster, with more unique signals, and an architecture we could grow on top of."

Kelly and her team came to The Kiln with three goals: cut the time from signal to action, build internal and unique signals beyond what every competitor was running, and create a scalable foundation that could turn signal-based selling into a continuous growth program rather than a static set of plays.

THE BUILD

The Kiln rebuilt AdRoll's signal-based selling foundation from scratch in Clay, designed around a shared function architecture. Instead of building each signal as a one-off workflow, five shared functions handle the heavy lifting: a Company Function qualifies every account against ICP, agency, tech stack, and Salesforce status checks; a Contact Function sources the right contacts at qualified accounts; a Contact Validation sub-function confirms titles and emails; an Email Enrichment function fills gaps from Clay-sourced contacts; and a Push to SFDC function handles contact creation, campaign membership, and personalized ad copy generation.

More than ten signals plug into that foundation, spanning Salesforce-sourced plays (active pixel, closed-lost opportunities, previous customers, funding, tech install) and externally sourced plays (job changes, hiring, lookalikes, social listening). Every one of them flows through the same shared functions, which means a single change to the qualification logic propagates across the entire system instantly. Adding a new signal takes hours instead of days.

THE IMPACT

The impact was immediate. The 3-day lag disappeared; SDRs now get real-time alerts in Slack the moment a qualified company triggers a signal. Across the first batch of signals, the system processed [X rows], qualified [Y companies], and sourced [Z contacts] with personalized ad copy generated for each. The SDR team began booking opportunities directly from Slack alerts in the first weeks of operation.

For Kelly, the bigger win is the architecture itself. "Even just thinking about the time saved, I easily saved six, seven hours every single time I want to create a new table," she said. "What used to take days in MadKudu now takes hours. And the way you've set it up means we can turn this into a growth program. If a signal is performing well, we pour resources in. If it's underperforming, we sunset it and test something new. We can iterate on it, grow from it, and learn from it."

AdRoll's outbound team is now operating off a much narrower, higher-fit account list, with a clear goal in front of them: double the connect rate of the SDR team by focusing only on accounts most likely to respond. The signal-based selling motion AdRoll once tried to build internally now runs on infrastructure that will scale with the team for years.