Nearly every demand gen motion runs on signals these days. The accounts you prioritize, the segments you target in paid, the content you put budget behind, the triggers that move someone from nurture into active sales — all of it is downstream of a set of signal assumptions.
But most teams have never verified those assumptions.
👋 Hi, it’s Kaylee Edmondson and welcome to Looped In, my newsletter exploring demand gen and growth frameworks in B2B SaaS. Subscribe to join 2k+ readers who get Looped In delivered to their inbox every Sunday.
The standard approach: buy an intent tool, define an ICP, pick a few firmographic filters, and build your motion around whoever surfaces. Over time you layer in more signals — job postings, technographic data, review site activity, LinkedIn engagement. The stack gets more sophisticated. The assumptions underneath it never get tested.
If the motion produces pipeline, the signals get credit. If it doesn’t, the copy gets blamed, or the sequence, or the channel. The signals keep running though.
The GTM Alpha Approach
A few months ago I built a custom AI skill called GTM Alpha — named for the concept Clay popularized — to identify non-obvious data points that predict buying behavior for a specific market. The stuff on the right side of the uniqueness spectrum: hard to find at scale, specific to your buyer’s world, more predictive than what your competitors are looking at.
You feed it your ICP, your product, your competitive landscape, and your best-customer profile. It generates ranked signal hypotheses — job posting patterns that indicate a company is mid-migration, review site signals that correlate with buying urgency in your category, scrapable web-based insights, that kind of thing.
Generating hypotheses is half the play though. The other half is the backtest. And almost no one does it (similarly to your ICP backtest).
What a Backtest Shows You
Last month I ran this with a client — an enterprise customer evidence platform (software that helps revenue teams collect, organize, and activate customer stories and proof points). They had decent intent coverage and a full demand gen motion running: outbound, content, paid, ABM. The motion was producing. But they couldn’t tell which signals were doing the work.
We took the GTM Alpha output and backtested it against their 30 most recent Closed Won accounts. The question: did accounts that eventually closed actually show these signals before they entered pipeline?
Three findings:
Job posting patterns held up. Accounts that posted for a Customer Marketing Manager or References Manager role in the 90 days before entering pipeline converted at a much higher rate than accounts without that signal. Nobody was tracking this before we ran the backtest.
G2 review spikes did not hold up. The team had treated G2 activity as a strong intent signal across the motion — it was influencing outbound prioritization and retargeting audiences. Fewer than 30% of their Closed Won accounts had shown elevated G2 activity in the 6 months prior to pipeline entry. The signal looked valid in the vendor’s aggregate data. It wasn’t predictive in their actual cohort.
LinkedIn content patterns held up at mid-market. Not at enterprise. Completely different stakeholder dynamics at each tier. Applying the signal universally was adding noise to segments where it didn’t belong.
One backtest changed how the entire demand gen motion was structured. The signal assumptions were wrong in specific, fixable ways. They just hadn’t been checked.
Where to Start
Most teams only pull their Closed Won accounts when they do this exercise. That’s too narrow. The dataset you actually want is four cohorts working together:
Closed Won — what signals did these accounts show before they entered pipeline? These are your positive examples. Look back 60-90 days pre-opportunity and document everything: job postings, LinkedIn activity, tools in their stack, review site behavior, leadership changes.
Closed Lost — same exercise, different outcome. Where did the signals diverge from Closed Won? Were they showing the same intent signals but missing something structural? Or did they show different signals altogether? Closed Lost accounts tell you which signals are necessary but not sufficient.
DQ’d MQLs that sales rejected — these accounts cleared your marketing qualification threshold but sales looked at them and said no. That’s a false positive. What signals were present that made marketing think they were ready? Understanding what triggered qualification without triggering a real opportunity is how you find the noise in your signal stack.
Leads that never reached MQL threshold — these came in but stalled before qualification. What was missing? Sometimes the most predictive signal is the one that separates accounts that make it to MQL from ones that don’t and why.
Think of these four cohorts as a Venn diagram. The signals that show up consistently in Closed Won and are absent or weak in the other three — that’s likely where your GTM Alpha lives. Not signals that correlate with intent broadly. Signals that are uniquely predictive of accounts that you close.
That’s the backtest. And once you know which signals hold up across all four cohorts, you build your motion around them and keep testing as new data, insights, signals come in. The idea is that your GTM Alpha should be fluid. This is how you outrun your competitors.
The GTM Alpha Skill accelerates the hypothesis generation. Give a try for yourself. I’d love to know what you find.
See ya next week,
Kaylee ✌
The G2 finding is very interesting. Nothing against review platforms but I’ve always found people over index there.