It wasn’t all that long ago when any signal on ICP accounts felt like gold. Not anymore.

Fundraising announcements, office openings, job posts, new hires, tech changes, you name it. Third party data told us that the prospect might be ready to buy. They tell all our competitors, too.

There’s been such an arms race in intent signals, it feels like all my old standbys suddenly became a commodity. I don’t mean to trivialize this data (or the difficulty of wrangling it for GTM plays). But the bar for building a moat around distribution, what Clay calls GTM alpha, keeps getting higher.

The way through it isn’t ignoring the data. It’s being more creative with how you use it. Where I’m betting right now: combining third-party (3P) data with your own first-party (1P) data sources. And the potential combinations are literally endless.

Today I’m teaming up with Clay, the maker of infrastructure to launch and run self-learning GTM plays, to explore exactly how to do this. My rule of thumb: the plays that last are the ones where the first-party side is genuinely yours (a champion relationship, a lost deal’s context, usage data sitting in your warehouse) and the third-party side just tells you when to use it.

Here are the four places I’d start. Start with these, then make them your own.

There aren’t many better feelings in prospecting than when an existing champion announces a new job at a VIP prospect. Your gut instinct: immediately send a message that says “Congrats, we should catch up.”

We’ve all gotten those emails, yes? And we’ve probably all ignored them.

Data combination:

Existing champion or exec buyer (1P) + job change (3P)

The old approach:

See a job change announcement on LinkedIn (hopefully!). Immediately reach out and ask for a meeting. Continue to pester the champion for the next week, then move on.

A better approach:

This is a rare instance where it pays not to be first. The champion is overwhelmed and still in the thick of onboarding. Their inbox is getting bombarded, but that won’t last after the first few weeks.

Try waiting one month for the champion to settle into their new role. Then reach out via phone and email. Reference their past product use in the outreach (“I know your previous company used X, might be relevant here”).

Changes:

From manual to automated, always on, and intelligently sequenced.

How to do it:

How to take it further:

How to create a learning loop:

Adjust when to contact them, who qualifies, the channel sequence, and which prior relationship/usage to reference.

Only about one-in-four qualified opps become customers. And that’s already the highest-converting stage of the funnel. Go back further to SQLs or MQLs, and win rates normally fall under one-in-ten. That leaves nine warm prospects who can be reactivated.

Closed-lost deals are probably the best source of pipeline, yet too often they’re neglected to relegated to a “follow-up in 9 months” CRM reminder.

Data combination:

The old approach:

Set a follow-up reminder for 9 months after the closed-lost date. The rep is hopefully still around 9 months later and writes a manual follow-up note. The follow-up note ends up being pretty generic (“Hey, just checking in…”)

A better approach:

Closed-lost opps are already first-party data available in your CRM. The key is to gather all the needed context around the loss and to pair this with relevant external signals that indicate it’s time to re-engage.

Many (most) losses come down to things like timing, lack of budget, or not a priority. These are all things that can change, and you want to be the first to know about when they’ve changed. Three of my favorite data sources for this are leadership changes, new strategic initiatives (via job posts), and tech changes.

Changes:

From one-off and time-based to continuous and signal-based.

How to do it:

How to take it further:

How to create a learning loop: Test which new signals actually overcome specific reasons a deal was lost. For example, if the loss reason was “lost for budget” you might find that a funding event or CFO change is the best signal. Then only reactivate when the right signal appears.

My Instagram feed is essentially a never-ending scroll of tech products that I researched for this newsletter, even though I’m a solopreneur without much budget. Most retargeting ad dollars get wasted on people like me.

Ad retargeting does work. But the old approach is inefficient, and isn’t generating enough incremental pipe relative to what’s possible.

Data combination:

Website visitor (1P) + buyer groups at target accounts (3P)

The old approach:

Retarget anyone who’s been on your website in the past 30 days. Advertise across Google, Meta, and LinkedIn. Then get disillusioned because of the high costs.

A better approach:

The first step is to de-anonymize website visitors who’ve viewed high-value or paid landing pages. Look up their company and qualify the account tier (non-ICP, Tier 1 account, Tier 2 account, etc.).

If the company is on your target account list, find all relevant buying group personas. Then push this to LinkedIn and Meta for ad retargeting. Tailor the offer from awareness (ex: book a demo, read a case study) to conversion (ex: gift offer) once the intent signal has been confirmed.

Changes:

From individual visitors to buying groups at high-intent, ICP-fit accounts.

How to do it:

How to take it further:

How to create a learning loop:

Gauge which pages and visit patterns best predict buying intent for different accounts and personas, then continuously re-score behaviors. For example, maybe three docs visits are more predictive than a pricing page visit. Test who gets retargeted, with what offer, and on which channel.

54% of Anthropic’s new enterprise logos came through the self-serve funnel in 2026, Head of Industries Eleanor Dorfman shared at SaaStr AI earlier this year.

Turning product users into enterprise pipeline was perhaps one of the earliest signal-based plays combining first-party and third-party data. This felt like a big source of alpha in the earlier days of product-led growth, especially in 2020 and 2021. Now it feels like just another pillar of an increasingly complex GTM playbook.

But it’s changing (again). AI products in particular are getting inundated with personal email signups. In my experience about 75-90% of new signups are personal emails. These almost always got ignored by the sales team.

Data combination:

The old approach:

Either block personal emails entirely or essentially ignore them in the enterprise pipeline. Focus GTM campaigns on business domains only.

Look for a combination of product usage signals (ex: reach a usage limit, growing usage week-over-week) and ICP-fit (ex: above 500 employees, in the target industry). Reach out directly to the power user and pitch them on a team or enterprise-wide expansion.

A better approach:

De-anonymize personal emails to find the work email and all the relevant context around the user. Then roll-up personal email and business email users who work at the same company to get an account-level view.

Match accounts to the ICP account tiers, and focus on the best-fit accounts. For these best-fit accounts, identify the buying groups and enroll in ad retargeting (mentioned above). Track third-party signals of buying readiness like new job posts and recent news.

Meanwhile, monitor a combination of product usage data such as:

AI agents ingest all the context and propose when to reach out, who to reach out to, and which message to use. This is where agentic GTM can finally live up to its potential, eventually becoming self-learning as these micro-campaigns run.

The message might go to a product user (ex: if there’s a user with a Director+ title) or might go to an executive buyer. Reference specific product usage activity in the outreach.

Changes:

Expands coverage to personal and business email signups. Combines product users and buying groups. Shifts from regimented (and arbitrary) outreach to continuous monitoring, full context, and relevant outreach.

How to do it:

How to take it further:

How to create a learning loop: Understand which usage patterns precede expansion and who inside the account is best to approach.

The recipe is simple: combine 3rd party signals (hiring, tech, company news) with 1st party data (champion movement, closed-lost deals, website activity, product usage) for high-converting GTM plays that can’t be easily copied.

The examples shared here are (admittedly) just a jumping off point. You can and should take these much further with the data sources that are most applicable to your exact situation.

One more thing before you build any of this: decide up front how you’ll know it worked. Track them like a GTM channel: pipeline sourced or influenced per play, win rate for signal-triggered outreach versus your normal baseline, and cost per opportunity once you count the enrichment and tooling spend. If a play can’t show its work against a baseline within a quarter, it’s not a repeatable GTM motion.

Where things are going in the future is even more exciting. All of the plays mentioned above prescribed in advance (a) which data matters and (b) how that data should be used. What if agents simply ingested all the signals, then surfaced up patterns and suggested plays we hadn’t even thought about?

This future state would be autonomous, self-learning, and even harder for competitors to copy. Yes, this all means more work for less outreach. The upside is way less spray-and-pray, and way more relevance.