Hello {{First name|Predictable Revenue community}},

Book update: I’m going back to an old-school (last summer tactic), if you want the full guide that this newsletter is based on, send me a copy of your receipt and I’ll share the full 12-page guide.

Community update: our next GTM Club is March 13th (next Friday) at 930am PT, this is a 90 minute Q&A where everyone shares their top revenue/PMF challenge and I do my best to help. I do it because I like helping people with stuff like this and it’s fun to get a chance to meet the folks that read the newsletter. Hit reply if you want to join us.

Onto the newsletter…

I've been helping a few companies figure out their outbound lately, and I've noticed I keep having the same conversation. When I find myself explaining the same thing for the fourth or fifth time, I get frustrated and start to realize I need to write it down. This is one of those things.

Some of this almost made it into The Terrifying Art of Finding Customers. When I first started writing that book, I was actually writing a follow-on to Predictable Revenue. But at the time, Clay wasn't a thing. Outbound wasn't very fun or interesting. There wasn't much new to add. For years, the constraint on outbound was data. You worked with what ZoomInfo gave you, a list, a title, a phone number, and you built your targeting around the edges of that. I did write some of this framework back then, but it wasn't compelling because all of these triggers had to be processed manually.

A couple of years ago, we literally had a team of people we called sales research reps. Their job was to go and find these signals by hand. Xiadani led that team for us, and they were amazing. But they were human. They couldn't look at 32,000 rows in five minutes.

Clay changed that. And then cheap AI models changed it again. When OpenAI’s 4.1-mini model came out, running a sheet of 25,000 companies through a basic classifier cost less than $5. That's a total no-brainer amount of money to spend. Before that, the same query with O3 Mini was $30 or $60 depending on the complexity. I still did it at that price, but I was careful with it. With 4.1-mini, I stopped being careful. I ran everything. And that model fundamentally changed how I thought about and used Clay.

That's what made this piece worth writing again. The tools can finally do what I always wanted them to do. But here's what happened to most teams in the meantime: they took all this new capability and went wide. More contacts, more campaigns, more signals, more tools. What they never did was go deep. They never asked the basic question that every outbound motion needs to start with: who are we actually going after, and how much are we willing to invest in each of those accounts?

This framework is how I see modern outbound, how I organize the signals, the inferred intent, the plays. Whether you're using SDRs to call, email, or prospect on LinkedIn, I think this is the way. I'm trying to bring some clarity to something that I think a lot of people got lost in when they made the jump from the ZoomInfo world to the Clay world.

Before you buy a signal tool or enrich a single contact, you need to define your total addressable market. Your full universe. Every account you could ever sell to.

I know. It's not sexy. Nobody gets excited about building a TAM spreadsheet. But here's why it matters: TAM defines your total spend envelope. Until you know how many accounts are in play, you can't make rational decisions about what each one is worth. And if you can't answer that, every downstream decision is a guess.

Once you have the universe, you score it. Firmographic scoring, not signal-based, not intent-based. Static fit. Does this account match the profile of companies that actually buy from us? Industry, headcount, revenue, geography, business model, historical win rates against similar accounts. The boring stuff.

That score produces a tier. And the tier controls everything else.

Most teams get this backwards. They start with signals and work backwards to accounts. The framework that actually works starts with accounts and works forward to signals.

Your account tier determines how much signal monitoring you invest. Top tier accounts get frequent scans across expensive signal sources. Bottom tier accounts get lighter monitoring on a less frequent schedule. This is a cost allocation decision, and it only makes sense if you've already decided which accounts are worth the investment.

Tier also governs contact sourcing depth. Top tier accounts get every buying influence sourced with full data: email, phone, LinkedIn, social handles. You might even monitor individual contacts at those accounts for social activity or podcast appearances. Mid tier gets primary buyers with core data. Bottom tier gets a single primary contact.

Then there's play selection. Plays have a cost. You don't deploy an expensive play against a low-value account. A top tier account with a new CMO gets a fundamentally different outreach motion than a bottom tier account with the same signal.

There are two types of signals and they work completely differently.

Active signals are time-sensitive. Something just happened. A company raised money, hired a new VP of Sales, posted a job with a relevant keyword, showed up at an event you're attending. These fire immediately. Any active signal triggers action regardless of account tier. The play you deploy will be calibrated to the tier, but you never ignore or queue an active signal.

Inferred signals are different. They're structural. They describe what's true about a company based on how it's built, not what just happened. Do they have an SDR team? Are they using a complementary technology? Are they missing a tool they'd be expected to have at their stage? These change slowly and you update them periodically, maybe annually, more often for top tier accounts. On their own, an inferred signal doesn't tell you to pick up the phone. But stack a few of them together and you start to understand how likely an account is to buy.

The two combine into a propensity score. Recent signal activity plus inferred signal quality equals a "likely to buy this month" ranking. Your firmographic score already told you the account is worth going after. The propensity score tells you whether now is the time.

The CRM. This is where the whole thing either works or falls apart.

All of it, the scores, the signals, the play assignments, needs to surface in Salesforce so reps know exactly where to focus today. If a rep has to leave the CRM to figure out what to do next, the system isn't working.

Separate pipeline views for separate funnels. Your Meet pipeline (active prospecting, sorted by propensity) should never be mixed with your Disco pipeline (active evaluations) or your Nurture pipeline (engaged but not ready). Mixing them destroys pipeline clarity and makes velocity metrics meaningless.

And every account needs a next action and an action date. (If you’re working opportunities instead of accounts, put the next action fields on the opportunity object.) That's what converts CRM from a historical record into a daily task manager. Reps start their day by looking at what's due today, not scrolling through a list trying to remember where they left off.

Accounts get reassigned to reps monthly based on current firmographic tier and propensity score. An account that was quiet last month might fire three signals this week and jump to the top of the priority stack. The list looks different every month because the data looks different every month.

That's the difference between a system and a spreadsheet. A spreadsheet is a snapshot of what was true when someone last updated it. A system recalculates on its own and tells you what's true today.

I've written the full framework document with the complete signal taxonomy, play types, contact tiering model, and customer watchlist logic. If you want the whole thing, reply with a screenshot of your book receipt and I'll send it over.

The principle underneath all of it is simple. Define your universe. Tier it. Let the tier control everything downstream. And make sure it all shows up where your reps actually work, not in some dashboard nobody opens.

Collin

PS - I just wanted to take a moment to appreciate you, thank you for reading this newsletter. Thank you for reading (or listening) and reviewing my book. The book crossed 30 five star reviews this week and it’s just a cool feeling to know that the two years of work I put into creating it is helping others.

PPS - as I write this, I’m about to ship my first Pull Request with real code for my CTO to review. I’ve vibe-coded a bunch of things but this is my first code submission that’ll touch a real code base. It’s exceedingly small (all front end stuff) but it feels like a cool moment for this non-technical co-founder.