Why CRM-based persona discovery beats job-title targeting (and how to switch in a week)

Most B2B paid-media plans start with a one-line audience: "VPs and Directors of Marketing at SaaS companies, 50-500 employees, US-based." It sounds rigorous. It's actually a guess.

It's a guess because nobody on the team checked whether the buyers who actually convert look like that. They started with a persona deck from last year's planning offsite and built every campaign downstream of it.

The cost of that guess shows up six weeks in, when CPA has crept up 30% and the team is rotating creative to make the budget stretch.

The job-title audience is a proxy — and it's a leaky one

Job titles are an easy field to filter on. LinkedIn's Campaign Manager has them. Google's customer match has them. Your CRM has them.

But "VP of Marketing" describes 50,000 people on LinkedIn alone. Some of them are:

• Solo VPs at 8-person startups burning through founder savings
• VPs of brand at $200M enterprises who never touch performance media
• Interim VPs on a 3-month engagement
• Career-changers still listed as "VP" from their last role
• Yes, also the actual decision-makers you're trying to reach

When you target all of them as one audience, you spread budget thin across people who'll never close. Your top-of-funnel volume looks fine. Your sales-qualified rate doesn't.

What your conversion data is trying to tell you

Look at your last 12 months of closed-won deals. Strip out the noise — exclude one-off small contracts, churned accounts, anything closed by a founder relationship. You're left with maybe 20-50 actual ICP wins.

Now look at those buyers. What's common across them is rarely their job title. It's a combination of things:

• Company stage — Series B specifically, not "Series A through C"
• Tech stack signals — they all use HubSpot, all use Mixpanel, all run their CRM through their RevOps lead
• Trigger events — new hire announcement 60 days before close, a fundraise within the trailing 6 months
• Behavioral signals — multiple stakeholders visited /pricing within a 14-day window
• Role context — not "VP Marketing" but "VP Marketing who also owns the demand gen budget"

That last one matters most. A VP of Marketing at a company where Growth owns paid media isn't a buyer. The same title at a company where Marketing owns it absolutely is.

The cheapest persona you can buy is the one your closed-won data already revealed.

How persona discovery actually works (the boring version)

The mechanics aren't magic. There are three steps:

1 — Pull the truth from your CRM

Every deal in your CRM has metadata: company size, industry, deal stage history, attached contacts. Most of it is junk. A small fraction encodes the pattern of who closed and why. The work is separating signal from noise.

2 — Match that pattern back to the ad platforms

Once you have a sharper persona definition — say "RevOps-aligned demand gen leader at Series B-to-D SaaS, primary tooling: HubSpot + Mixpanel" — you translate it into LinkedIn's job function filters, Meta's interest layers, Google's customer match seeds. Imperfect, but a 10x improvement over "VP Marketing."

3 — Keep the loop running

Personas drift. Your ICP six months ago isn't your ICP today — you've raised, your product has matured, your AOV has shifted. Persona definitions need to update with that. Most teams don't have time to redo this work quarterly. So they don't.

What changes when the personas are right

Three things, usually in this order:

1. SQL rate climbs first. Targeting tighter audiences means the leads who come through are closer to fit. Sales stops complaining about "garbage from marketing."
2. CPA drops second. Tighter audiences mean less wasted spend on people who'll never convert. ROAS improves without changing your bid strategy.
3. Sales velocity improves third. Deals close faster because the buyers showed up already aware of their problem.

The third one is the one the CFO notices. Pipeline velocity is the metric that lets a marketing team go to the next board meeting with a different story.

How to switch — without rebuilding everything in a week

You don't need to scrap your existing campaigns. The fastest path is:

• Spend a half-day reviewing closed-won deals from the last 4 quarters
• Write down the 3-4 patterns that show up across the wins
• Layer those patterns on top of your existing targeting as exclusions and inclusions — not as a full replacement
• Measure the SQL rate by audience layer over the next 30 days

If the SQL rate of the "matches persona pattern" segment is materially higher than the old broad audience, you have your answer. Promote the new persona to its own campaign, and let the old one die down.