Vetting a B2B Demand Generation Agency Before Launch

The hidden operational costs of a bad outbound partner

Hiring the wrong vendor typically burns capital and delays progress. Choosing the wrong B2B demand generation agency creates deeper operational wounds. Beyond the lost retainer, reckless outreach degrades your primary domain score, routes your emails to spam, and alienates key accounts with low-quality copy.

The real damage shows up in the recovery timeline. Fixing a broken domain reputation often requires twelve months of technical work. Rebuilding goodwill with burned accounts takes even longer. Once an agency mismanages your best prospects, those buyers stay quiet long after the engagement ends.

This failure pattern is common. A team hires a cheap provider, experiences an initial spike in outbound activity, and watches deliverability collapse by week six. When executives move to cancel the agreement, the technical sending setup is already ruined.

The Demand Department's 4-channel GTM motion only works because the infrastructure under it is built right. If the infrastructure goes wrong, every channel above it stops working. Vet the infrastructure first.

How to audit agency case studies for actual substance

Real case studies have four ingredients.

A named client. Not "a SaaS company." Not "a leading B2B services firm." A name you can search on LinkedIn.

Specific outcomes. Meetings booked, pipeline generated, close rate on those meetings, deals closed, revenue attributed. Three or four numbers, not one vague percentage.

A time range. Six weeks, three months, six months. Tells you the engagement length and the velocity.

Scope. What did the provider do? Email only? Multi-channel? Was content included? Was reply handling included?

A bad case study reads: "We helped a B2B SaaS company increase their pipeline by 300%." A good one reads: "Acme Corp, a 22-person dev agency in Austin, $150k MRR. Engagement Q3 2025 to Q1 2026. Built a 3-segment ICP, 4-channel campaign across cold email and LinkedIn. 47 qualified meetings, $312k pipeline, 9 closed deals, $84k new MRR. Founder signed a 12-month renewal."

If every case study reads like the first version, walk.

Inspecting technical deliverability and inbox setup early

Five questions, asked sequentially on the call.

How many sending domains do you buy per campaign? Right answer: 3 to 5 secondary domains. Wrong answer: "we use whatever you have."

What's your warmup protocol? Right answer: 10 to 14 days minimum, automated tool, ramping volume. Wrong answer: "we send slowly at first."

Do you use a separate domain from mine? Right answer: always. Wrong answer: any version of "sometimes" or "we recommend yours."

What's your daily send volume per mailbox? Right answer: 30 to 50 emails per day per inbox, capped. Wrong answer: 200 plus.

Which sending tool do you prefer and why? Right answer: a specific tool (Instantly, Smartlead) with reasoning. Wrong answer: "whatever the client uses."

If they fumble two of five, end the call.

Separating real market positioning from template personas

A real workshop runs 60 to 90 minutes live, with the founder and the operator both present.

Agenda covers: account criteria (firmographics, geography, tech stack, growth stage). Persona criteria (titles, seniority, function). Trigger criteria (hiring signals, funding events, competitor switches, content engagement). Disqualifiers (segments to actively exclude). Three to five segments minimum, each with its own positioning angle.

Output is a written ICP matrix, shared in a Notion or Google Doc, signed off by both sides before any list work begins.

If a provider says "we'll figure out the ICP from the onboarding form," the campaign will miss in week three. The list will be wrong. The copy will be wrong. The meetings won't book or won't qualify. By month two, you'll know the ICP work was skipped, and you'll be paying retainer to fix it.

In TDD's engagements with agency founders, the ICP workshop is week one, day three or four. Non-negotiable.

Red flags to watch for during preliminary discovery calls

Six instant disqualifiers. Each one ends the call.

Promised meeting counts before the ICP workshop. "We'll book you 15 meetings in month one" without seeing your ICP, offer, or market is salesmanship.

"We've worked with hundreds of agencies" with no names. Anonymous social proof is marketing fluff.

Refusing to share their own outbound numbers. If they don't run their own GTM, their advice is theoretical.

No clear offboarding process. Who owns the domains, the lists, the sequences, the data at exit?

Pushing a 12-month contract before a 60-day pilot. Confident operators offer pilots.

A "we handle everything, don't worry about it" tone. You should be involved in ICP signoff and copy review. If they want you out of the loop, it's so you can't see what's happening.

Reviewing sample copy to safeguard your brand voice

Ask to see three campaigns from the last 60 days, redacted of client names.

Look for five things.

Segment-specific openers. Each campaign has a different first line based on the segment. Generic openers (any version of "Hope you're doing well") signal a templated approach that won't convert.

Subject lines that don't trigger spam filters and don't read like sales copy. "Quick question on [specific business pattern]" works. "Save 30% on your customer acquisition cost" doesn't.

Clear CTAs. One ask per email. A specific time window. A direct link to a calendar.

Follow-ups that add value. New angle, new proof point, new question. Not "just bumping this" or "did you see my last email?"

Tone that matches the ICP. Agency founders read like agency founders. Healthcare practice managers read differently. The copy should reflect that.

If every email starts with "Hope you're doing well," move on.

Asking questions that isolate pipeline from vanity metrics

Four questions.

What's in your weekly report? Right answer: send volume, reply rate, positive reply rate, meetings booked, qualified meetings, pipeline created, channel breakdown, week-over-week comparison. Wrong answer: a Loom with no numbers.

Can I see a sample report from a current client (redacted)? Right answer: yes, here it is. Wrong answer: "we customize per client" with no example.

Do you have a live dashboard I can check anytime? Right answer: yes, with a link. Wrong answer: "we send weekly summaries."

How do you attribute pipeline to the engagement vs other sources? Right answer: separate columns for outbound-direct, content-direct, and mixed-touch attribution. Wrong answer: a single "influenced pipeline" number.

Vague answers mean vague reporting. Vague reporting means you won't know if it's working until month four when the retainer is gone.

Designing 30-day onboarding agreements to minimize risk

A pilot period (60 to 90 days) with a clean exit clause is table stakes.

The contract should specify five things explicitly.

Who owns the domains. Answer: you. Always.

Who owns the lists and sequences. Answer: you, with full export at exit.

What happens to the in-flight campaigns at exit. Answer: 30-day windup period, no abrupt cutoff that strands replies in their inbox.

What's refundable and what isn't. Setup fees usually aren't. Unused retainer days are.

Notice period. 14 to 30 days standard. Anything longer is a soft lock-in.

A confident provider offers all of this in the SOW. A worried provider buries it in the legal fine print or refuses to negotiate.

If they won't pilot, they're worried about month three. Walk.

Conducting back-channel reference checks with former clients

Always ask for two references. Speak to both. By phone, not email.

Ask each reference five questions.

What broke during the engagement? Every engagement breaks something. The reference's answer tells you whether the agency owned the failure or pointed at the client.

What took longer than promised? Hiring delays, copy approvals, infrastructure issues. The pattern matters more than any single answer.

What did you have to do yourself that you expected the agency to handle? Tells you the real scope vs the SOW scope.

What would you do differently next time? Honest references will give you tactical advice.

Would you sign again? This is the question. Yes with no hesitation: hire. Yes with a pause: ask why. No: do not hire.

If they offer only one reference, or stall on names, walk.

Reaching a clear hiring decision in forty-eight hours

Seven checkpoints. Score each pass or fail.

1. Named case studies with verifiable client names and specific numbers. 2. Infrastructure clarity on domains, warmup, sending tool, daily caps. 3. Live ICP workshop process with written output. 4. Clean contract with named ownership of assets at exit. 5. Willingness to pilot for 60 to 90 days. 6. Real reporting with weekly cadence and a live dashboard. 7. Two named references who answer "would you sign again" with an enthusiastic yes.

Pass all seven: sign within 48 hours. Pass five or six: ask follow-up questions, retest the failures. Pass four or fewer: walk.

The decision is binary by design. Most founders deliberate too long on the wrong providers and rush the right ones. Reverse the pattern.