The insider version. We run a cold email agency -- here are the patterns that separate agencies that deliver from agencies that waste your money.
Rees Bayba
Founder, Astra GTM
TL;DR
We run a cold email agency. We manage 100+ sending domains, send 50,000+ emails per month across 8+ clients, and operate across six industries. We also regularly inherit clients who just left another agency. When that happens, we see everything: the infrastructure they were handed, the copy that was written on their behalf, and the reporting they received.
This is the guide we wish every buyer had before signing. Not because we are better than every other agency -- but because the patterns of bad agencies are consistent, visible in advance, and entirely avoidable if you know what to look for.
Open rates in cold email are broken. Apple Mail Privacy Protection, which launched in 2021 and now covers roughly 50% of business email, pre-fetches email content and registers a fake "open" even if the recipient never looked at it. Microsoft and Google run their own scanning that inflates opens further. When an agency leads with "67% open rate!" in their first report, they are either unaware of this or hoping you are.
Open rates have not been a reliable metric since 2021. Any agency building their reporting around them is either uninformed or deliberately misleading.
What good agencies report: positive reply rate (replies that express interest, not just "unsubscribe me"), meetings booked, pipeline value generated, and cost per meeting. These are the metrics that connect outbound to revenue. If the first number in their report is open rate, ask why. If they cannot explain the Apple MPP problem, that tells you how closely they follow the space.
Sending infrastructure means domains, mailboxes, DNS records, and warmup history. Who purchases the domains? Who owns the mailbox accounts? Whose credit card is on file? This matters more than most buyers realize, for two reasons.
Want this built for your team?
We implement these systems end-to-end. First campaigns live in 14 days.
First, if the agency sends cold email from your primary domain (you@yourcompany.com), a deliverability problem burns your real business email. Your support tickets, your client communication, your internal email -- all of it can be affected. Cold email should always run on dedicated alternate domains (tryourcompany.com, yourcompanyhq.com) that are isolated from your primary operations.
Second, if the agency uses shared infrastructure across multiple clients, your reputation is tied to their other clients' behavior. One client sending aggressive volume or bad copy can tank the shared IP reputation, and your campaigns suffer. Dedicated sending domains per client, purchased in the client's name, is the standard. Ask explicitly.
Here is the real timeline for a properly built outbound program. Week 1-2: onboarding, ICP definition, account list building. Week 2-3: domain purchase, DNS configuration, mailbox provisioning. Week 3-5: domain warmup (the mailboxes send and receive low-volume emails to build reputation). Week 5-6: first campaign sends go live, initial volume ramps. Week 6-8: replies start coming in, first meetings booked.
That is a strong result -- and it still took 3 weeks of infrastructure setup before the first email went out. Fast results require patience upfront.
Any agency promising "20 meetings in your first month" is doing one of three things. Using pre-warmed infrastructure shared with other clients (risky). Skipping warmup entirely and sending from fresh domains at full volume (reckless). Or lying. Honest agencies set expectations: infrastructure takes 2-3 weeks, warmup takes 2-3 weeks, first meaningful results appear in month 2, and performance ramps from there.
| Timeline | What should be happening |
|---|---|
| Week 1-2 | Onboarding: ICP definition, account list building, CRM audit, competitive messaging review |
| Week 2-3 | Infrastructure: domain purchase, DNS setup, mailbox provisioning, warmup begins |
| Week 3-5 | Warmup period: low-volume sending to build domain reputation. Copy drafts written and reviewed. |
| Week 5-6 | First campaigns go live. Volume ramps gradually. Initial data starts coming in. |
| Month 2 | Optimization: first meaningful reply data. 3-8 meetings typical for a well-targeted campaign. |
| Month 3+ | Full velocity: 10-25+ meetings per month with ongoing copy testing and segment expansion. |
This should be obvious. An agency is sending emails on your behalf, from domains associated with your brand, to people you want as customers. You should see every word before it goes out. We show every client their copy rendered exactly as the recipient will see it -- subject lines, body text, variables filled in with real prospect data, follow-up threading, everything.
We have inherited clients who had no idea what their agency was sending. One client discovered their agency was using the phrase "I noticed your company" in every email for 6 months -- to every prospect, across every segment. No variation. No personalization beyond a first name and company name. The client was paying $6,000/mo for a mail merge.
Don't do this
Agency writes copy, adds it to campaigns, starts sending. Client sees the emails for the first time when a prospect replies (or complains).
Do this instead
Agency writes copy, sends rendered previews with real prospect data filled in. Client reviews, provides feedback, approves. Only then does the campaign go live.
Ask your agency: when do I see the emails? If the answer is anything other than "before they send," you have no quality control over your own brand voice. And ask to see them rendered with real data, not templates with placeholder brackets. {first_name} and {company_name} look fine in a template. "Hi there, " with a blank where the first name should be is what the prospect actually receives when data is missing.
Ask your agency what verification they run before sending to a contact. If the answer is "we use Apollo" or "we pull from ZoomInfo," that is sourcing, not verification. Those platforms find email addresses. They do not confirm the address actually works, the person still works at that company, or the email will not bounce.
Above 3%, deliverability damage accelerates. Every bounce tells inbox providers your list is bad. We verify every email to a 97+ deliverability score before it enters any campaign.
Here is why this matters to you specifically: the bounces happen on domains associated with your brand. The reputation damage accrues to your domains, not the agency's. If the engagement ends and you want to run outbound yourself or with a different partner, those burned domains follow you. A good agency runs multi-source enrichment (finding emails from multiple data providers) plus deliverability verification (confirming each email will not bounce) before any contact enters a campaign.
A quick test: ask the agency what their average bounce rate is across all campaigns. If they do not know the number immediately, or if it is above 3%, their verification process has gaps. If they cannot explain the difference between a catch-all domain and a verified deliverable address, they are not running verification at all.
"Hi {first_name}, I noticed your company {company_name} is growing..." is not personalization. It is a mail merge. Real personalization ties to something specific about the prospect or their company: a recent hire that signals a pain point, a technology decision visible in their stack, a funding round that changes their priorities, a competitive dynamic in their market.
Don't do this
Hi Sarah, I noticed Acme Corp is a leader in the logistics space. I wanted to reach out because we help logistics companies improve their operations.
Do this instead
Sarah -- saw Acme just posted 3 backend engineering roles after the Series C. Most teams scaling that fast find their existing deployment pipeline breaks around 10 services.
The test is simple. Ask your agency to show you 5 emails sent to 5 different prospects in the last 30 days. Read them side by side. If the structure, cadence, and arguments are identical with only names and company names swapped, you are looking at a template operation. If each email references something specific to that prospect's situation, the agency is doing real work.
That is a 17x difference. The effort to personalize is not optional -- it is the difference between a program that produces meetings and one that produces unsubscribe requests.
A 12-month contract with no performance minimums means the agency has already been paid regardless of whether you get a single meeting. They have zero financial incentive to fix something that is not working. Their downside is capped. Yours is not.
What confident agencies offer: month-to-month agreements, or a 3-month initial term with clearly defined performance benchmarks. At 90 days, both sides evaluate results against those benchmarks. If targets are missed, you can exit with 30 days notice. Some agencies (including us) use a hybrid model: a base retainer for infrastructure and operations, plus a per-meeting fee that aligns incentives directly with results.
| Contract structure | What it signals |
|---|---|
| 12-month lock-in, no performance clause | Agency is protecting revenue, not confident in results. You carry all the risk. |
| 6-month contract with 90-day performance review | Reasonable. Gives the program time to mature. The review provides an exit if results are not there. |
| 3-month initial term, month-to-month after | Agency is confident. They believe results will earn the renewal. Risk is shared. |
| Base retainer + per-meeting fee | Aligned incentives. The agency makes more money when you get more meetings. Both sides win or lose together. |
One nuance: a 3-month minimum is reasonable. Outbound takes 4-6 weeks to produce meaningful data, and 90 days is the minimum to run two full optimization cycles. An agency asking for a 3-month commitment is being honest about how long it takes to prove the channel. An agency asking for 12 months with no outs is being cautious with their revenue, not yours.
Red flags get the attention, but green flags are equally useful. Here is what to look for when evaluating agencies that have passed the red flag screen.
| Red flag | Green flag |
|---|---|
| Reports open rates prominently | Reports positive reply rate, meetings booked, cost per meeting |
| Uses your primary domain or shared infrastructure | Purchases dedicated sending domains in your name, separate from primary |
| Promises 20 meetings in month 1 | Sets honest timeline: infrastructure in weeks 1-3, first results in month 2 |
| You never see copy before it sends | Sends rendered previews with real prospect data for approval before every launch |
| Uses a single data source without verification | Multi-source enrichment plus deliverability verification with measurable bounce rate |
| Same template for every prospect | Situation-specific copy with prospect-level personalization signals |
| 12-month lock-in with no performance clause | 3-month initial term with defined benchmarks and performance-based exit options |
A few more green flags that are harder to spot but matter. They walk you through their infrastructure setup in detail during the sales process -- what domains they will buy, how many mailboxes, their warmup timeline. They proactively tell you when something is not working and suggest pivots instead of waiting for you to notice. They can name the specific person who writes your copy and manages your campaigns.
Here are the 12 questions that separate a thorough evaluation from a cursory one. Any good agency will answer these without hesitation. If an agency deflects, generalizes, or gets defensive, that is your answer.
Pay close attention to question 6. Asking to see real emails is the single most revealing request you can make. The copy quality, personalization depth, and strategic thinking all become immediately visible. An agency that hesitates to show real work is an agency that does not want you to see it.
Pricing varies, but there are ranges that signal where an agency sits. Below $2,000/mo, the agency is either a solo operator with minimal infrastructure or cutting corners on data quality and verification. Between $2,000-5,000/mo is the typical range for a dedicated engagement with proper infrastructure, copy, and campaign management. Above $5,000/mo usually includes multi-channel (email + LinkedIn + phone) or high-volume enterprise campaigns.
| Price range | What you should expect |
|---|---|
| Under $2,000/mo | Template-based campaigns, shared infrastructure likely, limited personalization. May work for simple, high-volume use cases. |
| $2,000-5,000/mo | Dedicated infrastructure, personalized copy, contact verification, regular reporting. The standard for serious B2B outbound. |
| $5,000-10,000/mo | Multi-channel execution, deep research and personalization, strategic consulting, potentially dedicated team members. |
| $200-500 per meeting (hybrid) | Base retainer for operations + per-meeting fee. Aligns incentives. Typical total cost: $3,000-8,000/mo when performing well. |
The cheapest option is almost never the best value. An agency charging $1,500/mo that books zero meetings costs you $4,500 over 3 months plus the opportunity cost of three months without pipeline. An agency charging $4,000/mo that books 8 meetings per month costs you $500 per meeting. Price per meeting is the metric that matters, not monthly retainer.
If you recognized multiple red flags in your current agency, here is the exit playbook. First, request ownership transfer of all sending domains and mailboxes. These should already be in your name -- if they are not, that is red flag #2 confirmed. Second, download all campaign data: contacts, send history, reply history. You need this for suppression lists regardless of what you do next.
Third, audit the damage. Check your domain reputation using Google Postmaster Tools and mail-tester.com. If domains are burned (spam rate above 1%), you may need to start fresh with new domains rather than trying to rehabilitate. Fourth, do not rush into a new agency. Run the evaluation checklist above. A 2-week gap between agencies is nothing compared to 6 months with the wrong partner.
How quickly can I tell if my agency is underperforming?
By the end of month 2, you should have data. Infrastructure and warmup take weeks 1-4, first sends go out weeks 4-6, replies come in weeks 6-8. If you are at the end of month 2 with zero meetings and the agency cannot show you positive reply rate data, ask direct questions. If you are at the end of month 3 with fewer than 3 meetings and no clear explanation of what is being changed, the relationship is unlikely to improve.
Should I ask for references from an agency's current clients?
Yes, but ask correctly. Do not ask for their best case study. Ask to speak with a client who has been with them for 6+ months and one who started in the last 90 days. The long-term client tells you about consistency and communication. The new client tells you about the current onboarding experience. If the agency cannot produce either, that is informative. Also ask references specifically about the red flags: do they see copy before it sends, what metrics are in the reports, how was the infrastructure set up.
Is a per-meeting pricing model always better than a flat retainer?
Not always. Pure per-meeting pricing can incentivize booking low-quality meetings that waste your sales team's time. A qualification definition needs to be in the contract. The best structure is hybrid: a base retainer that covers infrastructure, operations, and campaign management ($1,500-3,000), plus a per-meeting fee ($150-500) that rewards results. This ensures the agency is funded to do the work properly while also being incentivized to produce outcomes.
What if an agency says they cannot show me real emails due to client confidentiality?
Partially reasonable. They should not share another client's prospect names or proprietary messaging angles. But they can show you anonymized emails -- real copy with company names and personal details redacted. If they refuse even that, they either do not have examples worth showing or their copy is so generic that anonymizing it would make that obvious. Either way, it is a red flag.
How many sending domains and mailboxes should a good agency set up?
For a standard B2B campaign: 3-5 sending domains with 2-3 mailboxes each, for a total of 6-15 mailboxes. At 25-30 sends per mailbox per day, that gives you 150-450 emails per day, or 3,000-9,000 per month. This is enough volume to test multiple ICPs and copy variants without overloading any single domain. If an agency is running your entire program from 1-2 mailboxes, they are either sending too much volume per box or constraining your campaign capacity.
Can I audit my current agency's deliverability myself?
Yes. Ask for access to your sending platform or request they send a test email from one of your campaign mailboxes to mail-tester.com. The score should be 9+/10. Also check Google Postmaster Tools if your sending domains use Google -- it shows spam rate, domain reputation, and authentication status. If your agency resists giving you access to these tools, they may be hiding deliverability problems.
What is the difference between a cold email agency and a lead gen agency?
A cold email agency specializes in email-based outbound. They own the technical infrastructure, copy, deliverability management, and campaign execution for the email channel. A lead gen agency is broader -- they may use cold email, LinkedIn, cold calling, paid ads, or content syndication. Cold email agencies tend to be more technically proficient with deliverability and infrastructure. Lead gen agencies tend to be more diverse in channels but sometimes weaker in email-specific execution. Ask which channels they actually operate and whether email infrastructure is managed in-house or subcontracted.
We implement these systems end-to-end. First sends within 14 days.