Outbound for seed startups, in 90 days
A seed team can run outbound for under $150 a month, and the first 30 days produce no meetings, by construction. This is outbound to buyers, not investors: four phases, the stack at August 2026 vendor prices, the volume arithmetic, and where the honest answer is don't. Nobody here will sell you a meeting count.
By Kshitij Maheshwari, co-founder · Updated August 2026 · 14 min read
The first 90 days, and what each phase costs
Four phases, each with a price you can verify and a checkpoint you can observe. Nothing in the last column is an outcome we promise you.
| Days | What you build | What it costs | The checkpoint that says move on |
|---|---|---|---|
| 1 to 14 | Two sending domains, ten mailboxes, SPF, DKIM and DMARC set, warmup running. | About $28 a year in domains, about $30 a month in mailbox slots. | Authentication passes on a test send, warmup has run 14 days, nothing has gone out cold. |
| 15 to 30 | One ICP slice, 200 to 400 verified contacts, one offer, a four-step sequence. | Add the sending platform, about $39 a month. | The list verifies clean, and the offer is one sentence a stranger repeats back to you. |
| 31 to 60 | Send 20 to 30 per mailbox a day, and answer every reply the same day. | Add data and verification, about $49 to $99 a month. | Enough contacts through the whole sequence to read the slice, bounces under 2%, complaints at zero. |
| 61 to 90 | Read the market: keep the slice, change the slice, change the offer, or stop. | Flat. Nothing new to buy. | You can name who replied, what they said, and which no you kept hearing. |
Prices are the vendors' own published list prices, checked August 2026: Mailforge for domains and mailboxes, Smartlead for sending, Apollo and Findymail for data. Authentication requirements are Google's and Yahoo's own sender guidelines.
The three ceilings on a seed outbound motion
Your volume is set by whichever of three ceilings is lowest, and sending capacity is not one of them.
A ceiling is the input that runs out first. Outbound at seed has three: the companies that genuinely fit, the contacts you can research properly, and the replies you can answer the same day.
How many companies genuinely fit, exist and are reachable. If that number is 400, one pass through them is your whole program, not your first month. Working the other way, from a revenue target down to the meetings it needs, is a coverage calculation.
How many contacts one founder can put a real reason in front of per week. This one is measured in hours, and it does not move when you spend more.
How many replies you can answer well the same day. A reply you get to on Thursday is a reply you wasted, and no tool buys that back.
Sending capacity is the cheapest input in outbound. It is also the first one founders buy more of.
Work backwards from meetings, not forwards from sends
Pick the meetings you want, then divide backwards. The three rates on these cards are dials you set, not benchmarks anyone is quoting at you.
5 held meetings a month
Pick a number you could actually service this quarter. Five held is a real seed target and a hard one.
Two in three booked get held
So five held needs about eight booked. Replace this with your own no-show rate the moment you have one.
One in three replies is worth a call
So eight booked needs about 23 replies. Most replies are a no, and a fast no is still worth having.
3% of contacts reply, per contact
So 23 replies needs about 750 contacts a month. At 2% it is 1,125. At 4% it is 563.
Illustrative arithmetic, not a client result. Every rate above is yours to set; the touch-by-touch sequence lives in the founder-led sales play.
750 contacts through a four-step sequence is roughly 2,550 emails, about 128 on a working day, which is six mailboxes at 25 a day or two at a warmed vendor's number. The sends were never the hard part. Finding 750 companies worth writing to was.
The whole stack, named and priced
Five line items at the vendors' own August 2026 list prices. The lean build is about $69 a month plus $28 a year in domains, and about $118 once you add a data seat.
Two, and never your main one
Mailforge lists .com domains at $14 a year, charged once. Two is enough to start, and buying them is the cheapest decision on this page that you cannot reverse later.
Ten slots, three dollars each
Mailforge sells mailbox slots at $3 a month billed monthly with a ten-slot minimum, so $30 a month, or about $24 billed yearly. Google Workspace Starter is €6.80 per user per month if you would rather stay inside Google.
The platform that sends
Smartlead's Base plan is $39 a month for 2,000 contacts and 6,000 sends, mailbox connections unlimited. Instantly's Growth plan is $47. The rest are in our cold email tools roundup.
Where the contacts come from
Apollo Basic is $49 per seat per month billed annually. Findymail Starter is $99 a month and charges only for verified results. Both sit in our B2B data tools roundup.
Cheap, and not optional
MillionVerifier sells 50,000 verifications for $89 and says the credits never expire. Bounce rate is the one deliverability number a seed team can actually see, so protect it.
What you can skip at seed
Clay's Launch plan starts at $167 a month billed annually ($185 month to month), more than the sender and the data seat put together. Buy orchestration when your triage overflows, not before it does.
None of this is GTM engineering yet. It becomes that the day these parts have to talk to each other without you in the middle.
Which reply rate numbers are real, and per what
A rate per email sent and a rate per contact differ by roughly the length of your sequence. On a four-step sequence that is about four times, which is enough to make the same campaign look good or broken.
"The average cold email reply rate is 3.43%. We're sitting at 2%, so we're below average."
- ✕Compares a per-lead dashboard to a per-email benchmark
- ✕Nobody checked which one the report measured
- ✕Two denominators, quoted as though they were one
"All replies received, including follow-up responses, divided by total emails sent." Instantly's own definition of the 3.43%.
- ✓It is published, in the report's methodology
- ✓Hunter is also per email, net of bounces
- ✓lemlist's campaign overview counts each lead once
Instantly's 2026 report, 3.43%
Instantly's Cold Email Benchmark Report 2026 is where the number comes from: vendor data from its own workspaces, 1 January to 18 December 2025. The methodology defines the rate. Almost nobody quoting it reads that far.
What it actually measured
Instantly's methodology defines it: all replies received, follow-ups included, divided by total emails sent. Hunter's cold email guide publishes the same shape, replies over emails sent minus bounces. Both are per email.
Your dashboard may disagree
lemlist documents that in its campaign overview each lead is counted once however many steps they get, so the rates read against unique leads, not message volume. Same word, different denominator.
Do the multiplication yourself
A four-step sequence with normal drop-off sends roughly 3.4 emails per contact. Three replies from 100 contacts is 3% per contact and 0.88% per email. Both are true, and only one of them is comparable to 3.43%.
Founder-sent emails reply better
This one always arrives with a precise-sounding number and no study anyone can trace. Notice the direction too: it is a claim about who sends, and a reply rate is recorded against who receives.
The serious benchmarks all publish their denominator. The mistake is comparing yours to one without converting first.
Would you rather have the two of us run these 90 days?
Book a Fit CheckThe deliverability floor
You are not a bulk sender, so most of what you have been warned about does not bind you. Two things do, and one of them is invisible at your volume.
Send cold from your main domain
MX, SPF and DKIM on yourcompany.com, and the cold sequence goes out from there too.
- ✕Reputation attaches to the domain, not the campaign
- ✕Jumping to 150 a mailbox on a vendor's say-so
- ✕Chasing a spam rate Google will not show you
Separate domains, slow ramp
SPF and DKIM on every sending domain, DMARC at p=none, one-click unsubscribe, 20 to 30 a mailbox a day.
- ✓Warm two weeks before the first cold send
- ✓Keep bounces under 2%, because you can see them
- ✓Run a placement test before every ramp step
Every threshold you keep reading about is 5,000 messages a day, per provider: Google's bulk sender rules from February 2024 (Yahoo's rules landed the same month but name no volume), and Microsoft's Outlook requirements for domains sending more than 5,000 a day, enforced from May 2025. A few thousand a month is nowhere near any of them.
Sources: Google's email sender guidelines; Yahoo's sender best practices; Microsoft's Outlook requirements for high-volume senders; Google Postmaster Tools help; the Amazon SES Developer Guide. All checked August 2026.
Founder, SDR, or agency
Do it yourself until the motion is written down. The other two options run a playbook; neither of them finds one.
Founder-led
Costs hours, not money: roughly 10 to 15 a week at 750 contacts a month. Right from day one, because you are the only person who can hear a no and change the offer that same afternoon. It breaks when prospecting starts eating your closing time. The founder-led sales play has the sequence.
Your first hire
The Bridge Group's 2025 SDR report puts median on-target earnings at $80,000, average ramp at 3.0 months and median annual attrition at 40%, across 351 B2B companies whose median revenue is $47M. That is not a seed sample, and three months of ramp is not a discovery budget.
Ours, disclosed
We are one, so weigh this accordingly. Our pricing starts at 4k euros a month, the two founders run it, and the motion stays yours. We do not do cold calling, pay-per-lead, or guaranteed meeting counts. Anyone quoting you a meeting number is quoting you a fiction.
The long version of this decision, with the costs laid out on both sides, is in agency versus in-house.
- 1 Run it yourself until the motion is written down on paper.
- 2 An SDR runs a playbook. An SDR does not invent one.
- 3 Hours are the real budget line, not the $118 of tooling.
- 4 A guaranteed meeting count is a sales tactic, not a forecast.
When outbound is the wrong motion at seed
Four cases where the arithmetic says wait, and one thing that is not a reason to.
If nobody can repeat your offer back in one sentence, outbound will tell you that slowly and at full price. Fix the offer first.
If 400 accounts is the whole market, one pass at 3% per contact is about 12 replies and the list is gone. Outbound needs a list you can re-enter with a new reason.
A cold reply becomes a call, and a call needs something to sell. If the honest answer is a waitlist, spend the hours on the product instead.
At 3% per contact, a held meeting costs about 150 researched contacts. At five minutes each that is roughly 12 founder hours. If a meeting is worth less, wait.
Having no logos yet is not on this list. Nobody's first ten customers arrived with a case study attached.
Never send cold from your primary domain
Authentication, reputation and the spam rate all attach to the sending domain. Burn it on one campaign and you have burned your investor updates, your support mail and your billing receipts along with it.
What a seed team can watch for free
LinkedIn saved searches, a funding digest, the careers pages on your list, and whoever visits your own site. None of it costs anything. Getting the timing right is a separate question, and it is answered properly here.
Buying signal timing windows
How long each signal stays warm, with the sources. Every window figure this playbook would need lives there, not here.
Read the postThe signal library
Job changes, funding, hiring, intent: what each one is, where to watch it free, and when it misleads you.
Browse the signalsSignal-based selling
The worldview underneath all of it: why a reason to write beats a title match, and how the motion gets built.
Read the guideHow we run a seed ramp
Three sequencing choices we make on every seed ramp, and the reasoning behind each. This is how we run the motion, not a measured result.
We ramp at 20 to 30 a mailbox rather than 150, because the downside is a domain and the saving is a few dollars. One slice, not four, so a no means something. Replies get answered the same day, always.
The 20 to 30 comes out of a disagreement between two vendors. Smartlead's own guidance says a mailbox warmed a fortnight can safely run 100 to 150 a day; Instantly's says 20 to 30 in week one and under 50 for two weeks. Same product category, four times apart, both labeled safe. We take the low one and ramp from there.
Questions founders ask
How much does outbound cost a seed startup per month?
How many emails should a seed startup send a day?
How many meetings will I get in the first 90 days?
Should I use my company domain for cold email?
Do the Google and Yahoo bulk sender rules apply to me?
What reply rate should I expect at seed?
Co-founder of Real Good GTM. He has been the first business hire and Chief of Staff at seed-stage B2B startups, building outbound pipeline from before there was a playbook to copy. This is the version of that first 90 days he wishes someone had handed him, with the prices attached.
Connect on LinkedInThe three pages this playbook leans on
The offer that has to exist first, the read you are running the 90 days for, and how to design one slice test.
Build the offer first
Outbound cannot rescue an offer nobody can repeat back. How to write one worth a stranger's attention.
Read the guideTurn the replies into a read
Day 90 is a read, not a meeting count. What the replies tell you about your market, and how to write it down.
Read the guideThe ICP slice experiment
How to design one slice test, size it honestly, and read the result without fooling yourself.
See the playWant the 90 days run by the two of us?
Book a fit check. We'll size your list, name the slice worth starting on, and tell you straight if outbound is not your motion yet.
Book a Fit CheckNo hard sell. No fake numbers. Real good work speaks for itself.