At Marshland Software I run what I call a tech concierge service for small businesses. Think less "dev shop that builds whatever you spec out" and more "the technical person you never had on staff." I work closely with small businesses to figure out what they actually need, implement it, and keep it running.
The interesting side effect of that work is pattern recognition. You start noticing the same gaps across clients in the same industry. When a gap shows up consistently enough, it stops being a client problem and starts looking like a product opportunity. So I build SaaS products to fill those gaps — ones that can eventually scale well past any individual client relationship.
That's the model. And like most engineers who've shipped something real, I eventually hit the wall where building alone stops being the bottleneck.
Users are.
You can have the cleanest architecture in the world, but without actual people using your product you're really just guessing. You don't know what the market needs, which features matter, or how different customers want to interact with what you built. That feedback loop only opens once real people are in the system.
So I had to learn outreach. From scratch. As an engineer. It was uncomfortable.
What I'm actually doing
I'm targeting local service businesses for a current SaaS product. Think plumbers, HVAC shops, landscapers — hyper-local, owner-operated, Phoenix-area first. My goal is to get these businesses into a beta program, which means cold email is the primary acquisition channel right now.
This isn't a post about scale. I'm not doing millions in ARR. I'm one person, treating this like a real business, but watching every dollar.
Why cheap matters (and why cheap usually fails)
There's a reason people spend a lot on outreach tooling: the cheap path is full of traps that silently destroy your results.
The biggest one is reputation. Cold email is a numbers game, and you're going to land in spam sometimes. That's fine. What's not fine is burning your main domain. Once a domain gets flagged as a spam source, it's very hard to recover — and every email you send from it, including transactional product emails, suffers.
So the first rule: never send cold email from your primary domain.
The second trap is cheap inbox providers. I looked at Zoho Mail, for example. It's inexpensive, but it has a poor sender reputation. If you're going to do the work of writing campaigns, building lists, and setting up automation, you need your emails to actually land in inboxes. A bad provider burns all that effort before anyone reads a single word.
The third trap is skipping warmup. A brand new inbox with no email history looks exactly like what it is: a fresh account created to send spam. Email providers know this. You have to spend two to three weeks sending real-seeming email activity before you start campaigns. Most people skip this and wonder why their open rates are abysmal.
I know what skipping it costs because I did it right the first time, by hand. When I started, marshland.software was my sending domain and I warmed it manually: email a few friends, ask them to reply, have them fish my message out of spam, mark it not spam, respond. Repeat every day, increase volume a little each time, for about three weeks. It works. It also takes around 20 minutes a day of tedious, careful effort. Doing that for one inbox was enough to understand both why warmup matters and why I was never doing it by hand again at any larger scale.
Cheap but working means navigating all three of these without spending $200/month just to get started.
How I worked through this
I started by doing a lot of reading. There's a meaningful difference between "people who write about cold email" and "people who've run cold email at scale," and most of the advice online conflates them. The gotchas are real and worth researching before you buy anything.
My general approach was: find what the professionals use, then find the cheapest thing that gets you 80% of the way there. For some parts of the stack, the cheap option is genuinely fine. For others, cutting corners kills your results.
I also leaned on my existing strengths. I'm an engineer, so anything involving scraping, APIs, or custom automation I can just build. That changed the math significantly compared to what a non-technical founder would need to spend.
The stack
Cloudflare for domains. I use AWS for almost everything, but Route 53 costs $15/year for the domain plus $0.50/month for a hosted zone. Cloudflare is around $10/year for a .com and includes DNS. When you're buying multiple cold email domains (and you should be buying several), the savings add up. I grabbed a handful of variations on my brand — go[brand].com, try[brand].com, get[brand].com, that kind of thing.
Litmail.ai for Gmail inboxes. This one took some research. Google Workspace is the gold standard for inbox reputation, but $7/month per inbox gets expensive fast. Running 10 inboxes — which is reasonable for a real campaign — puts you at $70 to $140/month before you've sent a single email. Litmail is a contracted Google reseller, which means you get real Gmail inboxes with real Gmail reputation, for $37 for the first 10 inboxes and $3.50 per inbox after that. Roughly half the price.
Fair warning: Litmail has some rough edges. If you use Cloudflare (which you should), you can't point your nameservers to them, so you'll need to manually request all the DNS records and add them yourself. Avatar uploads and inbox connections also required back-and-forth with their support. The support was quick and helpful every time, but it's not a polished self-serve setup. For the price difference, I'll take it.
Smartlead for warmup and campaign management. Smartlead handles two things: inbox warming and campaign orchestration. The warming feature runs automated email exchanges between inboxes to build reputation before you start real outreach. The campaign side lets you manage multiple inboxes under one dashboard, which is table stakes when you're sending from 10 addresses at once. Plans start around $40/month and go up to $100. At my current scale, the lower tier covers everything I need.
There are other options in this space, but Smartlead had the best feature-to-price ratio for small-volume senders. The big players in this category charge for scale I don't have yet.
DataForSEO + Hunter.io for leads. DataForSEO has a business search API that's cheap and returns solid data: business name, category, location, phone, website. Hunter.io fills in the email field when the website doesn't have an obvious contact. For hyper-local businesses, Hunter's match rate isn't spectacular — these companies often just use a personal Gmail rather than a branded domain — but it's still useful as a second pass.
I built my own scraper on top of DataForSEO rather than paying for something like Apollo or Hunter's full platform. That's an engineer's shortcut, and it's not necessary if you don't want to build it yourself.
How it all fits together
The sequence is straightforward once the pieces are in place.
The domains point to Litmail, which gives you real Gmail inboxes. Those inboxes connect to Smartlead, which starts the warmup process immediately. Two to three weeks of that and you have sender reputation.
While warmup is running, the lead pipeline works in parallel. DataForSEO pulls business data on Phoenix-area service companies matching my target profile. My scraper visits each business's website to look for a direct email. If it finds one, great. If not, it hits Hunter's API for a fallback. The result is a CSV of enriched contacts that goes into Smartlead.
Campaigns run out of Smartlead, rotating across all ten inboxes so no single address carries the full sending load. When replies come back, Smartlead aggregates everything into one view.
I also built a small AI agent on AWS Lambda, deployed via SST, that sits in front of the reply inbox. Right now it's running in a man-in-the-middle mode: it drafts a response and surfaces it to me for approval before anything goes out. I'm not ready to fully trust automated replies yet, but the scaffolding is there. Once I've reviewed enough outputs to trust it, I'll flip it to autonomous.
The whole thing costs roughly $74/month in tooling: Cloudflare domains, Litmail inboxes, Smartlead on the starter plan, and API usage on DataForSEO and Hunter. The Lambda and infrastructure cost is essentially zero at this volume.
For a solo operator doing B2B outreach on local SMBs, this is a complete stack. It's not flashy, and a few pieces required some engineering work to glue together — but it covers the full funnel from lead generation to inbox management to reply handling without spending real money until the results justify it.
Marshland Newsletter
Building in public, one tool at a time.
Practical notes on SaaS, outreach, and the tooling decisions that actually matter — written for founders figuring it out as they go.