Build it yourself, and what it costs
You can assemble what CampaignStack does from tools you can buy today, and three different ways. This page prices each one for an agency, with the parts every estimate leaves out: proxies, the inbox poller, retry and dedup, the per-account budget, and the hours when a vendor changes something. The number is next to ours. Do the subtraction.
Every number on this page follows these four sliders.
Same job, three bills
Each one is a real stack agencies run. The wiring is the part the vendors' pages skip, so it is written out, and the bill follows the sliders above.
Gojiberry for signals, drafts and LinkedIn; Instantly for email; Zapier between them; Claude for review.
- Type the website into Gojiberry, pick the signals, connect two LinkedIn senders.
- A Zap moves Gojiberry's enriched leads into an Instantly campaign for the email steps.
- Copilot mode, or Claude over Gojiberry's MCP server, approves messages one by one.
- Running in an afternoon
- Signals and drafting are the vendor's problem
- Cheapest at one client with two senders
- Two senders per plan, one plan per client
- Email lives in another tool with its own memory of who replied
- No per-client reporting or exclusions across the two
- Gojiberry ProOne plan per client, one more per two senders$990
- Instantly HypergrowthEmail sending, one plan across client workspaces$97
- Zapier ProfessionalThe LinkedIn-to-email hand-off$20
- Claude ProOne seat per team member who reviews$60
- Upkeep, 12 hoursTwo hours per client a month to keep the hand-off and the exclusions in sync, plus two for the account$900
- Total$2,067
- CampaignStack, same sliders$1,620
The capability matrix
A tick is not a scoreboard; read the note. Where a stack relies on a piece you build, the cell says so rather than scoring it a miss.
| Agent + sequencer | Workbench stack | Build on APIs | CampaignStack | |
|---|---|---|---|---|
| Engagement signals on competitor posts | Gojiberry's 30+ signals include competitor engagement | Trigify watches posts and pushes engagers to Clay or HeyReach | An Apify actor scrapes commenters on a schedule; you decide what counts as a signal | Watchers on posts, people, companies and groups; engagement scored into the campaign |
| ICP scoring | The agent vets against an ICP it inferred from your site; no per-client scoring you can read | Clay formulas and Claygent, per table | Yours to write | Every lead scored against every ICP, plus signal weights per campaign |
| AI drafts with a critic | Drafts, no critic; copilot mode is you | Claude through the n8n MCP client drafts; a second pass is another node you build | Claude API; the critic is a prompt you maintain | Every draft passes a critic before a human sees it |
| Human review queue | Approve each message in the app, one at a time | A Slack approval step or a sheet, built in n8n | Airtable or a sheet, plus the code that resumes the send | Priority queue: approve, edit, retry, reject; the edits teach the workspace |
| Per-account LinkedIn budgets and warm-up | Managed by the vendor, two senders per plan | HeyReach manages limits per seat | Unipile sends what you ask; the daily and weekly caps are your code | Per-action daily budgets, weekly caps paced across the week, family-scoped auto-pause |
| Email sending | Enriches emails, does not sequence them; hand off to Instantly | Instantly, or HeyReach's own email steps | Instantly | Your own Gmail, budgeted and warmed, in the same workflow |
| Reply cancels the rest of the sequence | Inside Gojiberry's own sequence | HeyReach stops its sequence; the email sequence hears about it only if a Zap tells it | The poller and the cancellation are yours | One router across LinkedIn and email: a reply reroutes, a booking cancels |
| One workspace per client, isolated | One plan per client, no shared view | HeyReach has client workspaces; Clay and Trigify are shared tables | Your schema | Workspaces with members, exclusions and reporting per client |
| An agent can drive the whole thing | Gojiberry ships an MCP server | HeyReach has an API; Clay and Trigify through n8n | Everything is an API, nothing is a tool | Over 300 MCP tools, one key |
Five lines that are never on the quote
A LinkedIn account that sends from a datacenter IP, or from three tools' IPs in one day, is the one that gets the security check. One static residential address per account, rented monthly, is the floor.
Something has to notice the reply, on every account, every few minutes, and tell the sequence to stop. Sequencers stop their own sequence; the email tool next to it finds out only if you wired it.
The same person arrives twice under two URLs. The send failed at 2am. The scrape returned a commenter who opted out in March. Each is a rule, and the rules live in whichever tool forgot them first.
How many invitations this account sent this week, and whether today is a day to send more. The API stacks do not know; the sequencers know for themselves and not for the tool beside them.
Four vendors, four changelogs, one person who understands the Zap. Set the slider to your rate and the hours to what you have actually spent this month.
Keep the parts that work
Clay, Instantly, Apollo, HubSpot, n8n, Zapier and Make all connect to CampaignStack. The usual move is to keep the enrichment you like and let one system own the sending, the budgets and the review queue.
Frequently asked questions
Are these the cheapest possible versions of each stack?
No, they are the versions that do the job for an agency: several clients, several senders, email as well as LinkedIn, someone reviewing. Move the sliders to one client and one sender and every build gets cheaper, including ours.
Why count hours at all?
Because every agency that runs one of these stacks pays them, and none of them show up on a pricing page. Four vendors change something every month. Set the rate to zero if you want licenses only; the ranking barely moves.
Can Gojiberry send email?
It enriches email addresses through a waterfall of providers, but it does not run email sequences. The agent stack hands leads to Instantly through Zapier, which is why both appear in its bill.
Is the API build really that much work?
Unipile gives you send and read on a LinkedIn account, and nothing else. The budget per account, the pacing across the week, the poller that notices a reply, the dedup against people who opted out: each is a few days the first time and an hour a month forever after.
What if I already run the workbench stack?
Keep the parts that are working. Clay and n8n both connect to CampaignStack, in both directions, and the guides for each are linked below. The usual move is to keep Clay for enrichment and let one system own the sending, the budgets and the review queue.
Every vendor price on this page was read on the vendor's own pricing page on the date in the header:
Get started
Start building your stack.
Your LinkedIn accounts will be safer with CampaignStack than doing it by hand. That's not a pitch. It's a measurable claim.