The buyer question in most first calls is about the rate. The question that actually decides is different: what is on your side of the table the day the engagement ends?
Rate is comparable across routes. The end state is not. An agency retainer, a freelancer and one engineer doing the whole path can all quote a monthly number; only one of them leaves you holding files.
The three routes, priced the same way
| Agency | Freelancer | Marketing engineer | |
|---|---|---|---|
| Live pages on your domain | Usually, licensing varies | Sometimes | Yes — your hosting, your files |
| Measurement setup | Their dashboard | Whatever they wired | Events + one report per page, documented |
| Who owns the outcome | An account layer | One specialist skill | One person, start to finish |
| What you pay for | Retainer + coordination + margin | Hours | Fixed scope, fixed price |
| Evidence the work happened | Report | Deliverables | Live URLs, open in front of you |
| Exit cost | Notice period, handover project | Depends who holds the keys | 14 days notice on the monthly loop |
Read the last row first. It is the row that costs the most and appears in no proposal.
The dropped step is the real expense
A funnel is not one job. It is offer, page, indexing, distribution, measurement — in that order, each one reading the output of the last. Split across vendors, the handoffs are where the money leaks:
- The copywriter writes for a page nobody measures.
- The ads person buys traffic to a page that was never indexed.
- The dashboard reports sessions for a channel that was never tied to a reply.
Each vendor ships their piece and is right about it. Nobody owns the seam. When you hire one person for the whole path, the seam becomes an internal detail instead of a weekly call.
What you can actually test before committing
You do not have to believe any of this at contract time. Insist on the first artifact being small and checkable:
- A defined readable unit. Not "leads" — the event that decides whether the channel continues (a booked call, an activated workspace, a first successful API call).
- A volume floor. A test that cannot reach its sample size is a coin flip with a report attached. Count how many contacts or sessions a verdict needs at your volume before you spend.
- A stop rule written before the spend. What result, on what date, ends it. If the rule is written after the data arrives, it is not a rule.
- Every address verified before the first send. The order matters more than the tooling: verification before send rather than after the bounce is the difference between a 2–5% bounce rate and the 10–35% a static list gives you.
If a route cannot produce those four things inside its first month, the rate was never the problem.
Why the end state is worth paying for
The practical difference shows up when you want to change something. With files on your domain and a documented measurement layer, you can fire the person and keep the system. You can hand the page to a new writer, point the events at a new tool, or run the next test yourself.
With a report and a renewal, the system stays where it was built.
Full comparison, including the price list this article reads from: packages and pricing — the same page carries acceptance criteria and the artifact list.
If you want the outbound and lead-generation path run as one job — segment, contacts verified over SMTP before the first send, touches, replies worked until a meeting — the first test is 100 verified contacts, so you judge the list before committing to a month. Packages start at $900: how to work with me.
What did you keep the last time an agency engagement ended?
Top comments (0)