Buy help for the gap you actually have.
If you know what to say but struggle to publish reliably, you may need a writer or an internal workflow, not a full-service agency. If you need audience research, launch planning and someone to handle replies, ask who will do each job. Treat paid ads and creator partnerships as separate lines of work unless the proposal explicitly includes them.
Give each finalist one audience, one offer, one desired action and the same constraints. Ask: ‘What would you do first, what would you leave out, and what do you need from us before you can start?’ A proposal built around post volume alone does not answer that question.
Ask for work you can inspect.
Request a public post, the original brief or a redacted equivalent, and an explanation of what the agency owned. Ask: ‘Who chose the angle, who approved the claims, and who handled distribution and replies?’ If the agency cites results, ask for the date range, metric definition, source and what else was running at the time.
For example, TrendBTC’s public tournament announcement on X gives readers steps to register and explains how ranking works. You can assess how the post turns an announcement into a clear next action. It does not establish who drafted or distributed it, how many people registered because of it, or whether it produced incremental results. XAMM’s case is an editorial reading of public posts, not independent evidence of campaign lift.
Agree who can post, pause and leave.
Ask: ‘Who writes the copy, who checks product claims, who can publish, and who can pause the queue when information changes?’ Keep account ownership and recovery with your team. Agree which access the agency needs, who approves exceptions, where drafts and reports live, and what you receive at handover.
Ask for fees broken out by strategy, production, community coverage, media spend and any third-party work. Put revision rounds, launch-day response coverage and cancellation terms in writing. Ask how partners identify their involvement; rule out purchased engagement and undisclosed accounts posing as independent customers.
Hire for the work you can inspect and the decision you can make at the end of a pilot.
Agree on the pilot before anyone posts.
Use a fixed-fee pilot for one audience and one offer. Set the review date, approval turnaround and deliverables up front. Decide what counts as a qualified enquiry before the first post. A useful weekly report puts post links, tagged visits and qualified enquiries next to one another without treating them as interchangeable.
Ask: ‘Can we walk through one draft from brief to approval and one enquiry from first contact to qualification?’ Agree how paid traffic, other campaigns and uncertain sources will be recorded. A short pilot can show whether the team delivers and learns; it rarely isolates how much demand the agency caused.
Worked example: decide against the retainer.
Suppose a B2B scheduling tool wants demo requests from operations leads who manage shift handoffs. The buyer commissions a four-week pilot on its brand account: a launch explainer, a workflow walkthrough and an objection-led follow-up. Sales qualifies requests against the buyer and use case in the brief. The marketing team uses tagged links, checks attribution with Sales and keeps final approval.
In this hypothetical review, the agency delivers on time and the walkthrough draws specific questions from operations leads, but Sales cannot verify any pilot-sourced qualified demo request. The buyer does not sign the retainer. It asks for a narrowly scoped follow-up to test a demo page that shows the same handoff workflow and fixes the attribution gap. If the next review still cannot establish qualified demand under the agreed rules, the buyer stops rather than paying for more posts.
Make the decision from the evidence.
Continue when the team delivered what it promised, respected your approvals and produced evidence that meets the decision rule you agreed to. Revise when execution is sound and the evidence points to a specific problem you can test. Stop when account access, claim review or reporting is unreliable, or when there is no plausible next test worth paying for.
Ask each finalist to explain what would make them recommend against renewing. Compare that answer with the scope, source material and handover plan they put in writing. A credible partner can tell you what the pilot will teach you even if the campaign falls short.
Compare finalists on the same questions.
| Check | Ask the agency | What to ask to see |
|---|---|---|
| Scope | What is included, and what costs extra? | Itemized proposal and named owners |
| Evidence | What did you own on this campaign? | Public posts, a redacted brief and a dated report |
| Access | Who approves posts and who can pause the queue? | Approval workflow and account handover plan |
| Measurement | What separates a click from a qualified enquiry? | Sample report and written qualification criteria |
| Renewal | What would make you recommend we stop? | Written pilot decision rules and an exit date |
Sources and further reading
Take this with you.
- The same brief for each finalist
- Work samples with a clear account of who did what
- Written approvals, account ownership and exit terms
- A fixed-scope pilot with a review date and decision rule
A little more context.
What should an X marketing agency include in a proposal?
Ask for named responsibilities, deliverables and exclusions, separate fees for third-party costs, the approval and access plan, sample reporting, and an end date with handover terms.
Should I choose the agency that guarantees the most followers?
No. A follower total does not show whether you reached likely buyers or met a business goal. Start by checking the work and test the outcome that matters to your campaign.