Bring a real account, not a generic prompt
Ask each vendor to work through one account in your target market. Show the same ideal customer profile, offer, exclusions, and success criteria to each. Have the vendor explain why that account and person were selected, what information was used, and what the message would say. A polished sample email is easy to prepare; a repeatable research and review process is the harder test.
Ask what happens when the available information is thin, stale, or wrong. The answer should describe how the team reviews and corrects output rather than quietly filling gaps with invented details.
Inspect the whole conversation
A first email is only one part of outbound. Test an interested reply, a polite objection, a referral to another person, and a request to stop contacting the account. Ask when the system responds, when your sales team steps in, and how prospect information and conversation history reach sales. See the lead information your salesperson would receive.
For Fully Managed clients, Silenium sets up, trains, and tests bots with the customer. Self-Service teams set up, train, and test their bots using Bot University. Testing and validation are guided by people.
Understand Who Manages What
Write down who chooses targets, configures the campaign, reviews messages, monitors email and domain health, handles exceptions, and changes strategy. Self-Service requires your team to manage campaign setup and ongoing changes in-house. Fully Managed provides a team to build, launch, monitor, and manage campaigns with you.
Silenium offers Self-Service and Fully Managed options. The comparison page explains who manages each part of the campaign and shows the public pricing.
Compare the full cost and the right denominator
Ask for the subscription, onboarding, data, sending infrastructure, domain work, support, usage limits, and commitment in one view. Also estimate your team's time to configure and supervise the program. Compare like with like: a contact researched, an email sent, a reply, a qualified conversation, a meeting booked, and a meeting held are different units.
Use a baseline from your own market. Calculate response rate from responses divided by sent emails, and qualified conversation rate from qualified conversations divided by contacted people or another clearly defined denominator. Then trace those conversations through meetings and customers. Our Metrics Library defines the measures, and the planning calculator lets you change assumptions without treating a projection as a promise.
Run a short, reviewable pilot
Agree on the ICP, exclusions, message approval process, volume, success criteria, and owner for each exception before launch. Review samples of research and copy, then inspect replies and lead delivery weekly. A low response rate might point to targeting, deliverability, the offer, or the message; a high response rate alone does not prove pipeline quality.
Ask vendors for referenceable customers and documented results that match your market. Use documented customer results, not sample dashboard data, when evaluating performance. Silenium's three-month technology company case study provides one example with a defined scope and reported figures.