Start with the uncomfortable part. If nobody at your company has personally booked a meeting from a cold list into your ICP in the last six months, buying software will not produce one. It will produce volume, and volume against a wrong list is a faster way to burn a domain.
Send a hundred emails yourself first. If the reply rate clears 2%, you have a motion worth scaling and the question below becomes answerable. If it does not, the constraint is the offer and the tooling question is premature.
Once you have proof, buy in this order
- Sending first. Instantly or lemlist. Cheapest, and deliverability is the thing most likely to silently kill a small programme.
- Data second. Apollo bundles a database with sequencing, which is one bill and adequate quality. Clay if your ICP is narrow enough that match rates matter more than convenience.
- Coaching third. Lavender, if your reps write their own email and the problem is quality rather than volume.
What you do not need at this size is a full-stack agent on an annual contract. Those economics assume a proven motion and a territory large enough to exhaust. A small team has neither yet.