Two philosophies, one channel
Cold email tooling has split into two camps. Volume-first platforms, the category represented by tools like Instantly, Smartlead, and Lemlist, are built to deliver large numbers of emails reliably: many mailboxes, rotation, warmup infrastructure, sequencing at scale. Research-first platforms like Furet are built to make each individual email worth sending, even if that means sending far fewer.
Neither camp is wrong. They optimize different variables, and the honest question for an agency is not "which tool is best" but "which variable is the bottleneck for our ICP." This comparison is about the categories; specific features vary by product and change monthly.
What volume-first tools do well
The volume-first model assumes you bring your own lead list, usually exported from a B2B database, and its job is throughput: thousands of sends per month spread across many inboxes, with deliverability infrastructure to keep that volume landing.
This genuinely works when three conditions hold:
- Your ICP is broad and homogeneous.If you sell to "B2B SaaS companies with 20-200 employees," prospects are similar enough that one well-tested message can be relevant to most of the list.
- You already have good list sources and the operations muscle to clean, verify, and segment them.
- You can absorb the infrastructure overhead: multiple sending domains and mailboxes, monitoring, and the ongoing risk management that high volume requires.
The trade-off is built into the model: when relevance per email is low, you compensate with quantity, and quantity is exactly what mailbox providers now police hardest.
What research-first does well
Furet collapses the pipeline into one system: it sources prospects from Google Maps, fetches and analyzes each prospect's actual website with an LLM to find concrete gaps, verifies the contact email before sending, and composes a prospect-specific message that must reference at least two real findings. Drafts are quality-scored 0 to 1, and only those above 0.80 go out automatically.
This model fits when the opposite conditions hold:
- Your ICP is local or SMB, where no two prospects look alike and a generic message is visibly generic. A dentist, a roofer, and a boutique law firm cannot receive the same email and believe it was written for them.
- You sell what the research finds.For marketing, web design, and SEO agencies, the prospect's website is both the lead source and the sales argument.
- You would rather send 50 strong emails a day than 500 weak ones.Furet's defaults (50 sends per day per campaign, 10 per hour, three touches maximum) assume quality is the lever, not volume.
The trade-off mirrors the other camp's: research costs more per prospect than a CSV row does, so this approach makes no sense for ICPs where a broad template performs fine.
Side by side
| Dimension | Volume-first tools | Research-first (Furet) |
|---|---|---|
| Lead sourcing | Bring your own list (databases, exports, scrapes) | Built in, sourced from Google Maps by category and area |
| Personalization depth | Merge fields and segment-level copy variants | Per-prospect website analysis; two-plus real findings per email |
| Sends needed per reply | Roughly 100-300 at typical sub-1% template reply rates | Roughly 17-50 at 2-6% personalized reply rates |
| Domain risk | Higher: volume is the model, managed with many domains and inboxes | Lower: capped volume, pre-send verification, quality gating |
| Setup effort | High: lists, segments, copy testing, multi-inbox infrastructure | Low: pick a category and geography, review early drafts |
| Best for | Broad SaaS and mid-market ICPs with proven messaging | Local and SMB ICPs; agencies selling web, SEO, and marketing services |
The "sends per reply" row deserves a caveat: these are category-typical ranges, not guarantees, and a great operator can beat them in either camp. The structural point stands, though. When each email carries more relevance, you need fewer of them, and fewer sends means less infrastructure and less reputation exposure.
The cost question
Volume-first pricing tends to scale with inboxes and sending volume, plus the separate cost of lead data, which is easy to forget when comparing sticker prices. Research-first pricing scales with research: Furet uses credits, where one credit covers a Gemini-researched prospect and five credits a Claude-researched one, with plans from a free tier (10 researched prospects, 5 sends) up to $799/mo for 10,000 prospects and 40,000 sends. Full details are on the pricing page.
The comparison worth doing is not tool price versus tool price but cost per qualified reply, all-in: subscription, lead data, inbox infrastructure, and the hours your team spends on list operations. Agencies are often surprised by which model wins once their own labor is on the invoice.
Use both when it makes sense
The two camps are not mutually exclusive, and mature agencies often run both. Sensible splits:
- By ICP: volume-first for a broad SaaS or ecommerce segment where one message fits thousands; research-first for local and SMB segments where every prospect is visibly different.
- By offer: volume for a low-commitment top-of-funnel offer (a free audit, a benchmark report); research-first when the first email has to justify a $2,000/mo retainer conversation.
- By client, if you run outreach as a service: match the model to each client's market rather than forcing one tool across the book.
If you are choosing your first system and your clients are local businesses or SMBs, start research-first; the setup is lighter and the failure mode (too few sends) is recoverable, while the volume-first failure mode (a burned domain) is not. If you are already running a tuned volume operation with healthy reply rates, keep it, and consider research-first for the segments where your template numbers sag.