Agency sourcing looks like in-house sourcing and pays completely differently. In-house, saved recruiter hours mostly buy relief. In an agency, a desk's hours are the unit of production: fewer hours per placement means more placements per consultant, and placement fees are 15–25% of salary. That changes which tools are worth buying and how aggressively.
It also imposes constraints in-house teams don't have: two customers (client and candidate), business development competing for the same hours as delivery, and a CRM that is the firm's actual asset. A stack that ignores those constraints gets bought, ignored by consultants, and quietly cancelled at renewal.
The four layers
| Layer | Job | Common choices | Est. annual cost |
|---|---|---|---|
| 1. CRM / ATS | System of record for candidates, clients, and activity | Bullhorn, Loxo, Recruit CRM, Manatal | $1K–$15K |
| 2. Discovery | Finding candidates outside your database | hireEZ, SeekOut, Juicebox, AmazingHiring | $5K–$25K |
| 3. Engagement | Multi-channel sequences for candidates and BD, with activity write-back | SourceWhale, Interseller, Gem | $5K–$20K |
| 4. Automation | Running whole searches: screening and outreach without a consultant driving | Noon, HeroHunt.ai, Fetcher | $1K–$15K |
Most agencies have layers 1 and 2 and treat 3 as optional and 4 as futuristic. In our view that ordering is backwards for the majority of desks: layer 2 is the most commoditised (the underlying indexes are similar), while layers 3 and 4 are where consultant hours actually leak. This is why our agency ranking puts an agent first and an engagement platform second, with databases behind them.
The agency arithmetic
Work it through for one desk. Assumptions on the table, and you should substitute yours: a consultant fully loaded at ~$55/hour of capacity, 15–25 hours of sourcing per filled role, and a placement fee of $18K on a $100K salary at 18%.
| Line | Manual sourcing | With automation layer |
|---|---|---|
| Sourcing hours per role | 15–25 | 3–6 |
| Hours freed per role | — | ~12–19 |
| Roles worked per consultant per month | 4–6 | 7–10 (capacity, not guarantee) |
| Hours freed per month | — | ~48–114 |
| Value if converted to one extra placement/month | — | ~$18K fee |
| Tooling cost for the layer | — | $1K–$15K/year |
The honest caveat: freed hours only become fees if there is client demand to absorb them. In a slow market the same hours become bench time, and the automation layer is a cost. That is the real question to ask before buying — is your constraint delivery capacity or client demand? If it is demand, spend on BD tooling in layer 3 instead, and note that the same engagement platforms run BD sequences as well as candidate outreach.
CRM hygiene is the integration requirement that matters
Agencies live or die on the CRM being a true record. The failure pattern is predictable: a tool sends outreach from its own inbox, the activity never lands in Bullhorn, consultants either double-enter or stop logging, and within a quarter nobody trusts the candidate record. The firm's asset degrades.
Ask these five questions of any layer-2, -3, or -4 vendor before signing:
1. What writes back, automatically? Emails, calls, InMails, replies, stage changes, notes. "Profile push" is not write-back.
2. Is it two-way? If a consultant updates a stage in the CRM, does the tool respect it — or keep sequencing someone who has already been placed?
3. How does it handle duplicates? Deduplication rules against existing records, or new records for everyone it finds?
4. Which CRM version and edition? Integration depth varies by edition and API tier; get the specific answer for yours, in writing.
5. What happens to sourced candidates if we cancel? Export rights on candidates and activity, in the contract.
This is the reason SourceWhale rates highly with agencies despite not finding candidates itself — its Bullhorn and Vincere activity logging is the cleanest in the category. Depth of write-back beats breadth of features here.
Three stacks that work
Solo or boutique (1–3 consultants, generalist). An affordable CRM with sourcing built in (Loxo or Recruit CRM), plus a contact-data tool, plus a self-serve agent for the roles you repeat. Est. $3K–$8K total. Avoid enterprise databases at this size; the seat minimums alone will exceed your tooling budget.
Mid-size, volume desks (5–20 consultants). Bullhorn as the record, an engagement platform for sequences and BD, and an automation layer for definable roles. Keep one or two database seats for the power sourcers rather than buying seats for everyone — consumption is heavily concentrated in practice. Est. $20K–$50K.
Technical or specialist search. Depth beats breadth: AmazingHiring or SeekOut for engineering signal, plus engagement, plus a small database for coverage. Automation helps less here because the qualified population is small and relationships matter more than throughput — see sourcing hard-to-fill roles.
Four agency-specific pitfalls
Buying seats for everyone. Database usage concentrates in a few heavy users. Start with the actual sourcers, then expand — and audit logins at renewal, since dormant seats are the most common wasted line item we see.
Sequences that outrun your brand. Agencies send more outreach than in-house teams, so degradation compounds. Cap daily sends per consultant, and read a random sample of what actually goes out monthly. Reply rate falling while volume rises is the early warning; our benchmarks give you the levers.
Ignoring your own database. The cheapest candidates you will ever source are already in your CRM — placed contractors, silver medallists, applicants from 18 months ago. Rediscovery has no marginal licence cost and near-zero acquisition cost. Most agencies we look at have tens of thousands of records and no systematic re-engagement motion; the rediscovery guide covers how to run one.
Buying discovery when the bottleneck is qualification. If consultants have lists and no time to screen and message, another database makes the problem worse. Measure where the hours go for two weeks before you buy — the diagnostic in our metrics guide takes an afternoon and routinely changes the purchase.
Where to start
If you are rebuilding the stack, start with the ranking that matches your model — staffing agencies or recruiting agencies — and read the two head-to-heads agencies most often run: Loxo vs Bullhorn for the record layer and SourceWhale vs Interseller for engagement. For the automation layer, Noon vs hireEZ is the clearest illustration of the agent-versus-database tradeoff, and cost estimates for every tool are in the pricing benchmark.