ATS for a small recruitment agency (2026): the two numbers no listicle prints

There are maybe forty pages ranking the best ATS for a small recruitment agency, and they are all built the same way. A table of features, a paragraph of praise per vendor, a per-user price copied from somewhere, and a verdict that changes depending on who paid for the placement.

None of them answer the two questions that actually decide it when you are three people with no ops person and no procurement department.

The first: what does your team actually pay to get in? Not the sticker price per user. The real floor, which is set by how many seats you have, whether the plan you need is monthly or annual only, which tier your headcount drops you into, and how many of the things in the demo are separate line items.

The second: what does it cost to get out? Your candidate base is the only asset in this business that compounds. Whoever holds it holds your switching cost. That number never appears in a comparison table, and it is the one that will cost you the most over five years.

This page is those two numbers, product by product, with prices read off vendor pricing pages during the week it was published. It also contains the thing the roundups will not tell you, which is that the map flipped this year: the vendor everyone still calls the opaque enterprise option now publishes a price with free implementation, while several products sold as small-agency-friendly put the part you are actually buying behind a quote or behind a seat band you do not qualify for.

Contents

• Quick summary
• Why should you listen to us?
• Step one: find your headcount on the tier chart
• What three seats actually costs
• The floor is built from four things
• The 2026 flip: published enterprise, quoted SMB
• The second number: what it costs to leave
• Six questions to get answered in writing
• When a small agency does not need an ATS at all
• Buy carefully, because the window shuts for a year
• Where Gemsift fits, honestly
• FAQ

Quick summary

• The per-user price is not your price. Your seat count decides which tier you are allowed to buy, and the tier decides what you get. Crelate publishes Essentials at $85 per user per month on annual billing, but it covers up to 2 users, so hiring a third recruiter moves you to Business at $119 per user. JobAdder sorts plans by headcount, and its AI features start at Essential, which begins at 6 users. A three person agency is in Lite, where they are not included.
• Three seats, first year, from published numbers as of August 2026: Manatal from about $540, Zoho Recruit staffing Standard about $1,080, TE Recruit Standard about $2,700, Recruit CRM Pro about $3,420 before add-ons, Bullhorn Starter about $3,564 with implementation free, Loxo Core and Recruiterflow Platform about $5,364 each. Gemsift Agency is $299 a month flat for 3 seats, so about $3,588, and the free tier is $0.
• The transparency story inverted. Bullhorn, the vendor small agencies are told to avoid on price, now publishes Starter at $99 and Core at $165 per user per month with free implementation on both. Recruiterflow publishes $149 for the platform and quotes separately for the AIRA plan where its AI agents live. JobAdder publishes no prices at all.
• Watch the à la carte column. Recruit CRM lists Pro at $95, Business at $135 and Enterprise at $215 per user per month, with a footnote that enterprise features are billed separately, then publishes them individually: workflow automation from $41 a month, job multiposting from $20, data enrichment at $250 a year, LinkedIn messaging at $12 per user, analytics at $25 per user. Crelate marks contact enrichment, email campaigns, texting and e-signature with "additional fee applies".
• Migration is a real project, not a checkbox. Manatal's own documentation puts an executed migration at 2 to 4 weeks after they receive your data. Recruit CRM offers free self-serve CSV import or paid migration engineers, priced by volume, taking 10 to 14 business days. Recruiterflow offers white-glove migration at a paid rate with the fee sometimes waived on multi-year contracts. Loxo gives free migration to Core teams of 3 or more seats and self-serve tooling below that.
• Getting out is the number to ask about. Almost every vendor will export a candidate list. The question is whether you also get the notes, the activity history, the submissions and the actual resume files, in a format another system can read, at no extra cost. Get the answer in writing before you sign, not after.
• You may not need one yet. One recruiter, one vertical, one inbox, a handful of live roles: a spreadsheet is still the correct answer and every hour configuring an ATS is an hour not billed. The trigger is not growth, it is the first time two people need the same candidate history at the same time, or the first time you cannot find someone you know you have.
• Gemsift is one of the options here and I build it. It is the AI-native ATS you move into with one export file, and it re-reads your whole base against every new role with a plain English reason per pick. Free, $99, $299, and from $899 a month. Full structured export on every plan including free. It is not the right answer if you need deep integrations, job board posting or enterprise compliance.

Why should you listen to us?

I build one of the products on this page, so read it as informed and interested rather than neutral, and check every number yourself before you sign anything.

Here is what makes it worth reading anyway. Every price below was read off the vendor's own pricing page or help documentation in the week this was published, not lifted from an aggregator that lifted it from a listicle. Where a vendor does not publish a number, this page says so rather than inventing a range. Where a third party pricing tracker reports a seat minimum the vendor's own page does not state, this page flags the discrepancy instead of picking whichever number reads better.

The framing comes from roughly forty conversations with agency and in-house recruiters over the past two months. One of them, a nursing recruitment manager who leads a small team, described his situation exactly the way this page is organised: "our system right now is a mess, looking at options on how to have a more permanent system or a robust ATS to help me and my recruiters have a more organized database of candidates." He was not comparing feature lists. He was trying to work out what it would cost to move.

Another recruiter, at an agency, replied to us with a sentence worth taping to the wall of anyone about to sign: "not looking for any new ATS, we just adopted a new ATS subscription last month." That is what the entry decision really buys. Not software. A year or more of not being in the market.

Step one: find your headcount on the tier chart

Before you compare anything, do this. Open each vendor's pricing page, find your seat count, and read across. You are not choosing a product yet, you are finding out which products will actually sell you the version you saw in the demo.

Three real examples, all from vendor pages as of August 2026.

JobAdder sorts its plans by team size. Recruiter Lite is 1 to 5 users. Recruiter Essential is 6 to 20. Recruiter Pro is 21 and up. Adder Intelligence, the AI layer with smart summaries, smart sync, smart job descriptions and smart floats, starts at Essential. So a three person agency is in Lite, and the AI is not in Lite. The company publishes no prices at all, stating instead that "rather than having a one-size-fits all approach to our pricing, we work to understand what you need from your recruitment software and create a tailored proposal for you."

Crelate publishes Essentials at $85 per user per month paid annually, and the plan covers up to 2 users and 20,000 contacts. Business is $119 per user per month with annual and monthly options. That means the moment you hire a third recruiter, your per-seat cost rises by 40 percent, because you are no longer eligible for the plan you started on. Third party pricing trackers also report a 5 seat minimum on Business. Crelate's own page does not state one, so treat the minimum as unconfirmed and ask them directly.

Loxo publishes Core at $149 per user per month billed annually and Professional at $199, and puts a seat number on a benefit rather than a price: "free migration for Core teams of 3+ seats, with guided self-serve tooling below that". A two person agency does the move itself.

None of that shows up in a features table. All of it changes your bill.

What three seats actually costs

Three recruiters, first year, using each vendor's own published numbers as of August 2026 and the cheapest plan a three seat team is eligible for. Annualised so you are comparing the same thing. Confirm all of it on the vendor's site before signing, because these pages change and several of them changed while this cluster of articles was being written.

Two things fall out of that table that no roundup says out loud.

The cheapest published options, Manatal and Zoho, are cheap in subscription and expensive in your own time, because a flexible generic platform hands the configuration work to whoever owns the business. That is a real cost at three people, it just does not appear on an invoice.

And the spread between the cheapest and most expensive agency-native option at three seats is roughly ten to one. That gap is not a quality gap. It is mostly a question of whether the vendor built for your size or is willing to sell down to it.

The floor is built from four things

Whenever a small agency tells me the bill came in higher than the demo suggested, it is one of these four.

Seat bands. Covered above. Your headcount decides your eligibility, not just your multiplier. Crelate's 2 user Essentials cap and JobAdder's 1 to 5 Lite band are the clearest published examples.

Billing term. Several published prices exist only on annual billing. Loxo's $149 and $199 are both "billed annually". Crelate's $85 Essentials is "paid annually" with no monthly option listed. Manatal is the honest opposite: it publishes both, $15 annual against $19 monthly, so you can see exactly what the commitment is worth. For a three person agency, annual-only is not a discount question, it is a cash question, and it is also the thing that keeps you from leaving when the tool turns out to be wrong.

Tier gates on the thing you came for. This is the one that stings. Read the previous article in this series on what to ask an AI-native ATS and you will see the pattern repeated across four vendors: the published price buys the filing cabinet, and the part that reads it lives in a tier with no number. Recruiterflow publishes Platform at $149 and puts every AIRA agent, including Matchmaker, on a custom priced plan. Bullhorn publishes $99 and $165 and puts Amplify in quote-only Pro. JobAdder puts AI in a band that starts at twice your headcount. If the reason you are buying is the AI, check which tier it is in before you compare per-seat prices at all.

À la carte line items. Recruit CRM is the most transparent example precisely because it publishes them: Pro at $95, Business at $135, Enterprise at $215, with the note that enterprise features are billed separately, followed by a price list. Workflow automation is $41 a month for 1,000 tasks. Job multiposting is $20 a month for 7 jobs. Data enrichment is $250 a year for 3,000 credits. LinkedIn message sync is $12 per user per month. Analytics is $25 per user per month. All of those are free on Enterprise, which is the actual pricing argument being made. Crelate does the same thing less explicitly, marking native contact enrichment, email campaigns, text messaging and e-signature with "additional fee applies", and pricing its data sync service on volume and frequency. Neither is dishonest. Both mean the number on the card is a starting point.

The 2026 flip: published enterprise, quoted SMB

If you read small-agency ATS advice written even a year ago, the shape of it was consistent. The scrappy SMB vendors published their prices, the enterprise vendors hid theirs behind a sales call and a five figure implementation, and the advice was to stay away from the second group.

That is no longer what the pricing pages say.

Bullhorn, the vendor most often named as the one that will quote you into next year, now publishes a small agency section with Starter at $99 per user per month and Core at $165, and states free implementation on both, and on Pro as well. Whether Bullhorn is the right system for three people is a separate argument, and the honest answer is often no, because a configurable platform assumes an administrator you do not have. But the pricing objection specifically has been answered, in public, with a number.

Meanwhile the products positioned at agencies our size increasingly quote the part that matters. Recruiterflow publishes the platform and quotes the AI. JobAdder publishes nothing. Loxo publishes both tiers, which is genuinely better than it was earlier this summer, and prices the agentic tier at $199 per user per month with 2,500 credits per seat pooled across the team, plus paid Boosts at 2x and 5x if you run through them.

The useful conclusion is not that one group is good and the other bad. It is that the old heuristic, published price equals small-team friendly, has stopped working. Replace it with a narrower question: is there a published number attached to the specific capability I am buying, at the seat count I actually have? For a lot of vendors in 2026 the answer is no, and that is worth knowing before the demo rather than after.

The second number: what it costs to leave

Here is the part that gets left out entirely.

Your candidate base is the only compounding asset a small agency owns. Eight years of resumes, notes on who was good and who ghosted, which client rejected whom and why. The software is replaceable. That is not. And every month you run an ATS, the cost of changing your mind goes up, because more of the asset lives inside someone else's schema.

So the exit terms are a buying criterion, at the same level as search and workflow. What is actually documented, as of August 2026:

Manatal publishes a real migration process, which is more than most. Self-import for small datasets via CSV or JSON, plus a migration template for custom sources, and Manatal-executed migration for large or complex ones. Their documentation puts an executed migration at 2 to 4 weeks from when they receive your data, and tells you to export your full dataset from your old system including candidate attachments. That last instruction is a quiet warning about your current vendor, not about Manatal.

Recruit CRM splits it cleanly: free self-serve import from Excel or CSV, or paid expert data migration engineers, priced by the amount of data, taking 10 to 14 business days.

Recruiterflow offers white-glove migration at a paid rate, with the fee waiver available on multi-year contracts. Read that trade honestly: the migration discount is paid for with lock-in.

Loxo gives free migration to Core teams of 3 or more seats and self-serve tooling below that, and states on its pricing page that your data stays yours and your database is private.

Bullhorn publishes free implementation across Starter, Core and Pro. Implementation and migration are not the same word, so ask specifically whether bringing your existing base across is inside that.

Now the harder direction. Almost every vendor will let you export a candidate list to CSV, and almost every vendor's marketing treats that as data portability. It is not. A usable exit has four layers, and vendors often deliver only the first:

1. Records: candidates, contacts, companies, jobs.
2. Activity: notes, tasks, emails, submissions, who said what and when.
3. Files: the actual resume documents, attached to the right people.
4. Relationships: who was submitted to which client for which role, and what happened.

Lose layer 2 and you have lost your institutional memory. Lose layer 3 and you have lost the documents themselves. Lose layer 4 and you cannot prove ownership of a placement in a dispute.

I could not find a vendor on this page that publishes a clear, complete answer to all four in public documentation. That is not an accusation, it is just what the pages say. Which means you have to ask, and the asking is the test.

Six questions to get answered in writing

Send these before you sign. Not to be difficult, but because the answers are the actual product differences, and how quickly a vendor answers tells you as much as what they say.

1. At my exact seat count, which plan am I eligible for, and is the capability I saw in the demo included in that plan?
2. Is the published price available monthly, or annual only?
3. What is the total first year cost including any implementation, migration and add-on line items, quoted as one number?
4. Can you import my existing base including the resume files themselves, who does the work, and what does it cost?
5. If I leave, can I export every candidate, contact, company, job, note, activity and attached file, in a machine readable format, at no additional cost, and how long do I have access after cancellation?
6. What happens to the AI features if I stay on the entry tier?

Any vendor confident in its product will answer all six in an email. Hesitation on number 5 is the answer to number 5.

When a small agency does not need an ATS at all

Most pages in this category cannot say this, because they are selling one. I sell one and I will say it anyway, because getting this wrong costs you more than picking the wrong vendor.

A spreadsheet, a mail client and a folder of CVs is still the right setup if most of the following are true. You are one recruiter. You work a narrow vertical where you personally know your candidates. You are running a handful of live roles at once rather than a dozen. Your work is relationship-led and repeat-client, not volume inbound. Nobody else needs to look up a candidate's history while you are asleep.

In that shape an ATS adds admin and removes nothing. Every hour you spend configuring fields is an hour not spent billing.

The trigger is not growth and it is not revenue. It is one of two specific failures.

The first is simultaneity. The moment two people need the same candidate's history at the same time, a spreadsheet stops being a system and becomes a rumour. The classic version is two recruiters approaching the same person with two different pitches, and a client noticing.

The second is recall failure. The moment you know you have spoken to the right person for this role and cannot find them, you are paying for a database you do not have. That is the pain worth buying software for, and it is a different problem from tracking, which is why we wrote a separate piece on reactivating a candidate base you already own and another on what candidate rediscovery software actually is.

One caution on the second trigger, from our own research. When we tested this messaging on working recruiters, one told us flatly that "the entire database is searchable, and I rarely start from scratch", and he was right. Searchable is table stakes. The failure is subtler than a missing search box: keyword and boolean search only return people who happened to describe their work in the words you thought to type. The strongest candidate in your base is often the one your search will never return, which is a reading problem rather than a storage problem. We wrote that argument out in full in why keyword filters bury strong candidates.

Buy carefully, because the window shuts for a year

The most useful sentence anyone said to us during this research was a rejection. An agency recruiter replied to our outreach with: "not looking for any new ATS, we just adopted a new ATS subscription last month."

That is the real cost structure of this decision. Not the monthly fee. The fact that once you sign, migrate a base and get your team into the habit, you stop evaluating for a year or more, whether or not the choice was right. Annual-only billing extends that. A paid migration you got discounted for a multi-year term extends it further.

Which is the argument for two things. Start on something you can leave, and if you can, start on something you can run alongside what you have rather than instead of it. A free tier you can test with a real role, on your real base, with no card and no migration project, tells you more in an afternoon than four demos will. If the tool cannot show you value before you commit your data to it, that is information.

For the broader argument about why this category is shifting at all, the pillar piece is the shift to AI-native recruiting. If you have already decided you want an agency-native system and are just choosing between them, Recruiterflow alternatives works through that comparison directly.

Where Gemsift fits, honestly

Gemsift is the AI-native ATS for agencies whose current ATS has become a filing cabinet. You move in with one export file, and it re-reads your whole candidate base against each new role, returning a client-ready shortlist with a plain English reason per pick. Free, Solo at $99, Agency at $299 for 3 seats, Scale from $899 a month, verified on our own site the day this was published.

Measured by the two numbers this page is about:

On the floor: there are no seat bands gating capability. The scoring with reasons and the re-read of your base are in every plan including the free one. The limit is base size, 100 candidates on free, 1,000 on Solo, 5,000 on Agency. Agency is flat for 3 seats rather than per user, which is why it lands mid-table at three seats and better than most as you add people. There is no implementation fee because there is no implementation: you upload a file.

On the exit: full structured export of the base, including every profile field and every per-role evaluation, in CSV or JSON, on every plan including free. It is not gated behind a tier and it is not a support ticket. The honest limitation, since this page is about honesty on exactly this point: the export gives you the structured data and the reasoning, and the original documents are downloadable per candidate rather than as a single bulk archive today. If you want the whole base including files as one download, ask us and we will tell you where that stands rather than pretending.

Where Gemsift is the wrong answer, plainly. If you need deep integrations with a payroll or back office system, job board posting and distribution, or enterprise compliance and procurement paperwork, keep a traditional ATS: those are real requirements and we do not serve them. If you are a pure outbound headhunter with almost no inbound, your problem is sourcing, not judgment on a base, and a sourcing tool is the better buy. If you get single digit applicants per role and know all of them personally, you do not need this or anything else on this page.

If your problem is that you have thousands of people you already paid to attract and no reliable way to find the right ones when a new role lands, that is the problem we built for.

Start free on your own base, no card

FAQ

How is this different from the AI already in my ATS?

Your ATS ranks on keyword and field overlap against the record it parsed when the resume arrived, and it runs your pipeline. Gemsift does the judgement call: who is worth a call and why, read against the context of the specific role including the unwritten requirements you keep in your head and never put in the public job post, with the reasoning in plain English so you can agree or overrule it.

There are two honest paths from here. Move into an AI-native system where that reading is the spine, or keep your current ATS and add a layer that does the reading. Both work. What does not work is expecting a keyword index to surface someone who described the same work in different words.

Half the applicants use AI to write their resumes now. Won't AI-written resumes just fool it?

That is what it is built for. Half the pile is now near-identical AI-written applications, and honest candidates write to the filter too, so signal gets buried on both sides.

Gemsift judges the substance under the wording instead of rewarding the best prompt. An AI-written resume is not a weak candidate, so nobody is disqualified for looking AI. We flag the low-signal filler and surface the actual person, so the strong applicant who wrote like a human does not sink.

I already do this in Claude or ChatGPT. Why would I pay for it?

If you already feed a model your role context for each role, you have proven the point: that context is what makes the call good.

What you would pay for is the part you rebuild by hand every time. Structure, so you are not reassembling prompts and files per role. Consistency, so the same criteria apply across a whole base and across your team rather than one careful chat. A persistent base, so last quarter's candidates are still there when this quarter's role lands. And an explanation you did not have to write yourself. If a tuned project already gives you all of that, keep it.

Do I have to migrate off my current ATS to try this?

No, and for a first test you should not. Export your base as one file, upload it, run one real role, and compare the shortlist to what your current search returns for the same role. That is the comparison that tells you something. Migrating properly is a decision to make after that, not before.

Is my candidate data safe?

Your uploads are used to score your own pile, and you can evaluate the reasoning on built-in sample data without uploading a real resume. If your organisation has data-governance rules, start there and ask us for a data-handling summary covering where data lives and how long it is kept. Several recruiters have told us they cannot trial anything with real client data at all, which is a fair constraint and the reason the sample path exists.

Do I get a real say, or does the AI decide for me?

You decide, always. It encodes your criteria and the role's unwritten requirements, applies them consistently across the whole base, and explains its reasoning so you can agree, overrule or recalibrate. It is there to make your judgement faster and more consistent, not to make the call for you.

---

Prices in this article were read off vendor pricing pages and help documentation in August 2026. Vendors change pricing, tiers and seat bands frequently, and several of these pages changed within weeks of publication. Confirm current numbers on the vendor's own site before you sign anything.