How many tools does a recruitment agency need?
Most agencies pay for somewhere between eight and fifteen pieces of software and use four of them properly. Here is how to work out what you actually need, what each extra tool costs you beyond its invoice, and which things are worth keeping separate.
Ask AI about this
Open the card statement and count the software lines. A CRM or applicant tracking system. A job board posting tool. A sourcing or contact-finding tool. An email sending platform for campaigns. A phone system. A video interview tool. An e-signature service. A scheduling link. A payroll or invoicing system. Possibly a separate reporting tool bought because the reporting in the CRM was not trusted. Then the spreadsheets, which are free and therefore invisible, and which usually hold the numbers the directors actually look at.
Very few agencies chose that stack. It accumulated. Each piece was bought on a specific Tuesday to solve a specific problem, usually by a different person, and nothing has ever been removed because removing software is somebody's whole week and nobody owns the job.
The number of tools you need is smaller than the number you have
A recruitment agency needs one system of record and a very small number of things that genuinely sit outside it. Everything a consultant touches while doing recruitment work belongs in the first category: candidates, clients, jobs, notes, calls, emails, messages, interviews, CVs, offers, signatures, adverts and the pipeline they all move through. Those are not separate jobs that happen to share data. They are one job, and splitting them across products is what creates most of the administrative work in a modern agency.
The things that genuinely sit outside are the ones with a different audience and a different owner. Accounting has an accountant, statutory deadlines and an auditor. Payroll for contractors has legal requirements of its own. Your website belongs to marketing. Those can be separate without hurting, because the handover between them and recruitment happens a handful of times per placement rather than a hundred times per day.
The useful question is not how many logins you have. It is how many times a day a consultant has to carry something from one product to another with their hands.
What does an extra tool actually cost?
The licence is the smallest part of the bill. An extra tool costs you the licence, the time spent moving data across the gap, the errors that gap produces, the admin of accounts and permissions, and a share of every future decision you make about anything else.
The data movement is the obvious one. Somebody exports a list, somebody imports it, somebody corrects a number in one place and not the other. The errors are less obvious, because they do not announce themselves. A candidate opts out of marketing in the campaign tool and stays subscribed in the CRM. An interview moves in the scheduling tool and the pipeline still says Tuesday. A signed contract sits in a signature service nobody logs into and never reaches the placement record.
The administrative cost is the one people forget. Every tool has users to add and remove. When a consultant leaves on a Friday, their access has to come off eleven systems, and the one everybody forgets is the one with the candidate database in it. Every tool has a renewal, an owner, a security questionnaire from your enterprise client, and a set of terms that changed last month.
Is it better to buy one system or the best tool for each job?
One system wins when the tools in question share the same records all day, and separate tools win when they do not. That is the whole rule, and it explains why the best-of-breed argument sounds right in general and goes wrong in recruitment specifically.
Best-of-breed works beautifully where the boundary is clean. Your accounting package does not need to know that a candidate replied to an InMail. It needs a placement and an invoice, once. The handover is small, rare and easy to check, so a specialist tool on each side is the better answer.
It works badly where the boundary is crossed constantly. A phone call is a candidate event. So is an email, a message, a video interview, a signature and an advert response. If those live in six products, then the story of a candidate lives in six products, and no single screen ever shows you what happened. Consultants respond by keeping the story in their heads, which works until they are ill, busy or leave.
There is a second reason, and it only shows up after a couple of years. An integration is a promise between two companies that they will keep agreeing about something. Promises hold until one side changes its product, its pricing or its owner. When that happens the failure is usually silent: records stop syncing and nothing turns red, and you find out from a candidate who says nobody called them back.
| Job | Belongs in the system of record | Why |
|---|---|---|
| Calls, emails, messages | Yes | Each one is an event on a candidate or client and belongs on their timeline |
| Video interviews and recordings | Yes | They are evidence about a person in a process, and need to be findable later |
| Offers and signatures | Yes | A signed document that is not on the placement is a document you will look for |
| Job adverts and applications | Yes | The application is the start of the pipeline, not a separate inbox |
| Sourcing and outreach campaigns | Yes | Consent, opt-outs and reply history have to agree with the CRM every time |
| Accounting and payroll | No | Different audience, different rules, a handover per placement rather than per hour |
| Website and public marketing | No | Owned by different people, changed on a different rhythm |
How to audit your stack in an afternoon
List every piece of software you pay for, then answer four questions about each one, and the list of things to cut writes itself. The four are: who owns it, how many people logged into it last month, what data it holds that exists nowhere else, and what would break on the day you stopped paying.
The login question finds the dead ones straight away. There is almost always a tool bought two years ago that two people use, one of whom is the person who bought it. The data question finds the dangerous ones, which are not the same list: a tool nobody uses may still be the only place a signed document or a consent record lives, and cancelling it without moving that data is how agencies lose things they are legally required to hold.
Then walk one placement end to end and write down every time somebody moved information by hand. Advert posted here, application landed there, CV downloaded and re-uploaded, interview booked in one calendar and typed into another, contract sent from a fourth place, invoice raised from a fifth. Each of those handovers is a candidate for removal, and the ones that happen several times per placement are worth more than the ones that happen once.
What to ask a vendor about what they actually own
Ask which parts of the product their own engineers build and which parts are somebody else's, then ask what happens to the second list if that relationship ends. Vendors answer the first question happily and the second one tells you far more.
It matters because a resold component behaves like a separate tool wearing your vendor's logo. The support escalation goes through two companies. The feature you want is on somebody else's roadmap. The price you pay moves when a third party changes their pricing, and the part of your data that lives there is subject to somebody else's terms.
On our side the phone system, the video recording and narration, the e-signature, the sourcing, the campaigns, the candidate and client portals, and the desktop and mobile apps are built by us rather than resold. That is a straightforward claim to check, and I would rather you checked it against everyone on your shortlist than took my word for it. The answer tells you how many tools you are really buying when you buy one.
My honest view is that most agencies would be better off with one system, one accounting package and nothing else, and that the stack they have instead is not a strategy but a sediment. Cutting it back is unglamorous work with no launch day, and it gives your consultants more hours back than any feature on any roadmap.
Siva is an engineer at Recruitly.
