AI sourcing in plain English: from a broad search to a shortlist

Why AI sourcing in plain English works differently from boolean search

What boolean search asks of a recruiter

Boolean search revolves around searching with operators such as AND, OR and NOT. This requires a fair amount of preparation, because you need to think of job titles, synonyms and exclusions in advance. As a result it often takes a lot of time, especially for new roles or scarce audiences. If one important word is missing, a good candidate can drop out of view immediately. That's why many teams increasingly choose sourcing without boolean when they want to learn from the first results faster.

What searching in plain English changes

When you search in plain English, you simply describe the work, the context and a few clear conditions in ordinary sentences. You explain what the candidate needs to do, what type of role they've worked in, and roughly where they need to live. You then look at the outcome and adjust. This makes searching for candidates in natural language far more transparent for recruiters and hiring managers. On our page about AI sourcing in plain English we show how we support this, while the recruiter keeps full control over the criteria and the outcome.

Why a broad first search is useful

A broad start is often a smart choice. You want to see first how the system interprets the brief. The first results usually make it immediately clear whether you're leaning too heavily on job titles, whether the region is set too wide, or whether the desired experience is still described too generally. That's why searching for candidates with AI often works better if you don't lock everything down straight away. This gives you room to learn before refining your search for the candidates who form the perfect match.

How to start AI sourcing in plain English from a broad search

The example vacancy

Say you're looking for a field sales account manager in the Utrecht region. This role calls for experience with B2B sales, client visits and working with colleagues. You're looking for someone on a permanent employment contract, living within a reasonable travelling distance, with enough experience to get started independently right away. This is an excellent starting point for translating a vacancy into a search brief. The core of the work is already clear, without you getting bogged down in long lists of loose keywords.

The first search brief, written out in full

A first brief in plain English might sound like this: find candidates for a field sales role within B2B sales in the Utrecht region. The ideal candidate can build client relationships, plan appointments independently and is used to a commercial environment. Experience with account management or field sales is relevant. The candidate also preferably works on a permanent contract and lives in or around Utrecht.

Why this brief deliberately stays broad

This first version is deliberately kept broad enough to capture different job titles. Someone might be called an account manager, but also a sales consultant, relationship manager or business developer. If you search too specifically right away, you miss people who could do the job brilliantly but have a different job title on their profile. In this case, the first search brief returns 12,035 candidates. That's a lot, but it provides highly useful input for the next step within LinkedIn sourcing and iterative sourcing more broadly.

Which irrelevant candidates you'll still see

This first list will undoubtedly still include people who live too far away, work as freelancers or contractors, or mainly have experience in inside sales. You'll also see candidates with too little matching experience, or employees of employers you specifically want to exclude. At this stage, that's absolutely not a mistake. This noise is exactly what shows you where to make targeted corrections next. So start with hard exclusions. Add preferences only later, because otherwise you risk losing valuable candidates too soon.

Adjusting AI sourcing in plain English, hard filters first and preferences second

First adjustment: no freelancers or contractors

Start with a hard exclusion. Since you're looking for someone on a permanent contract, you specify that freelancers and contractors don't qualify. This is often a safe first step; it immediately removes a significant chunk of the noise without costing you strong candidates. For teams working with AI sourcing tools, this is a logical first correction that makes the outcome noticeably cleaner right away.

Second adjustment: a maximum 30-minute commute

Next, add a practical condition, such as a maximum travel time of 30 minutes to Utrecht. This helps filter the candidate list for feasibility. Do bear in mind that travel time always needs a manual check, since distance doesn't work out the same for everyone. Someone might live just outside the stated limit and still be perfectly reachable. It's a useful filter, but not a reason to trust the outcome blindly.

Third adjustment: at least 3 years of relevant experience

Next, make the required experience more concrete. Add that the candidate needs at least three years of relevant experience in B2B sales or account management in a field sales role. The word "relevant" is essential here, because not every year of sales experience is equally useful for this specific vacancy. At this point, you often see the results improve substantially in quality, and filtering the candidate list this way delivers far more than adding yet more job titles.

Fourth adjustment: excluding the current employer

Sometimes you'd rather not approach candidates from a specific employer. This might be down to client agreements, or because someone from that company is already in the process. You add this too as a simple hard exclusion. After these four steps, the list in our case drops from 12,035 to 642 candidates. This shows why refining your search for candidates works best when you follow a fixed order. First you remove the biggest sources of noise. Then you look at the results again and decide whether the list is workable.

• Step 1 removes freelancers and contractors, since the role calls for an employment contract.
• Step 2 limits travel time, because reachability has a direct impact on the chance of interest and a successful placement.
• Step 3 makes the experience concrete; overly broad sales experience otherwise creates too much noise.
• Step 4 excludes specific employers, so you honour existing agreements and avoid duplicate outreach.

What AI sourcing in plain English teaches you going from 12,035 to 642 candidates

Why 642 candidates still isn't a shortlist

Although 642 candidates is a substantial improvement, this still isn't a shortlist. It's a workable longlist. The list is small enough to assess properly, but large enough to compare profiles. Many recruiters make the mistake of filtering too far at this point. That can make exactly the one candidate with a less obvious job title or an unconventional career path disappear, even though they're actually a perfect match in substance.

When you should stop refining

Stop refining once the list is large enough to compare candidates and small enough to check each one individually. That's the point where the work shifts from searching to assessing. We support this process by scanning thousands of candidates and explaining, per profile, why someone is a good fit. Because the recruiter keeps full control over the criteria and the outcomes, the final decision always stays human and easy to explain.

Why refining for too long carries risk

Refining for too long quickly makes a search too narrow. Good candidates, after all, don't always carry exactly the same job title, and relevant experience can also come from an adjacent market. Sometimes a career path is a bit less linear, but the knowledge gained is more than valuable for the role. That's why it's smart to switch straight to human assessment once you've set the hard filters. This is the practical limit where AI sourcing in plain English delivers the most value.

How to keep borderline candidates separate

Put candidates who don't quite seem to fit into a separate group. Briefly note for yourself what's still unclear, such as travel time, exact field sales experience, or knowledge of a specific market. Then check whether you can quickly find out that information. This keeps you organised and stops genuinely strong candidates from dropping out of view too early. This is especially valuable when building a shortlist in recruitment, because many good matches only really stand out once you leave room for nuance.

From AI sourcing in plain English to a shortlist and first outreach

Work with a simple scorecard

A simple scorecard works brilliantly for building a strong shortlist. Assess every candidate on four clear points: hard requirements, strengths, open questions and the reason for priority. This makes your final choice much easier to explain to hiring managers and colleagues. You really don't need to overcomplicate this. A short, fixed assessment per profile is usually more than enough to quickly see the difference between strong matches and candidates who still need a bit of extra checking.

• Hard requirements: does the basis check out, such as the region, the type of employment and the relevant work experience?
• Strengths: does the candidate have extra market knowledge, strong relationship-management skills, or experience in a comparable field sales role?
• Open questions: what do you still need to verify before placing someone firmly at the top of your list?
• Reason for priority: why do you choose to approach this candidate first?

Which candidates you approach first

Start with the candidates who score well on the hard requirements and also have clear strengths. Next come the profiles with just one, solvable open question. Candidates with several outstanding questions go a bit lower on the list. This approach works more efficiently, because you approach the best-supported matches first. On our page how Elvatix works from shortlist to message we explain in detail how searching, assessing and human oversight flow smoothly into one another.

How you hand the shortlist over to outreach and follow-up

Sourcing obviously doesn't stop at filtering. You want to know exactly who's already been approached, who's replying, and which candidates can move on to the next round. That's why it's crucial that searching and follow-up connect logically. We position our platform as a complete environment for finding, approaching and following up with talent. Candidates already approached before are recognised automatically before new outreach starts, messages always stay under human control, and progress remains easy to track. Curious how teams use this more broadly? You'll find extra context in our real-world examples of AI in recruitment.

A repeatable process with AI sourcing in plain English for teams with many roles

Working together without relying on one specialist

In many teams, there's one colleague who's skilled at boolean search, and the rest of the team leans on their knowledge. That makes the process fairly fragile. Searching in natural language makes collaboration far more transparent and straightforward. Recruiters, consultants and sourcers understand the starting brief faster and in the same way. This makes it easier to assess together, hand work over more quickly, and clearly explain why a candidate does or doesn't qualify.

When iterative sourcing helps with niche profiles

Iterative sourcing simply means you search step by step and keep learning from every intermediate result. This works especially well for roles with a wide variety of job titles or backgrounds. You don't need to get the first brief perfectly right straight away. This method is enormously valuable, particularly for niche profiles or less common vacancies; you discover far faster which words, context and filters really make the difference.

How agency and in-house teams put this to work

Agencies, staffing firms and in-house recruitment teams may work in different contexts, but the underlying approach is often exactly the same. You start broad, filter out the obvious noise, stop once you have a workable list, and only then build a shortlist. For teams running many assignments at once, we have a dedicated page on AI sourcing for recruitment agencies. There we show how this way of working fits seamlessly into day-to-day recruitment processes.

Exercise for AI sourcing in plain English: turn a vacancy into four plain sentences

Sentence 1: describe the work the person needs to do

Start by writing down what a candidate actually needs to be able to do. Think of something like: "Find someone who independently builds client relationships, plans appointments and successfully turns commercial opportunities into results." Always start with the work itself. This gives you far more direction than a long list of loose keywords and helps you get a sharp picture of what the role actually involves.

Sentence 2: add one hard exclusion

Next, add one clear exclusion. For example: "No freelancers or contractors, only candidates open to a permanent contract." Choose a requirement here that's genuinely non-negotiable. Don't add several exclusions at once. If you do, you learn less from the initial result and increase the risk of ruling out good people too early.

Sentence 3: add a practical preference

Now it's time to add a practical preference, such as "living within about a 30-minute commute of Utrecht." Always leave room for a manual check here, since preferences like this can sometimes be flexible without a candidate immediately being unsuitable. This step in particular helps enormously in moving from a broad search brief to a genuinely more usable list.

Sentence 4: make the relevant experience concrete

Round it all off with the experience that really matters. For example: "At least 3 years of relevant work experience in B2B sales in a field sales role." Then read back through your four sentences critically and check whether they still read as normal, natural English. If they do, you generally have an excellent starting point for your sourcing. Want to try this straight away in a real workflow? Through the link try Elvatix yourself you'll see with your own eyes how this works in practice.

Frequently asked questions about AI sourcing in plain English

How broad can a first search brief be?

Often broader than most recruiters think. As long as the core of the role, the desired region and the basic context are clear, you can course-correct in a targeted way afterwards. The main goal of this first step is to see exactly where the noise sits and which filters are genuinely needed at a later stage.

When do you change the brief, and when do you change the filters?

Change your search brief if the substance of the results doesn't line up well. Use filters, on the other hand, when the substance is right but the list is still too large or too messy. Seeing too many inside sales profiles while you're looking for a field sales employee, for example? Then you need to sharpen the description of the work. But if you see fantastic candidates who simply live too far away, a practical filter is the right answer.

How do you turn a candidate list into a shortlist?

Use a scorecard with hard requirements, strengths and any open questions. Then assess, per candidate, why they're a good fit. Put the best-supported matches straight at the top of your list, and keep the profiles that don't quite seem perfect in a separate group. This makes your shortlist not only easier to explain, but also immediately more practical for the actual outreach.

What do you do with candidates who almost fit?

Set these profiles aside and clearly note what's still missing or unclear. Then check whether you can resolve that missing information with a quick check. This works considerably better than rejecting someone outright. Strong candidates simply don't always carry exactly the same job title or career path on their profile as described in your vacancy text.

Is AI sourcing in plain English a replacement for recruiter judgement?

No, absolutely not. It only helps you navigate faster from a broad search brief to a workable list. The recruiter always stays in control of the criteria, the outcome and the outreach. The quality of the final shortlist therefore always depends on sound human assessment and strong, personal follow-up.