# The Search-Breadth Score, to Widen One Element, Hold, or Tighten

*After this you can score your saved searches on five dimensions and leave with one decision: widen a named element, hold, or tighten.*

- Canonical URL: https://www.refolk.ai/candidates/guides/search-breadth-score-widen-hold-tighten
- Pillar: Applying at volume
- Format: Framework
- Published: 2026-10-10
- Last reviewed: 2026-10-10
- Reading time: 16 min

Running a search across many companies at once, you keep hitting the same fork: your criteria feel either too tight to feed your week or too loose to mean anything. Public advice splits unhelpfully between "cast a wider net" and "pick one target," with no way to tell which applies to you right now. This guide scores your actual saved searches on five dimensions and hands you one decision - widen a named element, hold, or tighten - plus the increment to change it by.

This is the breadth-of-criteria decision itself. It is not the alert-tuning playbook for reducing inbox noise, and it is not the channel score for where to apply. It is the question of whether your definition of the job is the right width, and which single filter to move.

## What the Search-Breadth Score measures

The Search-Breadth Score is a read on whether your search criteria are too narrow, about right, or too broad, built from five things you can measure about your own saved searches. It does not score the market. It scores the net you have cast into it.

The five dimensions, and what each one proves:

- **Inflow** - how many genuinely new postings a search surfaces per week. Proves whether there is enough raw material to feed your application capacity. It lies when the count is inflated by duplicates.
- **Fit rate** - the share of surfaced postings you would actually apply to. Proves whether the filter is pointed at your target. It lies when "open to anything" makes everything technically a fit.
- **Response rate** - interviews divided by applications sent. Proves whether the problem is breadth at all. It lies when ghost jobs depress it regardless of your criteria.
- **Message coherence** - whether the set of searches describes one legible candidate. Proves whether breadth has become incoherence. It lies when high inflow masks a search no one could help you with.
- **Allocation** - how your limited alert slots are spent across lanes. Proves whether broad lanes are evicting focused ones. It lies when you forget the platform caps are doing the choosing for you.

A score is not a grade. It is a diagnosis that ends in exactly one move.

#### The five dimensions of a breadth read

1. **Inflow** - New postings per week per search, de-duplicated
2. **Fit rate** - Relevant over total on surfaced postings
3. **Response rate** - Interviews over applications, the breadth-versus-materials test
4. **Message coherence** - Whether all your lanes describe one candidate
5. **Allocation** - How your capped alert slots are spent

*Read from the top: inflow and fit are measured per search, the rest across the whole set.*

## Is my job search too broad, too narrow, or about right?

Too broad shows as high inflow with low fit rate. Too narrow shows as inflow below the application volume your situation calls for. About right is enough relevant postings to feed your week without wading through noise. The trap is reading these from feel instead of from numbers.

Start with inflow, because the word "enough" has no public standard. No platform publishes a postings-per-week band that defines too narrow versus flooded. So you set the floor from your own capacity: how many applications you can send in a week.

| Situation | Suggested weekly applications |
| --- | --- |
| Employed, selective | 5-10 |
| Employed, serious move | 10-15 |
| Unemployed / active | 20-30 |
| Competitive field / pivot | 30-50 |

If you are unemployed and active, you need relevant inflow to feed 20 to 30 applications a week. A search surfacing six relevant postings is starving you, whatever it feels like. The explicit floor from the career literature is that under 5 applications per week in an active search is too slow for most people. Above that floor, your target is the volume your situation sets, not someone else's.

Then read fit rate alongside it. The qualitative rule is consistent across platforms: if filters are too narrow, results are limited; if too broad, you receive jobs that are not relevant. There is no published numeric relevant-to-irrelevant threshold, so I treat it as a per-lane judgement - tag a week of a search's output and look at the fraction. A lane returning mostly jobs you scroll past is too broad even if the raw count looks healthy.

> **Rule:** Measure before you move
>
> A breadth decision made on how the search feels is a guess. Set a floor from your weekly application capacity, count de-duplicated inflow against it, and tag a week of postings for fit rate before you touch a single filter.

## Reading response rate before you touch breadth

A near-zero response rate rarely means your search is too narrow. If inflow is adequate and fit is fine but almost nothing comes back, the problem is your materials and targeting, not the width of your net. Widening inflow in response to silence usually just adds more silence.

The base rates make this concrete. CareerPlug's look at more than 10 million applications found roughly 3% of applications reach an interview, and a good response rate is 3 to 5%, with 2 to 3% average for cold online applications. Above 3% means your targeting is working; above 5% is excellent. So "almost nothing comes back" is the normal condition of applying, not evidence that you defined the job too tightly.

**3% - Share of applications that reach an interview, on average**

From CareerPlug's analysis of more than 10 million applications - silence is the base rate, not a breadth signal.

Two numbers deepen the point. An estimated 20 to 40% of postings at any time are ghost jobs - already filled, on hold, or pipeline-building - so a chunk of your inflow was never going to respond to anyone. And Jobscan's report, pairing a survey of 442 seekers with over 2.5 million applications, found 44% got no interviews in the prior month. Silence is widespread even for people whose breadth is fine.

Channel changes the arithmetic more than breadth does. Response rate by method runs from 1 to 2% for Easy Apply, 3 to 5% for direct company sites, 10 to 20% for recruiter-submitted, and 30 to 50% for employee referrals. If you are at 1% through Easy Apply, no amount of widening fixes that; a referral does.

> **Watch out:** The 30-to-50 rule before you widen
>
> If you have sent 30 to 50 applications with zero interviews, do not add volume. The evidence points to targeting, resume fit, keywords, and source mix as the cause, not the width of your criteria. Fix those first; widening here just multiplies a broken funnel.

## The talent-pool tell, from Refolk's index

Before you widen a title or a market, check how crowded the lane already is. The size of the candidate pool for a title is a proxy for competition, and it moves far more than people expect when you shift titles or countries.

In Refolk's index of professional profiles, the counts for one role show how large a "small tweak" can be.

| Lane | Profiles in pool | Ratio vs US Product Manager |
| --- | --- | --- |
| Product Manager, United States | 67,737 | 1.00x |
| Product Manager, United Kingdom | 15,761 | 0.23x |
| Program Manager, United States | 140,069 | 2.07x |

Two readings matter for a breadth decision. First, switching from "Product Manager" to the adjacent "Program Manager" more than doubles the pool, from 67,737 to 140,069 in Refolk's index. That tells you the role competition - and likely the posting volume - also roughly doubles. Title adjacency is not a minor adjustment; it is the single largest breadth move available, which is exactly why you change it alone and watch what happens.

Second, the same title is about 4.3x scarcer in the UK than the US (15,761 against 67,737, derived from those counts). A geography-widening instinct that works in a dense US metro can starve a UK search. Set your breadth thresholds per market, not as one global rule. For US Product Managers, top current employers in the index include Ramp, Brex, Intuit, Wiz, and Formlabs; for UK Product Managers, they cluster at Google DeepMind, Meta, Coinbase, and Granola, concentrated heavily in London - a much thinner, more geographically pinned field.

If you want to see your own lane's depth and shape before you move a filter, [Refolk](/candidates) can map who already holds the title you are chasing, where they sit, and where they came from, so your breadth decision is made against the real population rather than a guess.

> Title adjacency is the largest single breadth move you have, which is the reason to change it alone.

## Which element to widen first, and by how much

Widen geography first, because radius is the only lever whose effect is quantified before you pull it. Widening a title or an industry changes result counts unpredictably; moving a radius changes them by a known multiple. That lets you predict the inflow change and attribute it afterwards.

| Change | Effect on result count |
| --- | --- |
| 5 to 10 miles | ~4x |
| 15 to 20 miles | ~2x |
| 20 to 25 miles | ~1.5x |
| Default starting radius | 25 miles |

Job boards default candidate search to 25 miles, with options from 5 to 100 miles, and LinkedIn defaults to 25 miles (40km). So most searches start at the default and can go either way. If inflow is short, the cheapest first move is a radius bump you can predict: 15 to 20 miles roughly doubles results.

Radius has a ceiling, though. Searching a dense metro at 100 miles pulls in another state's worth of irrelevant results - the engine documentation notes a New York City search at 100 miles returns jobs in New Jersey and Upstate New York, and a smaller radius raises relevance. Widen radius until fit rate starts to fall, then stop and switch levers.

When radius is exhausted, titles are next, widened along an archetype. Up to 30 job titles can belong to one job archetype, so "Product Manager" can legitimately absorb adjacent titles. But remember the pool math: Program Manager more than doubles the field, so add adjacent titles in small batches and re-measure fit, not in one sweep.

#### Inflow against fit, and what to change

Horizontal axis runs from Low inflow to High inflow. Vertical axis runs from Low fit to High fit.

| Quadrant | What it means |
| --- | --- |
| Low inflow, high fit | On target but starved - widen radius first, then adjacent titles |
| High inflow, high fit | About right - hold and re-measure next cycle |
| Low inflow, low fit | Wrong lane - tighten to one coherent filter, then rebuild |
| High inflow, low fit | Too broad - cut one filter, usually a generic keyword |

*Plot each saved search by its weekly inflow and its fit rate, then read off the move.*

## Running the score: inventory to decision

Here is the procedure end to end. It takes one week because inflow and response only mean something over seven days. Run it once, land on a decision, change one thing, and run it again.

#### From saved searches to one decision

1. **Inventory every saved search** - Pull all active searches and alerts across every platform into one spreadsheet with columns for title terms, location and radius, seniority, and industry.
2. **Measure weekly inflow per search** - Record genuinely new postings each search surfaced over seven days, then compare each count to your own weekly application capacity since no public inflow band exists.
3. **Score fit rate per search** - Tag a week of surfaced postings as relevant or irrelevant and compute relevant over total for each lane.
4. **Pull your application-to-response rate** - Divide initial interviews by total applications sent and compare to the 2 to 5% band.
5. **Diagnose breadth versus materials** - If inflow is adequate and fit is fine but response is near zero, the cause is materials and targeting - zero interviews after 30 to 50 applications means fix that, not volume.
6. **Pick one element to change** - Choose exactly one of title, industry, geography, or seniority; widen with radius first because its multiplier is known, or tighten by cutting one filter.
7. **Apply the change at a documented increment** - Make the smallest predictable move - radius 15 to 20 miles to roughly double results, or a small batch of adjacent titles within the archetype.
8. **De-duplicate and re-save** - Keep one to two alerts for your main goal plus one to two for closely related roles and remove overlapping lanes.
9. **Re-measure after one cycle** - One week later recount inflow, fit, and response; hold if metrics landed in range, otherwise repeat with the next single element.

Note that sources genuinely disagree on where volume sits. Some argue most seekers apply too broadly and that the best searches are narrow and relationship-driven. Others argue for a genuinely relevant set at meaningful volume, because a strong base application to many relevant roles beats a perfect application to a handful. The score sidesteps the argument: it tells you which side you are on this week from your own numbers, rather than picking a camp for everyone.

## Allocation: when the platform makes the call for you

Your alert slots are capped, so breadth is a zero-sum allocation whether you decide it or not. LinkedIn caps members at 20 job alerts at once, and Glassdoor allows a maximum of 10 alerts per day. Every broad lane you add evicts a focused one.

The documented discipline is a small, deliberate set, reviewed weekly. Indeed advises one to two alerts for your main goal plus one to two for closely related roles, to reduce overlap and keep results relevant. Practitioners recommend prioritising 3 to 5 alerts and reviewing the list weekly: broaden low-volume alerts, delete inactive ones, and add new targets. A spreadsheet tracks settings and frequencies, and a dedicated email reduces duplicate notifications.

The weekly review is where breadth stays honest. Without it, old broad lanes linger, dilute your fit rate, and quietly occupy slots a sharper search should hold. Treat the review as part of the search, not admin around it.

To run a tailored application at the volume a widened search produces without dropping fit, Refolk writes each application from your own history and tailors it to every posting, then scores how well you actually fit - which is what keeps a 20-to-30-a-week pace from collapsing into generic blasts.

## How the breadth read goes wrong

The breadth read fails in predictable ways, almost all of them a true number pointing at a false cause. This is the section to re-read before any decision, because every one of these failure modes leads a reasonable person to change the wrong element.

- **Inflow looks healthy but is duplicate volume.** Two overlapping alerts inflate the count, and you may get multiple alerts for the same job posted in different locations. The false positive is a "flooded" reading that is really the same jobs twice. De-duplicate postings before counting.
- **Low inflow misread as too narrow when it is a dead radius.** A tight radius near a city edge can show zero. The false positive is widening titles when the fix was geography. Test radius alone first: if nothing appears, expand the radius before anything else.
- **High fit but zero responses treated as a breadth problem.** This is materials and targeting, not width. Apply the 30-to-50 application rule before widening anything.
- **Widening more than one element at once.** You cannot attribute the change, so you may keep a move that hurt you. If two filters changed in a cycle, the cycle is void - rerun it with one.
- **"Open to anything" mistaken for broad coverage.** Telling people you are open to anything seems to create opportunities but makes it harder for them to help. The false positive is high inflow with incoherent messaging. Check message coherence, not just the count.
- **Flooded inbox read as strong demand.** Volume is what crowded postings already have; community-tested auto-apply batches of 100 to 300 often report roughly 0 to 1% meaningful response. Check your response rate, not your inflow.
- **Ghost jobs inflating apparent inflow and depressing response.** With 20 to 40% of postings estimated to be ghost jobs, discount your inflow accordingly and do not blame your own breadth for the silence they cause.

#### Why raw inflow overstates real breadth

| Stage | Figure | Note |
| --- | --- | --- |
| Alerts reported this week | 100 | Raw count across all lanes |
| After removing duplicates | 70 | Same jobs posted in multiple locations stripped out |
| After discounting ghost jobs | 45 | 20 to 40% are filled, on hold, or pipeline-building |
| Relevant to your target | 20 | The only number your breadth decision should use |

*The postings worth counting are a fraction of what an alert reports, so count the bottom, not the top.*

The pattern across all seven is the same: a number is real, but it is attached to the wrong cause. Discipline is counting the bottom of that funnel, not the top, and testing one lever at a time so cause and effect stay legible.

## Before you call the breadth decision done

Run this check before you lock a change and move on. If any item fails, you are about to change the wrong element or attribute a result you cannot trust.

#### Breadth decision sign-off

- [ ] Every active saved search is in one spreadsheet with title, radius, seniority, and industry columns.
- [ ] Inflow was counted after de-duplicating postings, not from raw alert totals.
- [ ] Inflow is compared against my own weekly application capacity, not a generic benchmark.
- [ ] Fit rate was scored from a tagged week of postings, per lane.
- [ ] Response rate was pulled and checked against the 2 to 5% band before blaming breadth.
- [ ] The 30-to-50 application materials check was applied if response was near zero.
- [ ] Exactly one element is being changed this cycle, and I can name it.
- [ ] A widening move used a documented increment I can predict, starting with radius.
- [ ] My alert set is one to two main-goal lanes plus one to two adjacent, with no overlaps.
- [ ] A re-measure date one week out is on the calendar.

To keep the read current, rerun the whole loop on a weekly cadence while you are actively searching. Markets move, ghost-job density shifts, and your own funnel changes as your materials improve - so a breadth decision that was right last week can be wrong this week. Hold when inflow, fit, and response all land in range. When any one drifts out, change one element, re-measure, and let the numbers, not the mood of the search, tell you which way to move.

**One-line breadth verdict**

```
Lane: [search name]
Inflow (de-duped, relevant): [n] vs capacity [n/week] -> [starved / adequate / flooded]
Fit rate: [relevant]/[total] = [%]
Response rate: [interviews]/[applications] = [%]
Diagnosis: [too narrow / about right / too broad / materials problem]
Decision: [WIDEN radius 15->20 / WIDEN +N adjacent titles / HOLD / TIGHTEN cut filter: ___]
Re-measure on: [date]
```

*Fill in from your spreadsheet after a full cycle. Keep the completed line in your tracker so the next cycle has a baseline.*

## Frequently asked questions

### Is my job search too broad or too narrow?

Score it on three readings before you decide. Measure weekly inflow against your application capacity, tag a week of postings as relevant or irrelevant to get a fit rate, and pull your application-to-response rate. Too broad shows as high inflow with low fit; too narrow shows as inflow below the volume you can feed with your own capacity. If fit and inflow are both fine but responses are near zero, the problem is materials, not breadth.

### When should I widen my job search criteria?

Widen when inflow falls below the weekly application volume your situation calls for and fit rate on the surfaced jobs is healthy. Start with geography because its effect is predictable: moving radius from 15 to 20 miles roughly doubles results. Do not widen in response to silence alone, because with ghost jobs and a 3% average interview rate, more inflow usually just adds more silence.

### Which element should I change first to get more results?

Geography, specifically radius. It is the only lever with a documented multiplier, so you can predict the inflow change before you make it: 5 to 10 miles is about 4x, 15 to 20 is about 2x, and 20 to 25 is about 1.5x. Title adjacency is the largest move but also the least predictable, so change it alone and only after radius is exhausted.

### My job search is not returning enough results. What do I check first?

Test radius alone before touching anything else. A tight radius near a city edge can show zero results while the titles are fine, so widening titles would be solving the wrong problem. If no offerings appear, expand the radius first, re-measure, and only then consider adding adjacent titles or industries.

### How many saved searches and alerts should I run at once?

Keep a small, deliberate set: one to two alerts for your main goal plus one to two for closely related roles. LinkedIn caps members at 20 alerts and Glassdoor at 10 per day, so every broad lane you add evicts a focused one. Review the list weekly, broaden low-volume lanes, and delete inactive ones.

### Why change only one element per cycle?

Because if you change two filters at once you cannot attribute the result. If inflow doubles after you widened radius and added titles, you will not know which move did it, and you may keep a change that hurt fit rate. The rule is one element per cycle; if two filters changed, treat the cycle as void and rerun it.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/search-breadth-score-widen-hold-tighten*