# How to reproduce "Job change rates by job function, industry and region"

Source: https://www.reversecontact.com/research/job-changes
Data as of November 2026. Published by Reverse Contact.

This guide gives everything needed to produce the same figures with the
Reverse Contact API. The database is refreshed continuously, so a run made
later returns figures close to the published ones, not identical ones.

## What is measured

For a group of people (a job family or function, in a region or an
industry, optionally at a seniority level), two numbers:

- **profiles**: the people in the group;
- **started**: of those, the people whose current position started within
  the last N days, for N in 90, 180, 365, 730.

The share published is `started / profiles`, the job change rate. A job
change is any current position with a recent start date: a move to another company and a change of
role inside the same company both count. A third measure, read on a sample,
tells the two apart (last section).

## The one endpoint

```
POST https://api.reversecontact.com/v2/search/people/count
Authorization: Bearer <API_KEY>
Content-Type: application/json
```

It takes the filters of the people search and returns `{"data": {"count": <number>}}`.
A count call costs 0 credits. It counts toward the per-minute rate limit of
the key, so pace the calls.

## The filters

| Dimension | Filter |
| --- | --- |
| Job family | `"currentFunction": "<family>"` |
| Function | `"currentSubFunction": "<function>"` |
| Seniority | `"currentSeniority": "<level>"` |
| Industry of the current employer | `"currentCompanyIndustry": "<industry>"` |
| Position started within N days | `"changedJobsWithinDays": <N>` (omit it to count every profile) |

Regions:

| Region | Filter |
| --- | --- |
| Worldwide | no `location` filter |
| North America | `"location": {"area":"MR:NAMER"}` |
| North America: United States | `"location": {"countryCode":"US"}` |
| North America: Canada | `"location": {"countryCode":"CA"}` |
| Latin America and the Caribbean | `"location": {"area":"MR:LATAM"}` |
| Europe, Middle East and Africa | `"location": {"area":"MR:EMEA"}` |
| Europe, Middle East and Africa: Europe | `"location": {"area":"GR:EUR"}` |
| Europe, Middle East and Africa: Middle East, North Africa and Turkey | `"location": {"area":"MC:MENA","countryCode":"TR"}` |
| Europe, Middle East and Africa: Sub-Saharan Africa | `"location": {"area":"GSR:SSA"}` |
| Asia-Pacific | `"location": {"area":"MR:APAC"}` |
| Asia-Pacific: Indian Subcontinent | `"location": {"area":"MC:INSC"}` |
| Asia-Pacific: Eastern and South-eastern Asia | `"location": {"area":["GSR:EAS","GSR:SEA"]}` |
| Asia-Pacific: Oceania | `"location": {"area":"GR:OCE"}` |

The main regions and the zones inside them are zones of the API's own
catalogue (`location.areaCatalog` in the available fields, below), which
lists the countries of each. A zone id and a country code in the same
filter add up. The zones of a main region do not cover all of it: a few
small countries belong to the main region only.

The families, the functions of each family, the seniority levels and the
industries are the values of the API itself. List them from:

```
GET https://api.reversecontact.com/v2/search/people/available-fields
```

- `currentSubFunction.allowedValueGroups.groups` maps each family to its functions;
- `currentSeniority.allowedValues` lists the 10 seniority levels;
- `currentCompanyIndustry.allowedValues` lists the 490 industries.

Nothing is regrouped or renamed in the study.

## The cells

1. **Industries kept.** Count every industry once, worldwide, with no other
   filter, and keep the 25 with the most profiles. In this edition:
   IT Services and IT Consulting, Hospitals and Health Care, Software Development, Financial Services, Retail, Higher Education, Government Administration, Construction, Banking, Business Consulting and Services, Real Estate, Motor Vehicle Manufacturing, Telecommunications, Insurance, Education Administration Programs, Advertising Services, Oil and Gas, Hospitality, Non-profit Organizations, Food and Beverage Services, Manufacturing, Truck Transportation, Pharmaceutical Manufacturing, Restaurants, Utilities.
2. **Families.** For each region and each family, 5 calls: no period, then
   90, 180, 365, 730 days. First with no industry filter (every industry), then
   once for each of the 25 industries.
3. **Functions.** The same calls for each function: for each region, with no
   industry filter and then for each of the 25 industries.
4. **Seniority.** For each region and each seniority level, 2 calls (no
   period, 365 days); and worldwide, for each family crossed with each
   seniority level, the same 2 calls. No industry filter.

"Everyone" in a region or an industry is the sum of its families: the
classification puts every profile in exactly one family.

## What is published

A cell is published only when its `profiles` count is at least 200.
Below that it is left out (shown as n/a on the page, absent from the files).

## Example

People in the Sales family, in North America, whose position started in the
last 12 months:

```bash
curl -s https://api.reversecontact.com/v2/search/people/count \
  -H "Authorization: Bearer $REVERSE_CONTACT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"currentFunction": "Sales", "location": {"area": "MR:NAMER"}, "changedJobsWithinDays": 365}'
```

Run it again without `changedJobsWithinDays` for the denominator.

## A script that runs as is

Python 3, standard library only. It writes one line per cell to
`counts.jsonl` and can be stopped and restarted.

```python
import json, os, time, urllib.request

API = "https://api.reversecontact.com/v2"
KEY = os.environ["REVERSE_CONTACT_API_KEY"]
LOCATIONS = json.loads('{"WORLD":null,"NAMER":{"area":"MR:NAMER"},"US":{"countryCode":"US"},"CA":{"countryCode":"CA"},"LATAM":{"area":"MR:LATAM"},"EMEA":{"area":"MR:EMEA"},"EUR":{"area":"GR:EUR"},"MENAT":{"area":"MC:MENA","countryCode":"TR"},"SSA":{"area":"GSR:SSA"},"APAC":{"area":"MR:APAC"},"INSC":{"area":"MC:INSC"},"ESEA":{"area":["GSR:EAS","GSR:SEA"]},"OCE":{"area":"GR:OCE"}}')
PERIODS = [None, 90, 180, 365, 730]
TOP_INDUSTRIES = 25
PER_SECOND = 5  # stay under your key's per-minute limit

def call(method, path, body=None):
    request = urllib.request.Request(API + path, method=method,
        data=json.dumps(body).encode() if body is not None else None,
        headers={"Authorization": f"Bearer {KEY}", "Content-Type": "application/json"})
    with urllib.request.urlopen(request, timeout=180) as response:
        return json.loads(response.read())

fields = {f["name"]: f for f in call("GET", "/search/people/available-fields")["data"]["fields"]}
families = fields["currentSubFunction"]["allowedValueGroups"]["groups"]
seniorities = fields["currentSeniority"]["allowedValues"]
industries = fields["currentCompanyIndustry"]["allowedValues"]

done = {}
if os.path.exists("counts.jsonl"):
    done = {json.loads(line)["cell"]: json.loads(line)["count"] for line in open("counts.jsonl")}
out = open("counts.jsonl", "a")

def count(cell, body):
    if cell not in done:
        done[cell] = call("POST", "/search/people/count", body)["data"]["count"]
        out.write(json.dumps({"cell": cell, "count": done[cell]}) + "\n"); out.flush()
        time.sleep(1 / PER_SECOND)
    return done[cell]

def body(region=None, industry=None, family=None, function=None, level=None, days=None):
    b = {}
    if region and LOCATIONS[region]: b["location"] = LOCATIONS[region]
    if industry: b["currentCompanyIndustry"] = industry
    if function: b["currentSubFunction"] = function
    elif family: b["currentFunction"] = family
    if level: b["currentSeniority"] = level
    if days: b["changedJobsWithinDays"] = days
    return b

# 1. The largest industries.
sizes = {i: count(f"industry|{i}", body(industry=i)) for i in industries}
top = sorted(industries, key=lambda i: -sizes[i])[:TOP_INDUSTRIES]

for days in PERIODS:
    for region in LOCATIONS:
        for industry in [None] + top:  # None is every industry
            for family in families:  # 2. families
                count(f"family|{region}|{industry or ''}|{family}||{days}", body(region=region, industry=industry, family=family, days=days))
                for function in families[family]:  # 3. functions
                    count(f"function|{region}|{industry or ''}|{family}/{function}||{days}", body(region=region, industry=industry, function=function, days=days))
for days in (None, 365):  # 4. seniority
    for level in seniorities:
        for region in LOCATIONS:
            count(f"seniority|{region}|||{level}|{days}", body(region=region, level=level, days=days))
        for family in families:
            count(f"family_seniority|WORLD||{family}|{level}|{days}", body(family=family, level=level, days=days))
```

Then, for each group, divide the count with a period by the count without
one, and drop the groups whose count without a period is under 200.

## New employer or internal move

The count endpoint cannot tell a new employer from a role change inside the
same company. For that part we read profiles, for each job family and each
function, and for each of the four periods:

1. `POST /v2/search/people` with `"currentFunction": "<family>"` (or
   `"currentSubFunction": "<function>"`), `"changedJobsWithinDays": <N>` and
   `"perPage": 100`, following `nextCursor` until 300 profiles are read for a
   family, 200 for a function. No location or industry
   filter. This endpoint is billed per profile returned.
2. A new position is an **internal move** when the profile's `experience`
   lists another role at the same company (same `companyId`) besides the
   current one. Otherwise it is a **new employer**.

The split is therefore known worldwide, in every industry, for each period. The margin of error is about 5 points for a
family and 7 for a function; a function with fewer than 100 profiles read has
no split. The page multiplies
it by the family's share to print, for example, the share of a family that
joined a new employer in 12 months: an estimate, not a count.

## Reusing the data

The data is published under the Creative Commons Attribution 4.0 International
licence (CC BY 4.0): https://creativecommons.org/licenses/by/4.0/
It is free to reuse in your own analyses, studies, reports and products.
When you publish something built on it, name Reverse Contact as the source
and link to the study:

> Source: Reverse Contact, "Job change rates by job function, industry and region", November 2026. https://www.reversecontact.com/research/job-changes

## Limits to keep in mind

- Profiles are self-declared: people who do not update theirs appear as not
  having moved, so the shares are conservative.
- How complete and how recent profiles are varies by region, by industry and
  by function, which affects comparisons between them.
- The share is a snapshot of who is in a group today. Someone promoted from
  one function to another counts as new in the function they joined.
- The profiles read for the type of move are the first the search returns,
  not a random draw.
- Only aggregated counts are published. No individual is named or
  identifiable.
