Reed jobs in Python: UK salaries come per hour, per day and per year
A Reed search page carries its 25 jobs as data, salary included. Of 150 jobs, 85 % showed a salary, in three units; another 15 % said negotiable or competitive instead, and most of those still had a number in the data. How to read it, and the one column you must not drop.
Reed.co.uk is one of the big UK job boards. Its search pages are built with Next.js, so the page you download already holds the jobs as JSON, in a <script id="__NEXT_DATA__"> tag: no browser and no login needed. robots.txt leaves the search and job pages open (with ?pageno= for the next pages), closes /api/, and sets a Crawl-delay of 3 seconds only for two named crawlers; we waited 3 seconds between requests anyway.
On 7 November 2026 we read two pages each of three searches, 150 jobs, and checked a sample against the job pages themselves.
The short version#
- One page, 25 jobs, plus the total: 984 jobs for Python developer, 2,479 for data analyst, 1,640 for warehouse operative.
- Most rows show a salary, in three units: 127 of the 150 rows (85 %), 82 to 88 % in each search. Python developer: 35 per year, 9 per day. Warehouse operative: 27 per hour, 14 per year. Data analyst: 38 per year, 4 per day.
- 22 more rows (15 %) say "Salary negotiable" or "Competitive salary" instead of a figure, and 18 of them still have a number in the page data. A flag in
salaryDescriptionmarks them. That number is not real pay; the posting never shows it. Count it as a salary and 145 of the 150 rows (97 %) seem to have one. - The unit is a code in
salaryType: 1 is per hour, 2 per day, 5 per year. We checked 15 of the rows that show a figure against the schema.org JobPosting on their own pages: all 15 matched in unit, lowest and highest pay. - The same job can turn up on two pages: 1 of the 150 rows was a repeat. Keep the job id and drop doubles.
The code#
import json, re, time, urllib.request
UA = {"User-Agent": "reed-reader/1.0 (you@example.com)"}
UNIT = {1: "hour", 2: "day", 5: "year"} # salaryType codes, checked against each job page's own JobPosting data
# salaryDescription flags for "Salary negotiable", "Commission" and "Competitive salary": the posting shows those
# words, not the figure, although the page data still carries one
HIDDEN = (16, 32, 64)
def page_data(url):
req = urllib.request.Request(url, headers=UA)
html = urllib.request.urlopen(req, timeout=30).read().decode("utf-8", "replace")
return json.loads(re.search(r'<script id="__NEXT_DATA__"[^>]*>(.*?)</script>', html, re.S).group(1))
def row(job):
d = job["jobDetail"]
hidden = [flag for flag in HIDDEN if d["salaryDescription"] & flag]
paid = bool(d["salaryFrom"] or d["salaryTo"]) and not hidden
return {"id": d["jobId"], "title": d["jobTitle"], "employer": d["ouName"], "location": d["displayLocationName"],
"posted": d["displayDate"][:10], "expires": d["expiryDate"][:10], "work": d["workingOption"],
"salary_from": d["salaryFrom"] if paid else None, "salary_to": d["salaryTo"] if paid else None,
"per": UNIT.get(d["salaryType"], f"code {d['salaryType']}") if paid else None, "hidden": hidden,
"url": "https://www.reed.co.uk" + job["url"]}
def search(slug, pages=1, pause=3):
"""Jobs from a Reed search such as 'python-developer-jobs': (total found, rows), 25 rows a page."""
total, rows = 0, []
for n in range(1, pages + 1):
data = page_data(f"https://www.reed.co.uk/jobs/{slug}" + (f"?pageno={n}" if n > 1 else ""))
results = data["props"]["pageProps"]["searchResults"]
total = total or results["count"]
rows += [row(job) for job in results["jobs"]]
time.sleep(pause)
return total, rowstotal, rows = search("python-developer-jobs")
print(total, "jobs found;", len(rows), "on page 1")
print(rows[0])
shown = sum(bool(r["per"]) for r in rows)
print(shown, "show a figure; hidden by flag:", {f: sum(f in r["hidden"] for r in rows) for f in HIDDEN})It printed 984 jobs found, 25 on page 1, and the first row: a hybrid Python developer job in Essex at £500 per day, posted on 28 October and open until 9 December. The last line says 19 of the 25 rows show a figure, 5 are flagged 16 and 1 is flagged 64.
A figure the posting does not show#
Reed lets an advertiser write "Salary negotiable" or "Competitive salary" instead of a figure. The posting then shows only those words, but the search data still carries a number in salaryFrom and salaryTo, and a flag in salaryDescription says it is not to be shown: 16 for "Salary negotiable", 32 for "Commission" (none in these 150 rows) and 64 for "Competitive salary". The field is a set of flags, so row() tests the bits with &, leaves the salary empty for a flagged row and keeps the flags in hidden.
Of the 150 rows, 22 were flagged: 9 with 16 and 13 with 64. 18 of the 22 had a figure in the data. One other row had neither a figure nor a flag. Counting every row with a figure gives 145 (97 %); counting only the rows that show one gives 127 (85 %). Do not read a flagged figure as pay: the posting does not show it, and some look like placeholders, such as 10,000 to 50,000 a year, which came up twice. We opened 4 flagged postings, 2 with 16 and 2 with 64: each page showed only the words, and its JobPosting had no salary.
Keep the unit next to the number#
A column of salaries without units sorts by the size of the number: hourly rates end up at the bottom and yearly salaries at the top, whatever the jobs really pay. An average of it means nothing. Keep per beside salary_from and salary_to, and convert only if you have to, with your own idea of hours a day and days a year.
row() keeps codes it does not know as code N instead of guessing. We saw only 1, 2 and 5 in these 150 jobs. Other codes may exist, and you should check a new one against the JobPosting on its job page before you trust it.
Work patterns differ by job#
workingOption says on site, hybrid or remote. All 50 warehouse jobs were on site. Of the 50 Python developer jobs, 17 were hybrid, 27 on site and 6 remote.
Every day, at scale#
For many searches on a schedule, our Reed Jobs Scraper on Apify does the reading. Search by keyword and place, or paste search links. Each job comes back with title, company, agency or employer, place, county, the salary range with its period, job type, full or part time, remote or hybrid, and posting and expiry dates. With "Include posting details" on, which is the default, it also has the region, postcode, sector, skills and the description; switch it off for a faster run that reads only search pages. It is billed per result.
written
5 min readSubstack API in Python: every post of a newsletter with the archive cursor
Substack's RSS feed stops at the newest 20 posts. Its archive endpoint has them all, 25 a page, and it now pages by a cursor: an offset, even offset=0, has answered HTTP 400 since early November. The code that works, measured on three newsletters with 907 to 1,510 posts.
10 min readComputrabajo salaries in Python: half the postings show pay, and A convenir comes out as 0
We read 36 Computrabajo search pages in six countries and 24 job pages. 49 % of 706 postings showed a salary, always as one monthly figure, from 21 % in Argentina to 68 % in Colombia. Where the pay is A convenir, the structured data says 0. How to read the pages, and what to clean before you take a median.
12 min readWorkable, Recruitee, Personio and BambooHR job board APIs: no key, and a date that costs a request
Workable, Recruitee, Personio and BambooHR give a board's open jobs with no key, whole in one request; only BambooHR's list has no date. The four addresses, the catch in each, and about sixty lines of Python that put all four into one shape.