Reading a competitor's strategy from its hiring

A competitor keeps its roadmap private, but its open roles are public: each names a function, a level, a place and often a tool. Read across all its postings and over a few months, hiring data shows where it is investing before the product, the pricing or the press release does. This guide explains what each part of a posting tells you, which datasets to read, and how to turn them into a monthly competitor brief.

What a posting reveals

Each reading is a dated observation that you read beside the others, and a single posting says little. The signal is a change in the mix over weeks. The hiring dataset holds every column named here, and the data dictionary defines them. It has four parts: Job Postings, with every role a company publishes; Hiring Activity, with daily open, new and closed postings; Sales Team Metrics, with weekly sales-organisation measures; and Sales Pay Benchmarks.

Firsts carry more information than volume. A company that opens its fortieth engineering role has told you little. One that opens its first sales role in a new country, its first enterprise-segment posting or its first director in a function has changed something. Sales Team Metrics records two of these firsts directly, and for the rest you compare the mix with the month before.

Four of the ten role flags repay a column of their own in the brief. ai_role marks AI roles, which ai_postings counts for each day. new_initiative marks a role tied to something new. urgent marks an immediate start or an urgent hire, and a cluster of them suggests capacity pressure. budget_owner marks a role that owns a budget. Flags are set on firm readings, so a set flag is a reliable marker to count.

Setting up the watchlist

List the competitors by company ID, with the brands and subsidiaries that publish boards of their own, and look at what each one publishes before you read anything. Brands and subsidiaries carry the identifiers of their listed parent, so group their rows by isin when you want a group's hiring as one. The query counts each competitor's postings in Job Postings by board.

select
  company,
  board_platform,
  count(*)           as postings,
  min(first_seen_at) as first_read
from job_postings
where company_id in (:competitors)
group by company, board_platform
order by company, board_platform;

The result shows which competitors publish a board and how many postings each carries. A competitor with a few rows may publish most of its roles somewhere else, so read its postings beside its announcements. The earliest first-seen date marks the baseline observation of the board, and every posting on it that day is a baseline and not an opening.

A monthly brief in six parts

1. Size and direction. Open postings at month end, and postings opened and closed in the month.
2. Mix. The function and seniority of postings opened this month against the month before.
3. Places. The countries with postings opened, and the countries new to its sales hiring.
4. Go-to-market. The sales share of open postings (open_sales over open_all), the segments, upmarket and the median on-target earnings.
5. Tools. The tools named in postings, and the movers since last month.
6. Against the sector. The same measures for the competitor's industry from the weekly market series (hiring_new, function_share, seniority_share), so that a competitor hiring faster than its sector is told apart from one rising with it.

Write one line of reading under each part and end the brief with the decision it implies for you. The first query counts the postings a competitor opened in a month by function and seniority. The second reads its sales organisation week by week, and the third gives the sector baseline from Market Series for part 6.

-- JSON operators differ by engine; this is the PostgreSQL form
select
  labels ->> 'job_function' as job_function,
  labels ->> 'seniority'    as seniority,
  count(*)                  as opened
from job_postings
where company_id = :competitor
  and not found_on_first_read
  and first_seen_at >= :month_start
  and first_seen_at <  :next_month_start
group by 1, 2
order by opened desc;

select
  week_start, open_sales, open_all, new_sales, closed_sales,
  by_segment, new_countries, upmarket, median_ote_usd
from company_sales_weekly
where company_id = :competitor
  and week_start >= :month_start
order by week_start;

-- :industry is the industry id the dashboard shows for the competitor
select as_of, count, cohort, rate, growth
from market_series
where metric = 'hiring_new'
  and dimension_kind = 'industry'
  and dimension = :industry
  and window_days = 30
order by as_of;

Set the competitor's postings opened in the last 30 days against its own previous 30 days, and compare that change with growth for the industry. A competitor up 60 per cent in an industry up 5 per cent is changing course. One up 8 per cent in the same industry is not.

The exclusion of postings found on the first observation matters. A posting already on a board at the baseline observation was not seen opening, so its first-seen date is not its opening date, and it is not counted as new.

A worked reading

The figures here are illustrative. Acme Robotics, an invented competitor, opens 40 postings in a month: 22 in software_engineering, 8 in data_science_ml, 6 in sales and 4 elsewhere. Four of the six sales postings are account_executive roles in the enterprise segment, new_countries lists DE and FR for the first time, and a data warehouse is named in nine postings.

The brief for that month reads as follows.

Read together, that is a product-led company building a data capability and opening an enterprise sales motion in Europe. The decision for you is whether your European enterprise accounts are about to meet it, and whether your own hiring in those countries is ahead or behind. For a closer reading of the sales organisation, see tracking sales team build-out.

How to read the postings

• Evergreen and reposted roles. Some roles stay open as a standing pipeline, and some are closed and reopened. Compare open and new postings, and treat a jump in new postings with no change in open ones with suspicion.
• Baseline observation. The first time a board is observed, every posting on it is marked found_on_first_read and none counts as new, so the record of openings for a board begins with its next observation.
• Board changes. A competitor that moves to a new applicant tracking system can look as if it stopped hiring for a few days. A fall to zero followed by a return to the earlier level is a change of board, not a freeze.
• One country per posting. country is the first location's country, so a role open in several countries counts once, in the first. The locations list holds all of them.
• A plan to hire. A posting records an intention to hire. Advertised pay is the range in the posting and not what the company pays.
• Seasonality. Graduate intakes, budgets and holidays move postings, so compare the competitor with its sector, which shares the season, and not with last month alone.

Roles, then the other half of the picture

The data shows roles opened and closed, with function, level, place and tools: the shape of a competitor's investment. A closed posting records that the role left the board, and it may have been filled or withdrawn, so read closures as a trend across many roles.

Job Postings carries every role a company publishes on its own careers pages, and the blog explains why postings are read at the source. Each label is produced under a frozen version, so a brief compares like with like from month to month. What a competitor announces is the other half of the picture, and tracking competitor launches and partnerships at the source covers it.

Frequently asked questions

The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.