Methodology

Sources and collection

Listings are collected daily from three source types: LinkedIn job postings, Seek (Australia), and company career pages via an ATS aggregation feed. Each listing is deduplicated within its source.

Counting unit

The index counts active job listings, not unique vacancies. The same role advertised on more than one board can appear more than once. This counting unit is applied consistently across every week of the index, so week-on-week movements are comparable even though absolute counts can include cross-board duplicates.

Classification

Every listing is classified by function (sales, marketing, customer success, revenue operations, enablement, partnerships, product marketing, growth, presales, GTM engineering, account management) and seniority using a large language model, with a keyword prefilter removing clearly non-GTM roles first. Classification model versions are stored against each record. Roles classified outside GTM are excluded from all published statistics.

Salary data

Salary figures come only from amounts explicitly stated in job listings. We never infer or estimate a salary for a listing that does not disclose one. Validation rules: annual amounts between $60,000 and $750,000. Because only a minority of listings disclose salary, all salary statistics describe the disclosed-salary population, which may differ from the full market.

What a stated amount means is not always the same thing. A published range can be base, can be OTE-inclusive, or can be a total package including superannuation, and the listing rarely says which. Until a semantic compensation parser is in place we therefore describe the figure as an advertised annual pay band rather than as a base salary, and we do not derive on-target earnings from it. The top of an advertised band is not OTE: across ANZ GTM listings the median band ratio is about 1.14 times, where the real sales base-to-OTE multiple observed on ads that state both figures runs 1.44 to 1.66 times depending on the role.

Three limits on the current population, stated plainly because earlier versions of this page overstated them:

• Window. Listings from the last 12 months are eligible, but collection only began on 30 March 2026, so the window currently holds about four months of data. No Pointer salary statistic covers a full year yet.
• Agency listings.Recruitment agency listings are excluded from the weekly index's headline medians. They are not excluded from the role and function benchmark aggregates behind the salary pages: roughly 2,400 agency-fronted rows sit in that pool and advertise about $10,000 above the direct-employer median. Removing them is queued work.
• Currency. Headline weekly index medians are computed from AUD listings only. The role-level benchmark aggregates still contain roughly 300 NZD-denominated rows that are pooled without FX conversion, which biases the affected role pools slightly low in AUD terms.

Benchmark blending

Role benchmark cards blend three inputs on the base row: published ANZ salary guide ranges (the anchor, with a fixed baseline weight of 300), live disclosed-salary aggregates from listings, and community salary submissions (weighted by count). The live listing weight is capped: it enters as the smaller of the real sample size and a published ceiling, currently equal to the guide anchor weight, so listing data can never contribute more than half of a published figure however large the pool grows. Each card states the real number of listings behind its base row, or says it is a pure guide estimate. Cards whose role cannot be matched to a specific live pool stay guide-anchored; there is no whole-function fallback pool.

The OTE row takes no listing input at all.The listing aggregates' on-target-earnings columns are percentiles of the top of the advertised pay band, so blending them into OTE understated it by 45% to 67% on every Account Executive segment. The OTE row is therefore a curated guide estimate, adjusted only by community submissions once there are at least ten of them for that role. It is not blended with listings and is not updated daily. Compensation structure that we can read directly from ad text, namely stated base/variable splits, quotas and uncapped-commission rates, is published separately on the quota and comp structure page.

Separately from the guide estimate, we extract on-target earnings from the text of ads that state one. That is genuine observed OTE and carries a real sample size, but it is thin: about 1,250 GTM ads state an OTE figure at all and only about 440 state both a base and an OTE. Observed OTE is published for a role only where at least 30 ads support it. Below that the figure is suppressed rather than shown as directional, so several roles have a guide estimate and no observed median. Account Executive, SDR/BDR, Account Manager and Sales Manager currently clear the floor; leadership, Customer Success, presales and partnerships do not.

Exclusions from the most recent observed-OTE build, published so a sanity filter cannot quietly hide a problem: of 1,256 OTE rows considered, 18 (1.4%) fell outside plausible annual bounds and 251 repeat postings of the same advertisement were collapsed to one. Of 443 rows carrying both a base and an OTE, 431 were kept: 3 had a base out of bounds, 7 implied a multiple below 1.0x and 2 above 3.5x. That is 2.7% of paired rows excluded, all counted. Observed OTE is reported for direct employers only, in a single currency per pool, and excludes Pointer's own advertisements.

A guide-anchored figure has no sample size, because it is a spread across curated published ranges rather than a sample of observations. Where no count can be shown we say the figure is a guide estimate instead of printing the anchor weight as though it were a number of data points.

Sample-size rules

In the weekly GTM index and our machine-readable data feed, salary statistics computed from fewer than 30 disclosed salaries are suppressed at source and never published. Statistics based on 30 to 49 data points carry a directional label. Statistics based on 50 or more data points are published without qualification. Role benchmark cards are guide-anchored blends (see above) and remain available at smaller live samples because the published guide ranges carry the estimate; those cards are labelled as estimates rather than given an N. Skill trend statistics in the weekly newsletter require at least 100 mentioning listings.

Weekly index snapshots

The weekly index is computed once per week and stored insert-only: a published week is never recomputed or overwritten, so any cited figure can be audited against the stored snapshot. Company hiring leaderboards exclude job boards, recruitment agencies, outsourcing and BPO firms, and anonymised advertiser placeholders.

Known limitations

• Salary disclosure is voluntary and varies by source, so disclosed-salary medians can move with source mix as well as the market.
• Cross-board duplicates inflate absolute listing counts; trends are comparable because the counting unit is constant.
• LLM classification has a non-zero error rate on ambiguous titles.
• Coverage skews toward Australia; New Zealand volumes are smaller.

Corrections

If a published number is found to be wrong we correct it in the next weekly issue and say so. Questions about methodology: ricky@pointerstrategy.com.au