Skip to content

100 Job Strategies · 45 of 100

The Public Dataset Build

Publish the data your industry wishes existed and the industry will come looking for you.

10,004 viewsHigh effortPays off in 4-9 monthsAnalysts and researchersCareer switchersConsultantsAnyone with no network
Share

In short

Every field has questions everyone asks and nobody has answered with data — what things cost, what people are paid, who the vendors are, how the market is structured. Compiling and publishing that dataset makes you the only source, which means practitioners, journalists and hiring managers find you rather than the reverse.

The situation

Every few weeks, in some industry forum, someone asks what the going rate is for a particular kind of contract work. The replies are anecdotes: I heard someone got this. My last client paid that.

Nobody knows. There is no source. The question gets asked again the following month.

Then somebody spends six weeks collecting three hundred data points, publishes a clean breakdown with methodology, and puts it online for free.

Within a year, that page is what people link to when the question comes up. The author did not become famous. They became the citation.

Why this works

Most industries run on anecdote in places where data would be obviously useful. Pricing, salaries, vendor comparisons, market structure, supplier performance — these are things everyone wants to know and nobody has compiled, usually because it is tedious rather than difficult.

Publishing that dataset produces a specific and durable kind of visibility. Unlike a blog post that is read once, a reference resource accumulates links, gets cited in conversations, appears in search results for years and becomes the thing people share when a question comes up. You stop competing for attention and start receiving it passively.

It also functions as an unusually strong work sample. Building a dataset demonstrates the whole chain: defining a question, designing a methodology, gathering messy real-world information, handling ambiguity, and presenting findings clearly. That is a more complete demonstration of analytical capability than any portfolio exercise.

The authority effect is what converts it. Being the person who compiled the definitive numbers on a niche makes you a known quantity to exactly the people who work in it. Journalists cite you, practitioners reference you, and hiring managers encounter your name in a context where you are the expert rather than an applicant.

The barrier is that it is genuinely laborious and nobody is asking you to do it. That is precisely why the space stays empty.

How to run it

  1. 1

    Find the question your industry keeps asking

    Look in forums and communities for questions that recur and never get a sourced answer. Recurring anecdote is the signal that data is missing.

  2. 2

    Scope it small enough to actually finish

    One narrow question answered thoroughly beats an ambitious survey abandoned halfway. Two hundred solid data points is a resource; two thousand planned is a fantasy.

  3. 3

    Use legitimate public sources

    Published listings, public filings, voluntary surveys, official registers, freedom-of-information requests. Never scrape private data or breach terms of service.

  4. 4

    Publish the methodology alongside the numbers

    Sample size, collection period, limitations, what you excluded. Transparency about weaknesses is what makes the rest credible.

  5. 5

    Make it free, linkable and easy to cite

    A stable URL, clear charts and a downloadable dataset. Anything gated or awkward to reference will not spread.

  6. 6

    Update it periodically

    An annual refresh turns a one-off into an institution, and each update is a fresh reason for the industry to look at your name again.

What to say

Copy, then make it yours
Hi Johanna, I published the first version of the UK freelance rate survey for technical writing last month — 340 responses, broken down by experience, sector and contract type. It's here: [link] I built it because the question comes up constantly in our community and the only answers available were anecdotes. I'm mentioning it because I'm starting to look for a permanent role, and I gather you lead the documentation team at Meridian. The survey is probably a better demonstration of how I work than my CV is — designing the methodology and handling the messy parts was most of the effort. If you're hiring, I'd welcome a conversation. — Ben

When it does not work

  • Scraping data you have no right to. Breaching terms of service or collecting private information turns a credential into a liability. Stick to public and voluntarily-provided sources.
  • Thin methodology. A dataset with an unclear or weak method gets picked apart, and the criticism attaches to your name. Be conservative and transparent.
  • Abandoning it halfway. A partially-built resource published as complete damages credibility more than publishing nothing.
  • Choosing a question nobody asks. Data on something the industry does not care about produces a tidy document and no attention.
  • Publishing salary data that identifies people. Aggregate carefully. Small samples can inadvertently expose individuals, which is both unethical and potentially unlawful.
The takeaway

Applications compete for attention. A reference resource receives it — for years, from exactly the people you want to reach.

Questions

What kind of data is worth compiling?

Anything your industry asks about repeatedly and answers with anecdote. Pricing and rate benchmarks, salary breakdowns, vendor or tool comparisons, market maps of who supplies what, and performance or reliability comparisons are all common gaps. The reliable signal is a question that recurs in community forums every few months and never receives a sourced answer, because that indicates genuine demand combined with nobody having done the work.

Where do I get the data legally?

Published listings, public company filings, official registers, freedom-of-information requests where applicable, and voluntary surveys you run yourself are all legitimate sources. What to avoid is scraping data in breach of a site's terms of service, collecting personal information without consent, or using information you obtained confidentially through employment. The legal and ethical route is usually sufficient, and building on a questionable foundation converts a credential into a liability.

How large does the dataset need to be?

Smaller than most people assume, provided the methodology is transparent. Two or three hundred well-collected data points on a narrow question, with clearly stated limitations, produces a genuinely useful resource. A thorough answer to a small question is far more valuable than an abandoned attempt at a comprehensive one, and scoping too ambitiously is the most common reason these projects never get published at all.

How do I avoid exposing individuals in salary or rate data?

Aggregate carefully and suppress small cells. If a particular combination of role, region and seniority contains only two or three respondents, publishing that breakdown can effectively identify them, which is both unethical and potentially unlawful under data protection rules. The standard practice is setting a minimum group size below which you do not report, stating that threshold openly, and avoiding any combination of variables granular enough to single someone out.

Build your free Drivetube Profile
Share

More from 100 Job Strategies

See all 100 in this series