Skip to content
Beta

For recruitment agencies and professional associations

Turn your salary survey into a benchmark people use all year.

People in your sector answer a short chat, verify their email and see where their pay sits. Your team decides which answers count.

Request accessSee how it works

RefBench is in private beta.

An example, played in two panes. A visitor to Acme Recruitment's benchmark describes their role in their own words, gives their base salary, verifies their email with a 6-digit code and sees their pay against the median of 26 verified responses in Vancouver. In the console, their answer waits in review until a person approves it, which moves the group median from $71,500 to $72,000.

Example figures for an invented company. The chat is shortened here.

You set it up. Your sector fills it in.

Two short journeys that meet in one place: answers count once your team approves them.

Your team

People in your sector

  1. 1Describe their roleIn their own words. The chat matches it to your roles, and they confirm.
  2. 2Answer a few questionsYour questions, one at a time, in your order.
  3. 3Verify their emailA 6-digit code, valid for 10 minutes.
  4. 4See where their pay sitsTheir pay against the median of people like them.

Approved answers become the benchmark. A group shows a figure once it reaches your minimum.

One comparison, from people like them

This is what makes people take part. After the chat they see where their pay sits, and every part of the answer says where it came from.

  1. 1Their pay against the medianIn money and as a percentage.
  2. 2The middle halfHalf the group earns between $64,000 and $83,500. Their own figure is the dot.
  3. 3How many people26 verified responses. A figure never shows for fewer than your minimum.
  4. 4Which groupThe role and the place it compares with. When that group is too small, it widens to a larger area and says so.
  5. 5When it was updatedFigures change as your team approves answers, and the date says when.
  6. Extras such as bonus or work arrangement show once enough people report them. Until then the result says how many it has.

How a figure gets published

The same four checks run on every answer, for every company. They are how the product works, not settings you have to find.

  • Verified by email

    Each person enters a 6-digit code sent to their inbox before anything they said is kept.

  • Approved by your team

    Answers wait in review. Auto-approve exists for clean, normal answers, and stays off unless you turn it on.

  • A minimum on every figure

    No figure shows for a group below your minimum, on any page or widget. It is 10 unless you change it.

  • Figures come only from answers

    The chat asks your questions and never answers with a number. No AI figure enters a benchmark or an export.

When a group is small, the comparison widens

It moves to the next group in your order that has enough answers, and the result says which one it used.

Senior Interior Designer, groups tried in order

  1. Victoria6 responsesToo few
  2. British Columbia36 responsesUsed
  3. Canada109 responsesNot needed

The result then says: “Not enough responses for Victoria yet, so this uses British Columbia. It narrows as more people take part.”

Example groups. You choose the order in Settings.

When no group is big enough, nothing is shown

The person sees how close the group is and can help it get there.

What your team sees

Where you have enough answers to show a figure, what is waiting for review, and what changes when you approve it.

  • Visitor funnelFrom the first visit to a shown result, with every button click.
  • Export to CSVApproved answers, with or without each person's email.
  • Delete by emailRemove everything one person gave when they ask.
  • A history of every changeWho approved, edited or deleted what, and when.
  • Roles for your teamAdmins, reviewers and viewers, each with their own access.
  • Several benchmarksOne company, a benchmark for each sector or survey.

Works with what you already use

Put the benchmark on your website, show your open roles beside each result, and bring last year's survey with you.

Paste into your site
<script src="https://refbench.ai/embed.js"
        data-benchmark="acme/interior-design"
        data-variant="card"></script>

Your website

One script tag adds the benchmark as a card, a bar, a tile or a chat button. Or link straight to its own page.

Refari

Connect your Refari job board, and each result can show open roles like theirs, the consultant to talk to and a link to set up a job alert.

Your files

Bring last year's survey as a CSV or Excel file, and export approved answers to CSV whenever you need them.

Made for the people who already run the survey

Recruitment agencies

You publish a salary guide once a year. Keep it open all year instead.

  • Each person verifies their email, and you see it with their answers.
  • Your open roles can sit beside their result.
  • When a client asks what a role pays, you have your sector's answer.
Request access

Professional associations

You run a salary survey for members. Let them add their pay any time.

  • Members see where they stand in a few minutes.
  • No figure shows for a group below your minimum.
  • Your brand, your questions and your reviewers.
Request access

Took part in a salary survey?

If a company's page sent you here, this is how your answers are handled.

Who sees your answers
The company named on the page sees them, with your email. Other companies on RefBench cannot see them.
What other visitors see
Only figures for groups that reach the company's minimum. Never one person's answers.
Removing your answers
Ask the company that ran the survey. They can delete everything linked to your email.

Questions

Anything else, contact us.

Do we need data to start?

No. Importing last year's survey helps, because those answers count from day one. You can also start empty and share the link.

Who sees each answer?

Your company sees each answer with the person's email. People are told this before they answer. Other visitors only see figures for whole groups.

What happens when a group has too few answers?

No figure shows. The comparison widens to a larger area if one has enough answers, and says so. If none does, the person gets a link to share and can ask for an email when it is ready.

Is any figure made up by AI?

No. Every median and range comes from approved answers, and no AI figure enters a benchmark or an export. If you turn on estimates for extras such as bonus, each one is labelled as an estimate and kept apart.

Can it go on our own website?

Yes. One script tag adds a card, a bar, a tile or a chat button. You can also link to the benchmark's own page.

Can we approve answers automatically?

Yes, if you turn it on. Only answers with no flags, in the normal range of a group that already shows a figure, are approved that way. The rest wait for a person.

What is the minimum, and can we change it?

By default a group needs 10 approved answers before any figure shows. You can set it anywhere from 5 to 25 in Settings, and every page and widget follows it.

Can we start now?

RefBench is in private beta. Companies on the beta list create their benchmark and invite their own team. To join, contact us and tell us about the survey you run.

Start with the survey you already have.

Import it, see where you have enough answers, and publish when you are ready.

Request accessSee how it works

RefBench is in private beta.