Money for time: the real cost of job searching

The real cost is not the subscription. It is the waiting.

Here is the thing that people forget when they look at a monthly charge for a job search tool and wince.

Every day you are out of work, or stuck in a job that is draining you, you are paying. Mentally. Financially. In opportunity cost. In the slow erosion of your confidence and your savings.

I would much rather pay whatever token costs I need to reach the destination sooner. And I am saying that as someone who is genuinely happy to spend 5 to 10 pounds per day, not per month, if it means I get hired faster.

Think about it honestly. Would you rather stay unemployed for three more months by saving a few quid on tools? Or would you rather get the job next month and start bringing home 3,000 pounds a month?

That gap is not a few pounds. That is 9,000 pounds of lost income. Economists call this opportunity cost. I call it obvious maths.

The elephant in the room

To build a tool that actually helps you beat that waiting game, we have to talk about what it takes under the hood.

Let me be upfront about something. The biggest cost of running OutRung is not servers, databases, or hosting. It is AI model tokens.

If you have been living under a rock, here is the short version. The good language models, the ones that actually produce quality output, cost roughly $2.50 to $5 per million input tokens and $15 to $30 per million output tokens. And our work is token intensive. Especially on the output side, because we ask the model to reason through its choices rather than just spit out an answer. That reasoning step is a proven method for better quality results, and I refuse to cut it just to shave costs.

We have already optimised where we can

I am not just passing costs through without thinking. We have done a lot to keep things lean.

• We use a lower quality model for initial screening steps where full reasoning is not needed.
• We reduce output tokens where possible, without sacrificing the reasoning that makes results good.
• We use Python logic wherever we can, rather than relying on the LLM for every decision.
• We use prompt caching, which brings input token costs down to about 10% of the standard rate.
• We have tested the top three model providers (OpenAI, Google's Gemini, and Anthropic's Claude) and settled on one because you genuinely see quality differences between them. You cannot just pick the cheapest and hope for the best.

Why usage based pricing is the fair model

So OutRung does not charge you a flat rate. It charges based on usage.

This means you can go harder when you need to. If you are actively searching, scoring jobs, generating tailored CVs, and tracking applications across multiple roles, you pay for that intensity. If you slow down because you are waiting on interviews, your costs drop too.

In almost all cases, spending more to move faster is worthwhile. The return on investment is not abstract. It is your next salary minus whatever you spent getting there.

A caveat, because I am not going to pretend otherwise

If your industry is truly shrinking, or there are simply no roles being posted, then no amount of AI or effort will overcome that. You hit diminishing returns eventually. I am mostly speaking from experience in the AI and tech space, where opportunities are genuinely plentiful right now. If you are in that world, the bottleneck is almost never a lack of jobs. It is a lack of focus, a lack of tailoring, and too much time wasted on roles that were never a good fit.

OutRung helps with all of that. Not by magic, but by giving you a proper command centre for your search, so every hour you spend actually counts.