Tuesday you priced a reply. Today you cut the price.

Three levers pull your cost per reply down. Same connects, more replies.

Tuesday we priced a reply. About $22 in Engineering & Architecture, $71 in Web & Mobile. Same connects, different math. Upwork is already the cheap channel. And the most expensive Upwork category still beats everything off-platform:

You still asked how to make your number lower. Here is the honest answer. Three levers. Each one is the same idea: win more replies for the connects you already spend.

• Lever 1 · Bid clients who can pay. Filter for payment-verified clients with real spend history. Adds about +2.6 points of reply rate.
• Lever 2 · Bid the fixed-price jobs. Fixed-price posts reply better than hourly for most categories. Adds about +1.5 points.
• Lever 3 · Fix the cover letter length. The U-curve from Issue 002. Get out of the dead zone.

No single lever is magic. Stacked, they are. Web & Mobile replies at about 4% today. Lever 1 adds 2.6 points. Lever 2 adds 1.5. That's 8.1% before you touch the cover letter. Almost double.

We're stacking measured single-lever lifts, so treat the low 30s (in cost-per-reply terms) as the good end of the range, not a guarantee. The direction is not in doubt.

Now a warning. There is a fourth thing your team touches every day, and most people believe it makes replies cheaper too. It does the opposite. That one needs a story.

Boosting does not make replies cheaper. The night I proved it.

Here's one we argued about for weeks. The boost. You spend extra connects to push a proposal higher on the client's list. Half the team said boost every bid. The other half said boost nothing. Both sides were loud. Neither had checked.

So one night I pulled 13,331 of our bids. The ones where I could see the boost and the connects the job asked for. I sorted them by boost size. Then I read the table twice.

A bid with no boost replied at 8.35%. A bid with a small boost. Less than the connects the job itself asked for. Replied at 6.73%. Lower. You spend extra connects and do worse than free.

So both loud sides missed it. The data did not crown a winner. It found a loser. The half boost.

Now the part the cheaper-reply crowd doesn't want to hear. Boost is a paid feature. Every connect you spend boosting lands on the bill. Boost can lift your reply rate. The chart proves it. But it never lifts it by enough to outrun what it costs.

Boost still buys something real. Speed, and a higher spot in front of the client. On a job you truly want, that is worth paying for. Just know what you're buying. You're buying position, not a discount.

And if you buy it, buy it properly. Boost nothing and you keep your connects. Boost hard on the job that matters and you get real position. Boost a little and you pay for neither. That's why GigRadar's Smart Boost sizes the boost to the job.

Meta cut 8,000 jobs to fund AI. Your clients are watching.

On May 20 Meta began cutting about 8,000 jobs, close to 10% of the company. It also cancelled 6,000 roles it had planned to fill. The cuts hit engineering, design, and cybersecurity. Around 7,000 people moved onto new AI teams. Mark Zuckerberg's stated reason was to fund AI. Meta will spend up to $145 billion on it this year.

Read what that is, and what it is not. Meta did not hand 8,000 jobs to contractors. It did not outsource anything. It automated.

So why does it reach you. Meta is not your client. But Meta is the example your clients copy. The businesses that hire agencies are reading the same headline, and they are starting to ask the same question. What here can AI do instead.

The agency that keeps the budget sells the result, not the roster. It uses AI itself and runs lean. That is a repricing of what an agency is for, not the end of the agency.

A cheap reply is not a deal. Do not lose it on the call.

The three levers get you a cheaper reply. A reply is a person who opened a chat. It is not money yet. The call is where most of that work gets thrown away.

• One. Open with your real price.
• Two. When they flinch, do not reach for a discount.
• Three. Never end a call empty-handed.

If your team cannot say what last month's connects bought, this issue is the conversation to start.