Why the cheapest proposal and the honest proposal are rarely the same document
We lost a deal this week. An energy company, an on-premise AI project, the kind of work we are good at. We came in over their budget, and they went with someone cheaper. I want to write about it while it still stings a little, because the reason we lost is the same reason I would bid it exactly the same way again.
Their tender came with a hardware baseline. A single GPU, speced to run the whole system. We read it and told them, in writing, that it was undersized for a real production AI workload. Not for the demo. Not for the first month. For the version of the system that has to hold up once real people depend on it every day. So we priced for hardware that would actually carry the load, and that made our number bigger than the one they had in mind.
You can always get to a smaller number. Quote the box they asked for, win the deal, and let the reliability problem show up later as their problem, not yours. It is a common move. It is also the move I have spent twenty years refusing to make.
The cheapest number is a story about the first month
A cheap proposal is not usually a lie. It is a selective truth. It tells you what the system costs to stand up, and stays quiet about what it costs to keep standing. The under-speced machine runs fine in a demo with a handful of test users. It struggles the day the workload is real. By then the contract is signed, the budget is spent, and the person who sold you the cheap number is somewhere else.
The honest proposal is a worse story to hear and a better one to live in. It prices the whole life of the thing, not the launch. It says the uncomfortable part out loud: this will cost more than you were hoping, and here is exactly why. That is a harder document to sign. It is also the one you are glad you signed in month six.
Why I would rather lose this way
Losing a deal is not free. There is real work in a proposal like this, and there is the plain fact of revenue walking out the door. I am not going to pretend that feels good. But there is a worse outcome than losing a deal, and that is winning one by hiding what I know.
If I quote hardware I have already told a client is too small, one of two things happens. The system underperforms and they blame us, correctly. Or it underperforms and we quietly spend our own margin fixing what we should have speced right in the first place, which means I lied my way into a loss. There is no version of that trade that ends well. The deal I lose by being straight is cleaner than the deal I win by being clever.
The number that makes this pay off
This is not only principle. It is how the business actually works. Close to 90% of ScalaCode's clients come back to us for their next project. That number does not come from winning every deal. It comes from the clients who did hire us finding out that what we told them up front was true. The spec was right. The timeline was real. The uncomfortable thing we said in the proposal is the thing that saved them later.
You cannot build a repeat rate like that and also be the cheapest quote in every room. The two goals pull against each other. Cheap wins the signature. Honest wins the second project, and the third, and the referral that arrives without a pitch. Since 2012 I have bet the whole company on the second one, and the number on my homepage is the receipt.
What to do with this if you are the buyer
If you are the one holding the tenders, here is the practical version. When one proposal is much cheaper than the others, do not treat the gap as savings until you know what it is made of. Ask what the cheap one assumed. Ask what happens in month six, not week one. Ask the expensive bidder to show you exactly where the extra money goes, and make them defend it. A good partner will have a clear answer, because they priced a real system on purpose.
The cheapest proposal and the honest proposal are rarely the same document. Sometimes you find one that is both. More often you have to choose, and the choice is really about when you want to find out the truth. Now, in the price, or later, in production. I would rather you find out now. That is why we lost a deal this week, and why I would send the same proposal tomorrow.
If you want a team that prices the real system and stays to see it through, that is what we do. See how ScalaCode works, or get in touch.