•
•

Topiclost a client because i quoted three weeks and someone else said four days (self.learnprogramming)

eingereicht von Specialist_Agent3599

Freelance, mostly laravel and vue for small agencies, seven years at it, and last month I lost a job I actually wanted because my quote was three weeks and the guy who got it said four days.

I know roughly what he's doing. I could probably do it in four days too, if I let the agent drive and didn't read most of it. The client isn't stupid either. He asked me directly why mine would take five times longer and I didn't have an answer that didn't sound like I was slow.

What I said was something about testing and edge cases, and it came out sounding exactly like what a slower person says.

My three weeks is mostly reading. That's where the time goes now. I run pest and coderabbit over everything before I even open the diff myself, which clears the mechanical stuff, and then I read it anyway, because none of that knows what the client's business actually does. The reading is the part I'm charging for and it's the part that's invisible on a quote.

A month later I'm still annoyed, mainly because I think he'll probably deliver something that works. It'll be fine for six months. Then it'll be somebody's problem, possibly mine when they come back and ask me to take it over.

I don't know how to price this anymore. Do you quote the fast number and eat the risk, or quote honestly and keep losing to people who won't

Alle 16 Kommentare
sortiert nach:
beste
TopNeuKontroversAltq&a
FormatierungshilfeVerstecke die HilfeNoun: Inhaltsrichtlinien

[–]misingnoglic [Bewertung versteckt]  (1 Kind)

You need to be better about explaining why your three weeks is better than someone else's 4 days. Maybe reach out to that person and see if 4 days really worked out.

[–]zomgitsduke [Bewertung versteckt]  (0 Kinder)

"I would hate for you to have a semi-working production crash at the critical hour and you come to me to emergency fix it at triple the rate"

[–]Primary_Concept_3147 [Bewertung versteckt]  (0 Kinder)

Explain to the customer the difference in in maintenance for a long term run between a system made in 4 days vs someone made in three weeks. If you make your customer understand that in the next iteration of the program they need to pay more because of poor quality of the system they would understand the importante of your time.

[–]PeteMichaud [Bewertung versteckt]  (2 Kinder)

You are competing with agents now, you have to figure out how to be faster. My main suggestion to stop spending 3 weeks tinkering with edges cases is to come up with a nice test harness / regime that ferrets out edges and tests them all.

[–]ichigox55 [Bewertung versteckt]  (0 Kinder)

mans competing with a bunch of buzzwords.

[–]Karyo_Ten [Bewertung versteckt]  (0 Kinder)

Or find new clients where bad quality = financial, regulatory or reputational damage

[–]CmpBreaker [Bewertung versteckt]  (2 Kinder)

Many of my clients were gained based on your scenario. Every time i was called after the short quote either did not deliver or when it broke. After five years i raised my rates significantly and continued for almost 30 years based on reputation alone

[–]Terreboo [Bewertung versteckt]  (1 Kind)

That’s pretty short-sighted. There’s a big difference between someone doing it in four days now powered by frontier LLMs and someone doing it in x amount of days for the short quote 30 years ago. At least the short quote back then actually had to do it all by hand. The whole landscape has changed now, it’s not even the same planet.

[–]CmpBreaker [Bewertung versteckt]  (0 Kinder)

You are probably correct on these types of small apps. I did avoid going into a long, detailed description. The focus was more on the general consulting market. There will always be competitors that undercut and, as such, it can't get under one's skin.

In my space, I worked on ever growing scale of applications and systems where at the beginning of my career the clients were smaller more local companies. After five years, I was working on global enterprise apps with 12-18 month deployments cycles. There were always competing bids from others for 30 days to 6 months. These frequently became my call backs due to non-delivery and tight remaining deployment dates.

So, I would say my comparison is more of scale than short-sightedness. However, the lesson of it not getting one down applies in general. Being a good consultant is a marathon and not a sprint. [pun intended]

[–]TheGrolar [Bewertung versteckt]  (0 Kinder)

There are a lot of ways to address this.

1) Better client screening. If the guy likes four days instead of verified code, he is probably never going to go for verified even when it breaks. His fault, so he'll blame something else, probably even believing that his reason is the real reason.
2) Follow up with the potential client six months from now. FInd out if the code actually broke. Don't do this expecting a job. You're trying to get a better sense of whether your initial hypothesis was right. If you can nail down what "good enough" looks like, nail it cold, then it gets a LOT easier to move into "really good", where you find the people who want that and charge them accordingly.
3) Be a LOT clearer about what might break and why. Your client may not care, but (here's the kicker) you make him FEEL like a boss because he "decided" whether to take on a risk based on his "boss decision making." You make him complicit, which is one of the most powerful consulting tricks. Those folks often become long term clients after a while.
4) Offer a warranty. If it breaks within six months (whenever) after implementation, you'll fix it free and return the fee. Ask innocently if Mr Four Days offers this kind of deal. Smile a little sadly. Of course, define clearly what "breaks" means. This will also make you look more like an expensive expert.
5) (the least important) Think about speeding up your process. This is mainly for your velocity, which is critical for a consultant for reasons too detailed to go into here. It's remarkably consistent advice from the many successful consultants I read, though.

[–]gusbo_the_jam [Bewertung versteckt]  (0 Kinder)

If anyone ever asks why something takes time, the correct response is exactly this - "Fast, cost effective, high quality. You can pick two."

[–]KedMcJenna [Bewertung versteckt]  (0 Kinder)

Just two words to tell them as a reason would be enough: "Quality assurance". Also a litmus test for the kind of businessman you're doing business with.

One note of caution about this: in 6 months' time from now, it's likely that the agents of 2027 will be producing code that won't almost inevitably need support farther down the line. No more than with human code at least, and likely better. If we get to the end of 2027 with the current 'AI code eventually breaks' being true, it'll be a surprise. So this whole QA assurance has a shelf life too.

[–]azian0713 [Bewertung versteckt]  (0 Kinder)

Why are you beating around the bush for why it takes longer?

It takes longer because you’re being more thorough and efficient. Full stop. You said your client isn’t stupid. If that’s true, they would know that testing debugging generally takes 3-5x as long as writing the code.

So yeah, if one person does no testing and the other does, the latter will be 5x as long.

You lost a client because you’re unable to communicate, not because ai is eating your lunch.

[–]abd53 [Bewertung versteckt]  (0 Kinder)

This is how I do it with my managers-

I can do this in 4 days by skipping x, y, z but that has a high chance of leaving critical vulnerability or safety risk and if any such ever surface, you might end up with million dollars lawsuit. OR, I could do x, y, z which would take another 15-16 days but you'll be sure that there is very minuscule, if any, chance of an accident landing you in a million dollars lawsuit.

[–]KingMoonfish [Bewertung versteckt]  (0 Kinder)

These ai posts are going to get uncontrollable aren’t they

[–]zomgitsduke [Bewertung versteckt]  (0 Kinder)

When the job comes back to you, it's more expensive to pick up someone else's work.