Do you start selling/advertising the product before its complete

It is about the happy medium. The more people know about the product the more painful would be the fail (if the product fails).

On the other side starting to sell a product only after launch is even more dangerous - you can find out that there is no product/market fit and all your development endeavor will go down the drain.

My advice is to try to sell the product right now, but in rather small scope: if it's software maybe a smoke landing page will be enough. It will let you aggregate some interested potential customers.

And don't forget what Chris said: it depends on market, competition, target audience, etc.

UPD: the smoke landing page is to aggregate some contact information on interested customers, not to receive payments

Yes, i do (advertise). Simple product web page with some description. It takes time to rank in Google. Meanwhile I write blog posts about the product and some useful information not directly related to the product itself.

Need a whole lot of information before figuring out a recommendation on that. Space you're in? Competition? Target market/audience? Length of time to get to market with the product? Please share as much as you can.

If this is an actual software product and not SaaS, selling the product (in the US) before it exists violates federal law and if you take credit cards, violates the terms of your merchant agreement.

In the US, once you take an order you have 30 days to ship the product or send the customer a written notice giving them the option to cancel the sale. See the saga of Joe Sugarman and JS&A. In the 1980's a giant snow storm in Chicago resulted in non delivery of items ordered from his firm. The Federal Trade Commision gave him giant fines.

One more major consideration- if you offer to sell or sell your software more than one year prior to filing a patent application, your software becomes unpatentable. Offering Software For Sale Can Trigger A Bar To Patentability

This may sound like a punt, but I would really recommend spending some time familiarizing yourself with the "lean startup" movement and "customer development" concepts: start by reading Eric Ries, Steve Blank, Dave McClure, and Sean Ellis. You want to validate your concepts as early as possible, and that means talking to potential customers using whatever you have: idea, mockups, working prototype. Until you have validated your concept and feel confident that you have reached "product market fit", I think that it is best to limit any advertising spend to the kind of SEM landing page "smoke tests" others have mentioned.

NOOOOOOOOOO! NEVERRRRRRRR!

I know that a lot of people, including popular authors, will tell you that this is a good idea. It is not and the reason is simple. Your product will have problems that will take time to fix. I don't care how much customers might say that they understand you are in early development and that they will work with you, they won't. The first technical problem, the first release delay, and the countdown to doom will start.

You can definitely create buzz by pre-selling your product, but I can assure you with 99.99% assurity that it will be bad buzz. There is a reason that company's like Apple hide everything until it is nearly rock solid. That way, the customer never gets to see all of their mistakes off the blocks and Apple's rep remains intact.

I am saying this from experience... very recent experience. There are only two ways to explain what happened to me when I created a lot of buzz prior to having my tech ready: either I really suck at launching a product (possible but not probable) or customer's suck at being patient when things go wrong.

Hope this helps.

It really depends on the situation. These are just my opinions though and there are many pros and cons to both sides.

The big thing, to me, is to not lead anyone on about the status of the product. If the potential customer is thinking two-week availability when the product isn't alpha yet, then you will have issues. Remember that any sales you make, that you can't deliver on, mean that you will wind up not only refunding the money but will have increased liability. If they know that it will be at least X time before you can deliver, and it's spelled out in the agreement, then there should be no legal issues.

If it's something where you would have a lot of lead time OR you are right on the cusp of completion, then I say go for it. If you have the research and product-market fit, then there's less likely to be a problem. If the tech is a little, um, "flexible" then it is likely a bad idea to sell.

If you're grant-writing or aiming for government contracts, it's very possible to sell the product before completion because of large lead times that they expect. But, again, you open yourself up to major liability on the off-chance that you can't deliver. Be very sure before you go that route.

As for advertising, don't bother with paid advertising until the product is almost ready to ship. That is money that will be better spent elsewhere. However, responding to reporter inquiries can be a really good idea to build initial interest. With most reporters, you can leave out some details and be a bit vague without raising eyebrows. It's a good feeling to be able to say "Our ground-breaking (whatever) has been featured in (where ever)."

I can talk about software products (or services). I can only tell you one thing: release a free beta very soon. You'll discover a lot of things that are broken, that are totally useless or simply are too complicated to be understood by the user. Mind you, the beta should be decent enough to be used without any major issue, but it doesn't need to be feature-complete.

Getting early user feedback can save you a lot of time, and will surely improve the quality of your product.

If you are worried about pissing beta users or turning away potential customers, know this: if you're on the market, even in beta, they will at least consider you, otherwise they'll just go somewhere else.

Other than that, offer something back to beta users, such as a discounted price on the final product.

Like a number of commentators here I'd also have to say 'it depends'. Certainly, selling something you don't have and that's not yet available is a bit of a grey area (pre-sales signup, maybe with a deposit depending on the product, is ok. Outright sales with an expectation of receiving goods when there's no product .. not so much).

However, gaining some fast feedback; a few friendly customers; early beta program or any other way to both get some traction and help iron out problems that are going to arise once the full launch comes are usually very positive (caveat: depends somewhat on the market/competition etc). Good to help tune your business model too.

Sometimes though the 'Gabbo' approach works really work. Just make sure you don't then let a Cuil out of the door!