As a startup you don't have the money to be fooling around with all sorts of expensive marketing stuff, especially the kinds that produce dubious results, right?
There are entire blogs that devote themselves to go through various topics of startup marketing. Here are a few:
http://www.startup-marketing.com/
http://www.balsamiq.com/blog/?p=198
Over the years I have learned (often through trial and error) what works in startup marketing and what doesn't. Here's my take on that:
The stuff that does not work:
1. Tradeshows / Conventions / Big organized events. These typically cost too much for a startup anyways. But even if you would get some government money or something and would be able to scrape together the budget to go there; even then they are too much of an effort for a startup to properly organize. They often require a full team of people to be there for multiple days, and that can ramp up quite a cost. Unless your product is very mature and already successful in some other market (and you are just bringing it over to market of the Tradeshow) I would not recommend attending. Not even if you get to "piggyback" with a bigger company inside their stand. Not worth it.

CC Attribution: TechShowNetwork@Flickr
2. Roadshows and trips for a series of meetings. Startups often go on one of these trips way to early; your product is nowhere near complete enough and all you end up doing is getting some lukewarm half-interested response that ends up wasting everyones time. If you are looking for partners and/or customers, there are better ways to do that than for example: to organize a series of 15 meetings during an expo in another country. The amount of followup you have to do, and the magnitude of discipline it requires will most often ruin this opportunity for a startup.
3. Paid advertising in print, tv, or "traditional media". Again startups do not have the budget, and it very rarely works. Have you noticed that just about 100% of advertising in traditional media is for products and stuff that is already very much complete, perhaps successful in another markets, or at least already being adapted by a significant customer mass? There's a reason for that: nobody launches and introduces their products in traditional media for the first time. Because it is simply a very ineffective place to do that. Stay away as a startup.

CC Attribution: Mrs Jackson@Flickr
4. Old fashioned sponsorships. These are designed for well-known larger corporations that already have a name and want to steer their brands more towards what ever it is that they are sponsoring. Nobody ever remembers a tiny unknown startup from the sea of sponsors the thing (what ever it is) is going to have anyways. Just about everything in this category is a waste of money and resources for a startup. Be very critical when anybody tries to sell you a sponsorship. The only kinds of events where that might actually work for a startup are something like www.slushhelsinki.com and even then; be critical in your assessment ;)
5. Big time "presence". Advertising agencies and such always claim that in their business the company office needs to have a certain kind of 'presence' and that it acts as a marketing tool for them. Maybe, who knows? For startups that certainly is not true. The focus on office space with startups should be your own team and employees, not marketing or really any outsider as such. Spending boatloads of green on your office deco just makes you look like an irresponsible spender in the eyes of who ever is financing you.
6. Fancy physical marketing materials. Another classic category of marketing that doesn't really turn into any incoming cash flow all that easy. Some startups notoriously spend a lot on making fancy and glossy marketing materials. I have never found this to be so useful - the material is outdated in a few months anyways, and I am yet to encounter a person that had made a purchasing decision ("I will buy something from this startup") based on the glossy materials the startup had. You can make materials, just make modest ones with actual information and substance in them. Not A3 sized pictures of your pretty office manager and a customer reference list that is hopefully going to be outdated in a few months anyways.
Stuff that does often work:
A. Going for leverage. Leverage = small bets, potentially huge gains if you get it right. Do small creative things. Lots of them. Figure out what works and do more of it.
B. Make yourself easy to be found. One of the best marketing things a startup can do almost for free is to make itself and its products as easy as possible to be found and discovered. Try to always think in terms of "how will the customer find us?", "where can we be discoverable where they are looking?" etc. Use what ever comes to your mind and followup on the different sources on how you customer actually did discover your; was it through Google? Through the company website address in your office window?
C. Great product is your best commercial ever. If you have good enough product you will be quite successful in turning all of your customers into active PR people and advocates of your product and brand. Just look at Apple how that's done. Getting people to passionately talk about your stuff only requires that it is truly excellent. www.IRC-Galleria.net originally spread like wildfire in the Finnish upper secondary school system (or high school) because of word of mouth, which was the direct result of people talking about what's going on there and how active the place is. Money spent on getting the customer experience right with your product is money spent on marketing.
D. PR and articles. This is the way to go for a startup. Usually requires no money, just plenty of planning, effort and time. Try to get as much free PR and articles out there as you can. Be open about what you do, be visible, be easy to discover, be loud, etc. Talk to journalists, issue press releases on important stuff - research and find out how this stuff is done and possibly get a partner to do some of it for you. Some of the recent PR successes in the USA by finnish startups have been possible because companies like www.tripsay.com or www.xihalife.com use a local PR agency there, that does this stuff for them. www.dopplr.com does their own PR, and are very successful in doing it. www.muxlim.com works with a Swedish partner on this. Hiring somebody to help you might increase the cost, but often it is worth it for a startup: quite minimal cost, and potentially big rewards.
E. Barter deals. Barter deal is where you exchange something (not money) to get advertising or something comparable to advertising. If you are a big huge website like www.irc.fi then this is something you can use very effectively to get all sorts of visibility stuff for zero money. Other startups that start to have some traction can also try to go with barter deals in different kind of things; banner exchanges, having your logo there, having your intro there in the partners newsletter, integrating a link to your partners website, etc. RunToShop does this frequently with many partners and it is both cheap and quite effective; gets you attention and allows customers to discover you.
F. Viral campaigns. Viral campaigns are perhaps more suitable for startups than anything else. Try to get as many of these going as possible. These are potentially almost zero money campaigns that may yield unexpectedly huge rewards. In 2007 iLike managed to get some heavy leverage going by jumping on the virality of the new Facebook apps and they notoriously got 300.000 new users every day and went from small numbers to 6 million users in no time at all. How to repeat that? Think about what could "go viral" with your product, and then try to make that happen. Be creative and try different things. Usually the things that utilize a paradigm shift just like iLike (eh..) did there are the ones that hold the highest potential virality.
G. SEO/SEM. Optimizing your search engine traffic and spending some on search engine marketing is "teh shit" these days, isn't it? The thing about SEO/SEM is that it really seems to work, and it is cheap enough for just about any startup to benefit from it. Do what you can yourself, but I recommend finding a good partner with references from your specific field to join forces with you. A good SEO plan takes at least 6 months to execute and for the results to show, so don't give up early.

CC Attribution: ByronShell@Flickr
H. Be your own brand builder. Talk about your stuff. Use every existing free service out there. Register to places such as www.stumbleupon.com, www.getsatisfaction.com, start a company Twitter, a blog if have the time (and somebody eager to write it). Mane a www.crunchbase.com profile for your company. Etc. Have conversations about your startup with people. Ask for feedback and opinion. Participate in industry events and meetups for some networking. Try to get into your customer segments' events. RunToShop team visited "ChildFare" this year for only a few hours with a mission to talk to potential partners exhibiting there. The result? Tens of contacts, about 10 signed partnership deals and a lot of attention for Run. Besides it was fun to talk to people while they were trying to put a fake tattoo on some kids arm for example.. a welcomed change for the daily startup life ;-)
I. Competitions, fun, and modern sponsorships. Competitions and fun stuff with startups work pretty good. Consider holding a couple of good compos. Preferably with prizes that you got free (or bartered) from your partners. Organize parties for the audiences that are important to your business. Consider sponsoring something in a non-traditional modern way. As an example: www.muxlim.com is sponsoring the TechCrunch Brunch at Slush.
---
..and that's it! I ended up with quite a list. The most important thing in startup marketing is to be agile; to decide-do-measure-evaluate-change and do that cycle over and over again rapidly. This is a bit of a stupid slogan but it really works: "do more of the stuff that works, and less of the stuff that doesn't".
Hi. My name is Taneli Tikka. This is where I preach what I practice. I'm a
serial entrepreneur and a startup activist of sorts. People usually know me
from my past and present consumer Internet service projects: IRC-Galleria,
Dopplr, Muxlim, StarDoll, RunToShop, Vakuutuskone.com, and a bunch of other stuff. My
"proper" bio is behind this link. Glad to see you here, thanks for browsing
around.