Link building is the area of SEO with the widest gap between what works and what is sold, and a new site is the most common target for the things that do not work. This is the honest version.

Start by accepting the timeline

A site with no history earns links slowly, because nobody knows it exists. There is no technique that compresses this, and every product marketed on the promise that it does is selling placements on sites that exist only to sell placements.

That matters more than it sounds. Google's spam policies treat buying links, excessive exchange, and large-scale article or directory submission as link schemes, and enforcement is applied to the site receiving the links. A new site with a sudden influx of links from unrelated low-quality sources is a recognisable pattern, and it is the pattern the systems were built to catch.

What actually earns links

Publish something that has to be pointed at. This is the whole game, and it is why most link building fails: you cannot earn links to a page that duplicates what fifty other pages already say. Things that get linked are original data you collected, a tool that removes a specific annoyance, documentation for something poorly documented, and thorough writing on a subject nobody has bothered to cover properly.

Be the best available answer to a narrow question. Narrow is the operative word for a new site. You will not out-resource a major publisher on a broad topic. You can comfortably beat everyone on something specific enough that nobody with a budget has bothered.

Do things offline that produce online mentions. Sponsor a local team, speak at a meetup, run a workshop, join a trade association. These produce links from organisations with real websites and real reasons to mention you, which is exactly the profile that is impossible to fake.

Be useful in places where your audience already is. Answering questions properly in forums and communities, over months, produces referral traffic and occasional links. It also fails immediately if you treat it as link building rather than as participation.

Methods that still work, used honestly

• Guest posting on sites with a real audience, where the editor would reject a bad piece. If the site accepts anything and charges a fee, it is a paid link scheme wearing a byline.
• Broken link outreach: find a dead link on a page in your subject, and suggest your equivalent resource as a replacement. It works because you are solving the other site's problem, not asking a favour.
• Responding to journalist requests through services that connect sources to reporters. Slow, competitive, and occasionally produces a link no amount of money would buy.
• Genuine directories and industry listings, meaning ones a human reviews and that people actually browse. Trade bodies, local chambers, specialist listings in your field. A handful of these is normal and unremarkable; five hundred automated submissions is a footprint.

What to avoid

Private blog networks, bulk directory submission packages, comment link automation, "DA 50+ guest post" marketplaces, and reciprocal link schemes all share one characteristic: the linking sites have no audience. A link is a reference from one thing to another, and a reference from a site nobody reads describes nothing.

The most reliable warning sign is a seller quoting a quantity and a price. Nobody selling real editorial coverage can promise you fifty of them by Friday.

A sane way to measure

Counting links is the wrong measure. A better set of questions: are links arriving from sites in or adjacent to your subject? Are they from pages that get read? Are they surrounded by relevant text rather than in a list of unrelated links in a footer? Would you be comfortable showing the page to a customer?

Ten links that pass those tests are worth more than a thousand that do not, and the thousand carry a risk the ten do not.