How to get the first backlinks for a brand-new domain
Links & Authorityeffort Mimpact 3/5
A new domain has no links because nobody has a reason to cite it yet, and outreach sent from that position mostly goes unanswered. Directories are the one link source that does not need anyone to say yes.
- Collect the free public directory lists first; several maintained ones on GitHub carry domain rating, pricing, and follow status per entry, so a paid list adds nothing.
- Screen before you submit anywhere. Keep a directory only if submission is free, the directory itself ranks for at least one keyword, and it has 50 or more referring domains.
- Build your submission kit once. With logo files, screenshots, and descriptions in three lengths ready, each form afterwards is a ten-minute paste.
- Submit to 30 or 40 of the screened directories over two weeks, and log each submission with its date in a sheet.
- A week later, open each approved listing and read the link in the page source; an approval email does not tell you whether the link went live or whether it carries nofollow.
- Stop when your domain rating stalls; the plateau tends to arrive in the mid-teens. Past the plateau, the same hours pay better in a tool or a data page that people link to on their own.
Why it works
Directory links are individually weak, and Google discounts links anyone can create, so the ceiling is real. But a brand-new domain is not fighting for weight; it is fighting a total absence of signals, and directories end that absence with crawl paths, brand mentions, and enough early authority that later outreach gets answered.
Evidence
Merging the 50 public directory lists in our research produced 2,821 unique directory domains; 1,066 accept free submissions, and 1,628 pass the screen in step 2.
How to get your site cited by ChatGPT and AI Overviews
AI Search (GEO)effort Simpact 4/5
A growing share of your audience asks an assistant and reads its handful of citations instead of ten blue links. Which pages get cited follows different rules from which pages rank, and your rankings do not transfer.
- Pick the one page whose question you most want to be the answer to, and write a direct, factual answer into its first 150 words.
- Back that answer with at least two statistics, each attributed to a named source.
- Add one short quote from a named person with stated credentials.
- Define the page's key entities, such as the product and the metric, in plain sentences near the top.
- Put a visible last-updated date on the page, and change it only when the content changes.
- Verify the site in Bing Webmaster Tools, and check that robots.txt is not blocking bingbot or GPTBot; ChatGPT's browsing runs on Bing's index.
- Re-ask your target question in fresh assistant sessions every two weeks and log whether you appear, because citation changes arrive much faster than ranking changes.
Why it works
Assistants assemble answers from passages they can quote with confidence, and a self-contained answer block with named sources and a visible date is the cheapest passage to quote. In Ahrefs' crawl data, only 12 percent of the URLs ChatGPT and Perplexity cite rank in Google's top ten. So a page that never reaches the first page of results can still be the source an answer names.
Evidence
A Princeton study of 10,000 queries in our research measured a 30-to-40-percent visibility lift in AI answers for pages that added statistics, named sources, and expert quotes; pages sitting around position five gained up to 115 percent.
How to do programmatic SEO without tripping Google's scaled-content policy
Content & Programmaticeffort Limpact 4/5
Programmatic pages built some of the largest sites on the web, and since Google's March 2024 scaled-content policy they are also what gets sites flattened. The difference between the two outcomes is whether each page shows something that exists nowhere else.
- Before building anything, write down what a single page will show that no other site holds. If the honest answer is reworded text from sources anyone can read, stop here; that is the exact shape the policy names.
- Find the dataset next, whether that is your product's usage numbers, a public dataset you compute something new on, or figures that change often enough for you to be the freshest source.
- Design one template in which the data is the page. A table, a chart, or a computed comparison carries the value, and the prose around it only explains what the reader is looking at.
- Generate a first batch of 50 to 100 pages, not 10,000, and submit it for indexing.
- Watch Search Console for six to eight weeks. A sample that cannot earn indexation and impressions will not be fixed by a thousand more pages of the same.
- Scale only what the sample proved, and schedule the data refresh, because a dataset page that goes stale turns into the template text you set out to avoid.
Why it works
Scaled template text is detectable and replaceable, so Google can discount it wholesale without hurting anyone's results. Data nobody else holds cannot be replaced, which is why the data sites grew through the same years the template sites collapsed. The small first batch tells you cheaply which kind you built.
Evidence
In the case-study record in our research, Wise's 1.7 million pages of live currency data reached roughly 54 million monthly visits, while template-driven G2 lost about 80 percent of its traffic by late 2025.