SEO backlink campaign: planning one, running it for ninety days, and measuring what it did

A backlink campaign is a time boxed project with four parts: a reason for anybody to link to you, a list of the specific pages that might, a sequence for asking, and a date on which you judge the result. Without the first part it is cold emailing. Without the fourth it runs forever at a cost nobody has calculated. Ninety days is the right box for a first campaign, because it is long enough for replies and placements to arrive and short enough that you will actually review it.

The question to settle in week zero is which page the campaign is for. A campaign pointed at the home page produces links that help the domain generally. A campaign pointed at one commercial page you want to rank produces a measurable outcome. Pick the second kind for your first attempt, and pick a page that is already in positions four to fifteen so there is something to move.

Part one: the asset that earns the link

Somebody linking to you is adding something to their own page. Four kinds of asset reliably give them a reason.

• Original data. A survey of your customers, an analysis of your own anonymised numbers, a price index for your sector. Journalists and bloggers need a figure to cite, and whoever publishes the figure gets the link.
• A genuinely useful tool. A calculator, a checker, a template pack. Tools get linked for years because they remain useful after the article that mentioned them is forgotten.
• The definitive explanation. The page people send to someone who asked. Harder to make than it sounds, and it has to be better than what the existing go to page says.
• Expertise somebody else needs. A quote, a review of a draft, a contributed section. The link comes attached to the credit.

Build or commission the asset before the outreach starts. Campaigns that begin with the email list and work backwards to the reason produce low reply rates and burn the contacts you will want later.

Part two: the target list, built in four passes

1. Who links to your competitors. Pull the backlink profiles of the three pages ranking above yours and keep the sites where a human editor made a decision. Skip directories and syndicated feeds.
2. Who already mentions you without linking. Search your brand name and your product names. These convert far better than cold targets because the relationship exists.
3. Who writes about your subject regularly. Trade publications, newsletters, association sites, university department pages, local press for a local business.
4. Who has a resource page. Pages that list tools, suppliers or further reading for your subject, where an addition is a normal request rather than a favour.

Score the list before you touch it: relevance to your subject first, then whether the site publishes new material, then whether a real person is named as the editor. Judging link quality honestly covers why a high authority score on an irrelevant site is worth less than a modest score on a site about your subject. A first campaign list of 80 to 150 well chosen targets is enough; a list of two thousand is a sign the filtering has not been done.

Part three: the ninety day rhythm

Weeks	Work	What done looks like
1 to 2	Finish the asset, build and score the list, write the sequences	Asset live, list scored, three email variants written
3 to 6	Outreach in waves of 25 to 40, one follow up each after five working days	Whole list contacted once, replies logged, first placements live
7 to 9	Work the replies, supply what was asked for, fix unlinked mentions	Every reply answered within two days, placements verified
10 to 12	Second touch on non responders with something new, then stop	Campaign closed, results written up, decision made

Waves matter. Sending the whole list at once means you learn nothing from the first batch, and the subject line you would have improved goes out to all 150 targets. Send 30, read the replies, change one thing, send the next 30.

On the emails themselves: short, specific to their page, and clear about what you are offering. The templates that get replies cover the structure, and the rule behind all of them is that the first line has to prove you read the page you are writing about.

Part four: measurement, and the decision at day ninety

Track five numbers and nothing else: targets contacted, reply rate, placements live, the position of your target page, and organic clicks to that page. Reply rate tells you whether the asset and the pitch are right. Placements tell you whether the targeting is right. The position and the clicks tell you whether the campaign did its job, and they move last, which is why the decision date is at day ninety rather than day thirty.

Write the result up honestly, including the targets that ignored you, and keep the list with its outcomes. The second campaign is substantially cheaper than the first because the list, the sequences and the knowledge of who responds already exist. Keeping that record is what backlink manager software is for once the volume grows past a spreadsheet.

What to avoid, and why it matters more than it used to

Google's link spam policies are explicit that buying or selling links for ranking purposes, large scale article and guest post schemes built only for links, and excessive link exchanges all violate its guidelines. Read them once before you start so that you recognise what is being sold to you. Vendors offering a fixed number of placements per month on sites they will not name in advance are selling exactly the pattern those policies describe, and a campaign that depends on them is a campaign with a cleanup project attached.

Where you would rather hire this out, the models and the vetting questions differ, and the three delivery models is the piece to read before you brief anybody. Be wary of any provider that promises a position in return for a number of links; the deliverable they can honestly commit to is outreach volume and placements, not a ranking.

Running a link campaign: the questions that come up in week two

How many links does one campaign need?

Fewer, better placed, than most plans assume. Look at what the pages above yours have and aim to close the gap in relevance rather than in count. Five links from sites genuinely about your subject outperform fifty from sites about nothing in particular.

What is a normal reply rate?

It swings with the quality of the list and the asset rather than the volume of sends. Treat your own first wave as the baseline, and improve it by changing the asset or the opening line rather than by sending more.

Should the links point at the home page or a specific page?

The page you want to rank, for a campaign with a measurable goal. Spread a few to the home page as well so the profile looks like a normal site's, which it will anyway if the asset is good.

How long before links show up in rankings?

The link has to be crawled first, then reassessed with the rest of the page's signals, which is why the review sits at ninety days rather than thirty. Positions on the target page are the last of the five numbers to move.

Can we run a campaign in house?

Yes, and the first one usually should be. One person can build the asset, score 120 targets and run the sequences in a quarter alongside other work, and doing it once teaches you what to brief an agency on later.

What should we do with the unlinked mentions we find?

Ask. A short, polite note to the author pointing out the mention and offering the right URL converts better than anything else in a campaign, and it takes ten minutes per mention.

Links your campaign wins are also sources an assistant cites

The sites on your target list have a second job you did not plan for. When somebody asks ChatGPT, Gemini, Perplexity or Claude for a recommendation in your category, the assistant builds its answer from a handful of pages it treats as reliable, and those pages are usually the trade publications, round ups, resource lists and association pages that a link campaign is already courting. A placement on one of them is both a link and a sentence about you inside the material those answers are made from.

That reorders a target list in a useful way. The sites worth the most effort are the ones the assistants already quote when a buyer asks your category's obvious question, and that is checkable rather than guessable. AI Knows Us keeps a record of which sites get cited in your category, which turns a link target list into evidence, and the overlap with your existing list is rarely complete. The sites in the gap are the ones to add to the next campaign.