How to generate traffic to your blog: harvest the archive first, then publish

The three lists hiding in your Search Console report, the internal linking sweep most blogs never run, one question per post, and the distribution routine for every new piece.

Published by AI Knows Us · Updated 12 October 2026

Traffic to a blog comes from five places, and only two of them involve publishing something new. The archive you already have is the largest and the most neglected: a small number of old posts carry most of the visits, several more are decaying quietly, and almost none of them link to each other properly. So the order of work is to harvest the archive first, then fix the internal linking, then publish into the questions you have not answered yet, then distribute each new post deliberately to the people who already read you, and only then chase links and syndication. Done in that order, a blog of thirty posts usually produces more than the same blog with sixty.

Start with the posts you already published

Open your Search Console performance report, set the range to the last six months, and sort pages by clicks. Two patterns will be visible immediately. A handful of posts produce most of the traffic, and a long tail produces almost none. Then switch the comparison to the previous six months and sort by the biggest decline. That gives you three working lists, and each list has its own treatment.

  • The winners. These get attention, not rewriting. Add the questions readers now ask that the post does not answer, put a clear next step in the middle rather than at the end, and link them to your newer posts, because they have the audience to lend.
  • The decliners. A post that peaked and has been sliding for two quarters usually has one of three problems: its facts have dated, a competitor published something more complete, or it is competing with a second post of yours on the same question. Fix in that order.
  • The silent tail. Check the impressions column before you touch these. A post with impressions and no clicks has a title and opening problem. A post with neither has a demand problem, and the honest options are to fold it into a stronger post with a redirect, or to leave it alone and spend the hour on something with readers.

Refresh in place. Keep the URL, update the content and the visible date, keep the original publication date somewhere on the page if your template supports it, and resubmit in Search Console. For the report mechanics, our Search Console guide covers what each number measures, and if the whole blog has dropped rather than individual posts, work through the traffic drop checklist before you write anything new.

The internal linking sweep, which is the fastest unpaid lever a blog has

Most blogs link outward from new posts and never backward from old ones. The archive is where the authority sits, so the links have to run from the old posts into the new ones as well as the other way. Make it a rule with a number in it: every new post gets links from at least three existing posts on the day it goes live, with anchor text that describes the destination in the reader's words. Keep a note of which posts you used, so the same three do not carry everything.

While you are in there, look for pairs of posts aiming at the same question. Two posts competing for one query split the signal and neither one wins it, which is the most common self inflicted blog problem. Choose the stronger post, merge the useful parts of the weaker one into it, and redirect. Our page on internal linking has the audit that finds the gaps.

Publish into questions, one question per post

A blog grows on questions, not topics. "Email marketing" is a topic and it cannot be answered. "How often should a small shop email its list" is a question, and it can be answered better than anybody else has answered it. Build the list of questions from four free sources: the Search Console queries where you already appear on page two, the questions your support inbox and sales calls repeat, the related searches and autocomplete on your core terms, and the questions inside the comments and threads where your audience gathers.

Then assign exactly one post per question and keep a spreadsheet of which post owns which question. That register is what stops you writing the same post twice in eighteen months, which is the other half of the cannibalisation problem. For the method behind building the list, see keyword research in five steps and content gap analysis.

Distribution, in the order that produces readers

MoveEffortTime to effect
Email the post to your own subscribers, with the answer in the email and the detail on the pageLowSame day
Answer the live threads where the question is being asked, properly, and link as supportMediumDays
Internal links from three existing postsLowTwo to six weeks
Pitch the post to one newsletter or roundup that covers your fieldMediumWeeks
Repurpose into the format your audience actually uses: a short video, a slide, a printableMediumWeeks
Outreach for links, but only for the one post a year that deserves itHighMonths

The email list belongs at the top of that table for a reason. It is the only channel you own, it gives a new post its first readers within hours, and those readers are the ones who share it. If the blog has no email capture on it, that is the first build rather than the last. The flows and list health side is covered in our email marketing page.

One rule for communities and forums: answer the question in the thread in full, then link only if the page adds something the answer could not hold. A link dropped without an answer gets removed and remembered.

A ninety day plan for an existing blog

  • Weeks one and two. Build the three lists from Search Console. Fix titles and openings on the posts with impressions and no clicks. Merge or redirect any pair competing for the same question.
  • Weeks three to six. Refresh the four biggest decliners. Run the internal linking sweep across the archive. Put email capture on every post if it is missing.
  • Weeks seven to twelve. Publish one post a week into the question list, with the distribution routine applied to each: email, three internal links, one thread, one pitch.
  • Review at day ninety. Clicks and impressions by post age cohort, email subscribers added, and the number of posts that moved from page two to page one. Those three tell you whether the system is working, and none of them is a monthly traffic total.

What the format of a post has to do with its traffic

Posts that earn links and get recommended have a few things in common, and they are all structural. The question is answered in the first paragraph rather than after six hundred words of context. There is at least one thing on the page nobody else has: a number you worked out, a procedure you tested, a comparison table you built, a screenshot of the thing itself. Claims that depend on a date carry the date. And the author is a named human with a reason to be believed. Those four choices decide whether other people cite you, and citations are what carry a post past your own audience. The writing workflow behind that is in our page on the SEO blog post.

Blog traffic questions worth a straight answer

How often should I publish?

Weekly if each post can be genuinely good, fortnightly if not. Consistency matters less than the register of questions: publishing twice a month into questions nobody else has answered properly beats twice a week into questions already answered better.

How long should posts be?

As long as the question needs and no longer. Length follows the answer, so a procedural question may need two thousand words of detail and a definitional one may be finished in six hundred. Padding a short answer to hit a word count is visible to readers in the first screen.

Should I update old posts or write new ones?

Update first for the next quarter. Refreshing a post that already has impressions compounds work you have done, and it shows results in weeks rather than months. Once the archive is current, go back to publishing.

How long before a new blog gets traffic?

Expect the first meaningful search traffic in the second quarter and a readable pattern by month six, faster on narrow questions with little competition. Use the first three months to build the email list and answer threads, because those produce readers while the search traffic arrives.

Which of your posts gets read back to somebody else

A blog post now has a second kind of reader. Somebody asks an assistant a question your post answers, and the reply is a short answer with two or three sources behind it. The posts that get used in that answer are the specific ones: a procedure with the steps in order, a dated fact with a source, a table somebody can lift, a number you calculated and showed your working for. General advice posts rarely get cited, because there is nothing in them that could only have come from you. That is a useful editorial instruction on its own, and it is also checkable: AI Knows Us reads the answers assistants give on a topic and lists the pages each one drew from, so you can see whether any of your archive is in there and what the posts that are have in common.

Read next

Or browse the whole archive.

See what AI says about you.

The first scan is free, it needs no account, and it asks the questions your buyers ask.

Free. No card. We ask 5 real buyer questions on 2 AI apps.

Already sure you want this? Join the waitlist. We open on 31 October 2026. It costs an email address. No card, and nothing is charged. Or talk to a person if you have a question first. Or read how we fix what it finds.