How to raise your Silotree Score, one part at a time.

A score is only useful if you can tell what would change it. The Silotree Score is built to be taken apart: it is four shares worked out from a crawl of your site, each with the counts behind it, and no model judges any of it. That makes raising it a matter of finding which count is costing you and changing it, not of guessing what an algorithm wants.

This guide takes the four parts one at a time, then puts the fixes in order and works through a site of 80 posts from 64 to 90. If you have not seen your score yet, the free internal link audit gives it to you in a few minutes, with no account.

Start with the part that costs the most

The score is the weighted average of four parts, each a share from 0 to 100:

Beside each part, the audit shows the points it cost: its weight times what it is short of 100. That figure is the most useful number on the screen, because it is exactly what fixing all of that part would add back. Start with the part that costs the most, not the one with the lowest score. A Working links score of 80 costs 4 points; a Linked score of 80 costs 8.

Under the parts, the audit lists the fixes worth at least a point, biggest first. Each is worked out by scoring your site again with that one thing fixed and nothing else, so “+7” is the formula’s own arithmetic, not an estimate. Two things to know when reading it:

• Each fix is counted alone. Do two and the gains can add up to more than the two figures, as the worked example below shows.
• Giving an unlinked post a link is counted as one link in, not three. That is the step most people take first, and promising the full part would overstate it.

Linked: give every post three links in

Linked carries 40 of the 100 points, more than any other part, because it is the part that tells sites apart. When the score was calibrated, Linked ran from 6 to 90 across the sites crawled, while Reachable sat near 100 on most of them. On most sites it is where the points are.

Each post counts in one of three ways:

What moves it, and what does not:

• Only links inside your writing count. A link in a menu, footer or sidebar reaches every page at once, so it says nothing about any one post. Adding a post to the menu does not raise Linked (it can raise Reachable).
• Several links from the same page count once. Three links from one article are one link in. Three articles that each link once are three.
• A fourth article adds nothing to the score. Once a post is linked from 3 articles, more links may still help readers, but Linked has already counted it in full.
• Every new post starts as an orphan. Publishing lowers Linked until another article links to the new post, which is why the score can drop in a week you did nothing wrong.

Google’s own guidance points the same way: its link best practices say every page you care about should have a link from at least one other page on your site. In 2022 John Mueller called internal linking “super critical for SEO”, as Search Engine Journal reported, and said ordinary HTML links between the parts of a site cannot be replaced with structured data.

How to fix it

For each orphan, find an older article on the same subject that already mentions what the orphan is about, and put the link on those words. On a recipe blog, a post on proving bread dough belongs in the sentence of the sandwich-loaf recipe that says “leave it to prove”. In a software help centre, the article on exporting invoices belongs in the billing overview where it says “you can export them”. One link moves the post from nothing to half; two more from other articles move it to full. Our guide to linking orphan pages covers choosing the source articles, and which orphans to retire instead of linking.

Working links: point links at the final page

Every link in your articles starts in good standing. Each one that does not arrive cleanly costs a set number of points per hundred links, so a site with more links can carry a few more mistakes for the same score:

So on a site with 1,000 links in its articles, 40 links through a single redirect leave Working links at 99. Ten broken links take it to 97, and forty broken links to 88. A broken link costs 12 times what one redirect does, because one is a dead end for the reader and the other is a short detour. Fix the broken ones first.

Single redirects are the most common kind, and usually come from a move nobody went back for. A plumber’s site that moved from http:// to https://, or changed its permalinks, often has every old link in its articles going through one. The redirect still works, and should stay in place for bookmarks and links from other sites; what you change is the link in your own article, so it points straight at the final address. Google’s crawlers follow up to ten redirect hops, according to Google’s crawling documentation, so a chain usually still arrives. It just arrives slowly, which is why it costs less than a broken link and more than one redirect.

The free audit’s CSV lists every bad link with the post it sits in. In the app, the Broken links screen does the same and, with the plugin connected, rewrites them for you.

Reachable: bring buried posts closer

Reachable is the share of your posts a reader can reach in 3 clicks or fewer from your home page. Here every link counts, menus and archives included, because this part is about the route a reader can take. A post that only your XML sitemap knows about counts as out of reach.

On most WordPress sites Reachable is high already: menus, categories and archive pages put nearly every post within a few clicks. When it is low, the usual cause is a long archive with only “Older posts” links, where each page of the archive is one click further from home. Our guide to click depth works through what that does to a 400-post archive.

How to fix it

• A hub page in the menu. A page linking the posts on one subject, linked from the menu, puts every one of those posts within 3 clicks at once.
• A link from a post near the top. Your newest and most-visited posts are usually one or two clicks from home. A link from one of them brings its destination within reach, and helps Linked too.
• Numbered pagination. If your theme only has older and newer links, switching to numbered pages shortens every route through the archive.

What does not work: putting every post in the footer. It raises Reachable, does nothing for Linked, and gives readers a wall of links nobody uses.

Organised: give each topic a hub

Silotree groups your pages into topics by what they are about. A topic of fewer than four pages is too small to judge and is left out: being small is not a fault. Every other topic counts one of three ways:

A hub is the page a reader would land on to understand the topic as a whole, which the topic’s other pages link to: a pillar page, a guide, or a category introduction with real writing in it. Silotree picks it mostly by the links the topic’s own pages point at it, then by how squarely it is about the topic and how near the top of its section it sits. A page that none of its topic’s pages link to cannot be the hub, however good it is. So there are two fixes for a topic without one: if a page already covers the subject as a whole, link the topic’s posts to it; if none does, write it. On a shop, eleven posts about caring for leather and no page on leather care as a whole is a missing hub, and the fix is writing that page and linking the eleven to it.

Competing means two topics in the same section of the site are about the same thing, which usually means one subject was written twice. The fix is to decide which post should rank, merge or retarget the other, and point links at the one you keep. Our guides to keyword cannibalization and merging, redirecting or retargeting cover how to choose.

Organised has no line in the audit’s fix list, because neither fix is a count anyone can change for you: a hub is something to write or choose, and a merge is a judgement about which post to keep. It is also the part to leave until last. Whether a topic has a hub, and how many of its links stay inside it, both depend on its links, so the linking in the steps before often settles part of it for you.

The order to fix things in

The audit sorts fixes by points for your site, and that is the order to follow. On most sites it comes out like this:

1. Give the posts nothing links to a link in

   Each one moves from counting nothing to counting half. Start with the longest and most useful posts, and retire the ones not worth a link.

2. Add links to the posts with only one or two, until 3 articles link to each

   This includes the orphans you just linked once, which is why this step is often worth more after the first than the list said before it.

3. Point bad links straight at the page

   Broken links first, then chains, then single redirects. Most are a two-second edit.

4. Bring buried posts closer to your home page

   Often done already by the links from steps 1 and 2, when they come from posts near the top.

5. Then the topics

   Write the missing hubs, link each topic’s pages to its hub, and settle the competing posts.

A worked example: from 64 to 90

A made-up site of 80 posts, scored by the same code that scores yours. The first scan found:

The audit’s fix list for that scan:

The owner worked down the list in that order, one weekend at a time, and ran the audit again after each:

Two things stand out. First, step 2 added 14 points, against the +7 the list promised before step 1: the 28 posts linked once in step 1 had joined the posts with one or two links, so there were more of them to bring up. Second, the four fixes together took the site from 64 to 90, and Organised still costs 10 points. Writing hubs for the topics that had none takes it to 96; tidying the topics that leaked or competed takes it to 100.

Why one fix moves some sites more than others

Every part is a share, so a site of 20 posts and one of 2,000 with the same shares get the same score: here, both 81. What differs is what one fix is worth.

On the small site, one link moves the total by 2. On the big one, it moves Linked by a fortieth of a point and the total, which is a whole number, not at all. It takes a hundred orphans linked to move it by 2. That is the formula being fair to both sites, not stuck: on a big site, look at the part scores and the counts behind them to see progress week by week, and expect the total to move in steps. It is also why there is no score below five posts: on a site that small, one link swings it by ten points or more.

What the score does not measure

The score measures how your pages link to each other, and nothing else. On purpose, it leaves out:

• Link text. Silotree flags generic anchors like “click here”, but does not score them. A score that punished them would teach people to stuff keywords into links, and our guide to anchor text explains why varied, descriptive text is better.
• Whether your content is any good, and what you have not written yet.
• Speed, titles, structured data and links from other sites. Other tools measure those well.
• Search traffic and rankings. The score is worked out from the crawl alone. It is not a Google metric, it makes no claim about rankings, and a band is only a name for a range of numbers.
• Other sites. A 70 is your own site’s counts, not a rank against anyone.

So raise it for what it does measure: posts readers can find from your other posts, links that arrive, and topics with a page that holds them together. Then keep it there with one habit: link each new post from two or three older ones on the day it goes live, and it never becomes an orphan. The internal link audit checklist covers the checks the score leaves out, and the methodology page has the full formula.

Questions people ask