Almost every blog we audit has 40–60 articles and close to zero commercial impact. Sometimes the traffic is even respectable — it is simply the wrong people landing on the wrong pages. The cause is nearly always structural: topics came from a "what could we write about" list rather than a demand map, and the articles are not connected to each other. What follows is the architecture that turns a blog into a lead channel: clusters, formats matched to intent, internal linking, trust signals, and the numbers that tell you whether any of it is working.

The blog exists, the enquiries do not

Open the analytics on one of these blogs and the same picture repeats from project to project. Five mistakes account for most of it, and each one costs months.

• Topics picked out of thin air. "10 tips for choosing a colour" gets written because it is easy to write, not because anyone searches for it. The check takes five minutes: if a topic has no query with at least 50 monthly searches behind it, it will produce nothing but a line in the content plan.
• Not a single commercial-intent page. A blog of 50 purely informational posts is a magazine, not a sales channel. Enquiries come from pricing breakdowns, comparisons and "how to choose a vendor" pieces — exactly the ones companies avoid writing.
• Articles live in isolation. With no internal linking, every new page starts from zero: no inherited authority, no context for Google, and a reader who finishes and leaves.
• Volume instead of depth. Three 700-word posts on one topic compete with each other for the same query and none of them ranks. One 1,800-word piece does.
• Nobody updates anything. A pricing article from 2023 with stale figures does more damage than no article at all — the reader concludes you are out of the market.

A cluster, not a list of topics

The working unit of content strategy is not an article — it is a cluster: one substantial hub page plus 5–12 narrower pieces around it, stitched together with links in both directions. Google evaluates how thoroughly a site covers a subject, not just the individual page. Twenty articles on one topic outperform twenty articles on twenty topics, and that is not theory — it is what shows up in Search Console six months in.

Sequence matters. Harvest demand and group the queries first — the mechanics are covered in our guide to keyword research and query grouping. Then draw the link map, and only then write. Do it the other way round and you end up rewriting half the drafts because they cannibalise each other. The stitching itself is a topic of its own, covered in the internal linking strategy guide.

Which formats actually produce enquiries

Not every article has to sell, but you should know the job each one is doing. Here is how the formats convert across our projects. The figures are submitted forms and direct emails from organic traffic, excluding assisted conversions.

Which leads to a conclusion plenty of people dislike: do not start a blog with dictionary articles. "What is UX" brings students. "What a redesign costs and how long it takes to pay back" brings a business owner with a budget. Write the definitions later — they exist to complete the cluster, not to sell.

Trust signals a human and an algorithm both read

E-E-A-T is not a score inside Google. It is the set of cues a quality rater and a ranking system use to decide whether this was written by someone who has actually done the work. Language models read the same cues when assembling answers in AI search: with no specifics, you simply do not get cited. The practical side is covered in our guide to optimising for AI search.

When to expect results and what to watch

Content is the slowest acquisition channel, which is precisely why it gets abandoned in month four — one step before it starts working. Here is a realistic schedule for a new blog on a domain with no history.

1. 1

   Weeks 1–4: indexation

   Pages enter the index within 3–20 days. The only metric that matters here is the number of indexed URLs in Search Console. There are no rankings yet, and there cannot be.

2. 2

   Months 2–3: first impressions

   Long-tail impressions appear at positions 20–60. Clicks are near zero, which is normal. Watch which queries you are being shown for — they are often not the ones you wrote for.

3. 3

   Months 4–6: movement into the top 20

   The strongest articles reach positions 5–15 and the first steady traffic arrives. This is where gaps in the cluster become visible — fill those in rather than starting a new cluster.

4. 4

   Months 6–12: traffic and enquiries

   The cluster starts behaving as a unit: the hub rises because of its spokes, and the commercial pages produce enquiries. Our working benchmark for B2B niches is 30–80 organic visits per article per month.

5. 5

   Year two onwards: updates over new posts

   From the second year, half the effort should go into refreshing what exists. Updating an article with current figures is usually cheaper and more profitable than writing a new one from scratch.

Watch three things and ignore the rest: how many queries the page ranks in the top 10 for, what share of the cluster has non-zero traffic, and how many enquiries had at least one blog article somewhere in their path. Total blog sessions is a vanity metric — it also goes up when you write about things nobody buys. For realistic timelines across the whole channel, see our breakdown of how long SEO takes to work.

Frequently asked questions about content marketing

In short

• Plan in clusters — one hub plus 5–12 spokes — rather than in individual articles.
• Pricing breakdowns and comparisons produce the enquiries; definition posts exist to complete the cluster, not to sell.
• Trust comes from specifics: a real author, honest ranges, and downsides named out loud.
• The first steady enquiries land in months 6–12; stopping at month four means paying for the work and collecting none of the return.
• From year two, spend half the budget updating existing articles instead of adding new ones.