Case study: 29 links in two markets for an online health retailer, with every decision explained

This is a delivery case, not a rankings case. The client's search visibility is still led by its own brand name, and we are not going to dress that up. What this case shows is how a budget turns into links when every placement has to clear five published checks, every decision carries its reason, and every link is watched afterwards. The figures come from our portal on 9 October 2026 and the client is not named.

The client

An online health retailer selling in Germany and the Netherlands. Two languages, two publisher pools, and one legal constraint that shaped every anchor: the company is not a pharmacy and must never be described as one, including by a publisher who means well.

It came to us directly, not through an agency, with a monthly budget and no list of sites. That is the whole brief we work from.

What the client was afraid of

• Links that go live, get invoiced, and quietly disappear in month seven.
• Sites that report strong German traffic but have almost no German readers.
• Anchors written by publishers that create a legal problem rather than a ranking one.
• Being one of five clients on the same five sites, with the same supplier.

The process, month by month

Candidates. We pulled 746 sites for the health topic across both markets and ran the five checks before anyone read a single one: traffic measured in the client's own country, a spam score that was returned and under 40, numeric relevance to the topic, no operator fingerprint across the portfolio, no rented links. One candidate in twelve got through. Own-locale traffic removed the most, as it always does. Sites reporting a combined German-speaking figure fell out when measured against Germany alone.

Outreach. 303 publishers were contacted, with 598 emails sent in total, under two per contact. 27.4% replied and 13.5% of those contacted came back with a price we could work with. The quotes ranged from 30 to 1,200 euro, which is why a flat price list would have lost money at both ends.

Decisions. Each order was chosen against the budget, not the other way round. The reason is written on the order: why this publisher, why a guest post rather than an insertion, why this page. Eighteen of the 29 links are insertions into articles that already ranked for the subject, because a page with history carries the link from day one. Eleven are new articles, written natively in German or Dutch, never translated.

Anchors. Thirteen distinct anchors across 29 links, 72% of them the brand or the bare domain. For a client that cannot be called a pharmacy, the brand is the safest anchor and, as it turned out, the one its profile was missing. Nine links point at the homepage and twenty at inner pages, so the inner pages the client actually sells from get most of the weight.

Three orders cancelled by us. A publisher who changed the terms after agreeing, a page that stopped being indexed between quote and draft, and a site whose traffic collapsed on re-check. Cancelled before money moved, with the reason on the order.

Live. The 29 links went live between 12 September and 8 October 2026. Twenty-seven are live and dofollow at the last weekly check.

What the checks found afterwards

Every live article goes through fifteen checks against what was ordered: reachable, indexable, canonical, in the sitemap, in the right section, title, description, headings, content, images, forbidden phrases, link in place, dofollow, anchor and target.

• On one article the editor reworded 8 of 21 paragraphs. Their right, and recorded as a warning rather than a failure, because nothing was cut and the link stayed where it was.
• On the same article the CMS had dropped the meta title and description we supplied and generated its own. Caught, reported, fixed with the publisher.
• Two anchor changes were picked up by the weekly check within a week of happening. Both were corrected.
• Seven older links on a previous domain were reclaimed at no cost: the publisher was asked to update the link and the anchor, so the client's new brand earns the link directly instead of through a redirect.

Google returned 21 of the 24 links we have checked so far. The remaining five are too recent to judge. A link not returned by day 60 is replaced or refunded.

The numbers

• 29 live links, 18 in Germany and 9 in the Netherlands, on 28 different publishers.
• 22 paid placements at an average of about 210 euro per link, plus 7 reclaimed links at no cost.
• 11 guest posts, 18 insertions; 9 homepage links, 20 to inner pages.
• 13 anchors, 72% brand; 27 of 29 live and dofollow at the last check.
• 746 candidates considered, 1 in 12 passed; 303 publishers contacted, 27.4% replied.

What the client sees

Every order in the portal with the reason it was chosen, the price, the live URL and the state found at the last weekly check. A monthly report with the same figures. The publisher invoices the client directly, and our fee is one line on its own. Nothing in the report, the article or the publisher conversation names us.

Where it stands

The budget keeps being spent through the same five checks each month, and the brand anchor profile the client did not have in September now exists. Rankings for non-brand terms are the next chapter, and we will write it only when the client's own search data shows it. Until then, this is what 29 links look like when every one of them can be checked.