Case studies: problems, work performed and measured results in real projects

The question asked before working with an agency or freelance team is often the same: what was done for a similar business, and what changed? A portfolio shows which websites were built, but it does not show why a website failed to appear on Google, where the advertising budget went or how traffic was preserved during migration. The following cases were compiled to answer that question.

The businesses in these cases are small and medium-sized companies in Istanbul and Germany: an automotive electronics service, a fish restaurant, a sign manufacturer, two e-commerce stores, an industrial equipment manufacturer, a pallet trading company, a guesthouse, a scrap metal company and the website itself. Industry and region information is provided only to explain the context of the work; company names, domain names and personal details are not included.

Automotive electronics service: a blog post competing with its own service page and 36 unlinked pages

The website of a service repairing vehicle ECUs and electronic units on Anadolu Yakası appeared for revenue-generating searches with a blog post rather than a service page. For the search ‘ECU repair,’ Google selected the ‘how long does it take’ article instead of the service page, while the service page remained between positions 35 and 44. In addition, 36 pages received no links from any page and had ‘Discovered – currently not indexed’ status in Search Console.

• Problem: pages describing the same topic competing with each other, unlinked pages, no structured data describing the business, a missing H1 heading in the blog template, and service pages in the menu remaining around 270 words.
• Work performed: seven duplicate pages were consolidated with 301 redirects, contextual links were added from 30 pages to service pages, service schema was added to 59 pages, and incorrect manually written schemas were removed. Three free tools were launched: a fault assistant, an interactive ECU board and a fault code decoder.
• Result: during October 1–16, organic clicks rose from 212 to 353 compared with September 15–30 (+67%, Search Console). The service page’s ranking will be measured on November 15–22.

Fish restaurant and fish market: a homepage consisting only of video

The homepage of a company operating both a restaurant and fish market on Anadolu Yakası contained only a background video: the page had 21 words, no H1 heading and no meta description. No Search Console property had been created either.

• Work performed (September 2026): Search Console was set up on the first day. An H1, introductory text and restaurant schema were added to the homepage, excess H1 headings were corrected, and titles and descriptions for 18 pages were rewritten. Thirteen pieces totaling approximately 28,300 words of content were added, including species and cooking guides linked to fish on the menu, a wholesale homepage and refreshed page bodies; images began being served in a next-generation format.
• Result: comparing August 16–September 12 with September 13–October 10, 2026, clicks rose from 76 to 224 and impressions from 1,290 to 13,100; average position remained at 6.7 (Search Console).
• An honest note: click-through rate fell from 5.9% to 1.7% because the new guides received many impressions but few clicks. Part of the increase came from searches using the brand name.

onurfreelance.com: from 34 to 361 clicks in 28 days

The website itself is also a case. At the beginning of September 2026, the website appeared on Google with few pages and lacked content for pricing, service and troubleshooting searches. The method we apply to our own website is the same as the one we apply to customer websites.

• Work performed: technical cleanup (redirects, duplicate meta descriptions, security headers), pricing pages, troubleshooting guides, structured data, free tools (price calculation, website checker and others), and a data study examining 847 local business websites.
• Result: organic clicks totaled 361 in the last 28 days, compared with 34 in the preceding 28 days. Impressions rose from 5,250 to 35,500 and the number of pages receiving Google impressions from 67 to 387 (Search Console, reading on October 22, 2026).
• Still unresolved: some pricing and service pages remain at positions 15–40. Most clicks come from informational searches; moving sales pages into higher positions is the work of subsequent months.

Electrical supplies e-commerce store: products without images and marketplace synchronization

At an electrical supplies store with an extensive catalog, a significant share of products had no images, while marketplace product pages contained inconsistent and incomplete information. Products received an ‘image too small’ warning in Google Shopping.

• Work performed: large manufacturer images were added to products lacking images, platform-specific rich content was written for more than 80 products on Trendyol, Hepsiburada and n11, and missing products were synchronized across marketplaces. A brand module was installed and brands assigned to 141 products, and the category structure was reorganized.
• Result: the number of products without images fell from 1,424 to 52; 1,450 products became approved in Merchant Center. The website received 341 organic clicks during September 15–October 15.

Computer accessories e-commerce store: URLs deleted by stock integration

At two computer accessories stores owned by the same company, stock integration periodically deleted and recreated categories and products. Each deletion caused old URLs to return 404 errors, and pages that had ranked on Google disappeared. The redirect table was empty; the larger website received 2,090 organic clicks in 28 days, and every broken URL took a share of this traffic away.

• Work performed: broken URLs were extracted from 16 months of Search Console data and redirected to the correct pages with 301 redirects. When it was discovered that the platform had deleted the redirect table, it was uploaded again and an automated check running every morning was established. The currency error in product schema was corrected, and a compatible adapter finder tool was added.
• Category content: selection guides and comparison sections were written for 300 categories on one website, the second H1 was removed from 1,989 of 2,061 product pages, and titles and descriptions were written for thousands of product, category and brand pages. Comparing August 15–September 11 with September 12–October 9, 2026, category page clicks rose from 58 to 166 and product page clicks from 76 to 247. Because some products were also unpublished within the platform during the same period, the increase is not attributed to a single change.
• Result: a total of 10,441 redirects across the two websites (599 and 9,842) were checked individually from outside, and all lead to the correct pages. 404 validation was initiated in Search Console.

Sign manufacturer: only 19% of the 16,300 URLs Google knew were indexed

A sign manufacturer in Istanbul had long been unable to reach the first page even for searches using its own brand name; average position for brand searches was 17.1. According to Search Console, Google knew approximately 16,300 URLs on the website, and only 19% were indexed.

• Root cause: tags on project photographs were automatically converted into clickable search links by the website’s platform. These links generated 10,143 ‘Blocked by robots.txt’ and 1,240 ‘Crawled – currently not indexed’ URLs, causing Google to spend crawl budget on meaningless pages.
• Work performed: 12 automatic links in product templates were converted to plain text, robots.txt and server rules were adjusted, and breadcrumb schema and missing H1 headings were added to product and category pages. Measurement of telephone and WhatsApp clicks began.
• Result: new meaningless URL generation stopped, leaving Google’s crawling focused on actual category and product pages. Ranking changes will be measured in early November.

Industrial hygiene equipment manufacturer: duplicate product descriptions and an unlicensed theme

A stainless steel equipment manufacturer selling through quotations said that even product content it had written itself did not appear in search. The review showed approximately 27% of the 102 URLs in the sitemap as indexed. Products were partially visible in categories with original descriptions, while almost no products were indexed in categories using template copy.

• Work performed: pages that could not be edited because of unlicensed themes and plugins were rebuilt with Elementor. Products were moved into an actual product structure across 21 categories, 96 product pages were rewritten with at least 600 words of original copy each, and new product images were prepared for 95 products.
• Result: the website’s duplicate product copy problem was removed and all product pages became editable. The indexing rate is being monitored through Search Console.

Pallet trading company, Germany: an advertising plan beyond the budget and a German website labeled Turkish

In the initial setup of a German pallet trading company’s Google Ads account, 17 campaigns with separate campaigns for each city and a total daily budget of 370 euros had been planned. The company’s actual budget was 30 euros per day. The website’s language tag appeared as Turkish despite German content, and the page carrying most ranking search terms returned a 404 error.

• Work performed: the advertising structure was reduced to one campaign with a daily budget of 30 euros; keywords were switched to exact match and each language’s ad group was directed to the purchasing page visited by people wanting to sell pallets. The website’s language and time zone were corrected for Germany, the 404 page was linked to the correct category through a 301 redirect, and GA4 and cookie consent were set up in German.
• Result: the budget was consolidated into one measurable campaign instead of being spread across campaigns that would exceed it; the URL that had carried the lost rankings became functional again.

Guesthouse in Kadıköy: a local search campaign replacing one with no conversions

In the advertising account of a guesthouse in Kadıköy, a campaign inherited from previous management was spending 4,000 TL per day while measured conversions appeared as zero.

• Work performed: the old campaign was stopped. A new search campaign was set up targeting Kadıköy and the surrounding 5 km, limited to 48 keywords and 51 negative keywords, with a cost-per-click ceiling.
• Result: the budget was directed toward local searches seeking accommodation. Search terms and click costs are checked regularly; conversion comparisons will be made once sufficient data accumulates.

Scrap metal company: a website that did not run on its server, moved to WordPress without altering the design

The website of a scrap purchasing company in Istanbul had been written with ASP.NET and SQL Server. Because the hosting was Linux-based, the website did not work; content remained only in a database backup.

• Work performed: content was extracted from the backup file without installing SQL Server. The website was migrated to a custom WordPress theme preserving the same design, and old URLs were connected to new URLs with 301 redirects. Approximately 50,300 words of content were then written for 67 pages, 3 blog posts and a new district page.
• Result: the website is live again, visitors arriving through old URLs reach the correct pages, and every service page contains text that search engines can read.

Experiment log: what was tried on onurfreelance.com, what worked and what did not

Case pages usually describe only successes. The following records are experiments conducted by the team on its own website, and whatever the outcome was is recorded, including failures. Each row contains the expectation before the experiment, the work performed, the measured result and the next step. Google updates, seasonality and other website changes during the same period can also affect the outcome; the figures should therefore be read as observational records rather than proof of cause and effect.

The log is updated with each measurement: new experiments are added, and rows with changed results are rewritten with dates. Lessons from experiments that did not work determine what should be done first on customer websites; one conclusion from these records is to carry out link and local profile work simultaneously rather than allocate the budget only to page content for a competitive service term.

What do these cases have in common?

In most of the ten cases, the problem was not a lack of content. Duplicate pages, broken URLs, platform settings generating meaningless pages or advertising structures that did not fit the budget prevented existing work from delivering value. Work always begins in the same order:

1. Measurements are taken from Search Console, the advertising dashboard and the website itself; the starting point is data rather than guesswork.
2. Technical problems losing money or preventing rankings are resolved first.
3. Content and advertising are built around queries people actually search for.
4. After every change, results are measured again from the same source and recorded in the report.

How were the figures measured?

Organic click and impression figures were taken from Google Search Console, product approval counts from Google Merchant Center, advertising spend from the Google Ads dashboard, and redirect counts from website administration panels and external URL checks. Comparisons use periods of equal length, and period dates are written in the case. For work whose results have not yet been measured, no figure is given and the measurement date is stated. Ranking and traffic results depend on Google’s decisions; these descriptions are records of past work, not guarantees for the future.

A quick preliminary check of a website in a similar situation can be performed using the free website checking tool . Service and pricing details are on the SEO service, SEO prices, Google Ads prices and corporate web design pages. Completed website projects are listed on the portfolio page.