Tecof • September 14, 2026
Meta Description and Title Tag Optimisation Guide
In Brief
The title tag is the headline a page shows in search results and in the browser tab; the meta description is the two-line summary underneath it. Neither decides directly where the page ranks — they decide how many of the people who see it click. When two results sitting in the same position get twice the clicks of one another, the difference usually lives in these two fields. As of 2026 the question is no longer “are our meta tags filled in” but “are we writing the text that appears in search results, or is Google rewriting it”.
Tuesday morning, 09.15. The Search Console report is open: 412,000 impressions over the last 28 days, 9,100 clicks, an average click-through rate of 2.2 percent, an average position of 8.4. Of the 180 product pages sitting on page one, 140 share a single title pattern: “Product Name - Store Name”. The meta description field is completely empty on 96 pages, identical to the first sentence of the product copy on 70, and on 14 the same text has been copied across three different products.
What is being lost here is not ranking but the payoff of ranking already earned. Taking 412,000 impressions at position eight means the search engine has already made the page visible. Lifting click-through from 2.2 to 3.1 percent means roughly 3,700 extra visitors a month without a single new backlink and without changing a line of code. This piece is about closing that gap.
What Title Tags and Meta Descriptions Actually Are
Both fields live in the page's <head> and are invisible in the page body once a visitor arrives. Their visibility is outside the site rather than inside it: in search results, in the link card shared on social media, in the browser tab and in the bookmark list.
Definitions
- Title tag: the page's formal headline. It is the blue clickable line in search results and the text in the browser tab.
- Meta description: the summary beneath the headline. Not a ranking signal, but it decides how much space the result occupies and what the reader decides.
- H1: the visible main heading inside the page. It does not have to match the title tag, and usually is better when it does not.
- Slug: the readable tail of the address. It appears in the breadcrumb line of the result and should be consistent with the headline.
Do they affect ranking?
The title tag is a weak but real ranking signal; the words in it describe what the page is about. The meta description is not a direct ranking signal. Both, however, affect click-through, and click-through in turn feeds the user-behaviour signals that influence ranking over time. So saying a meta description “does not affect SEO” is wrong; the effect is indirect and delayed.
Do AI search surfaces use these fields?
They do, but differently. In a classic result list the reader is a person; in an AI summary it is a model. Summarising the page, the model reads the title and description as the first signal of what the page promises. So an exaggerated description that does not match the page no longer merely disappoints — it lowers the chance of the page being cited at all. We covered this in detail in our piece on AEO and GEO.
Character Limits Are Really Pixel Limits
The usual advice is “60 characters for the title, 155 for the description”. It is roughly right, but the actual limit is measured in pixels rather than characters. A search result is truncated at a fixed width and letters are not equally wide. “Illicit” and “WWWWWWW” have the same character count and occupy very different space.
Practical limits
These ranges are safe for Turkish copy: 50-60 characters for the title, 140-155 for the description. Turkish fills the space faster than English because of its longer words and suffixes, so writing near the lower bound is safer than leaning on the upper one. Put the brand name at the end and keep it short: “ | Store” takes less room than a three-word brand.
Mobile versus desktop
The large majority of e-commerce searches in Turkey come from mobile. On mobile the headline can wrap to two lines and more characters survive, while the description is truncated earlier. The practical rule: fit the core message into the first 60 characters and the information that moves a purchase decision — shipping, instalments, returns — into the first 100. Everything after that is a bonus.
| Field | Safe range | Critical opening | Common mistake |
|---|---|---|---|
| Title tag | 50-60 characters | First 35 characters | Putting the brand first |
| Meta description | 140-155 characters | First 100 characters | Copying the product copy |
| Open Graph title | 55-70 characters | First 40 characters | Never defining it |
| Slug | 3-5 words | All of it | Burying the product code in it |
The Turkish character question
Capital İ, dotless ı, ş, ğ and ç cause no problem as long as the source is encoded correctly. Trouble appears in templates that force headlines to uppercase: a phrase like “ışıklı tabela” can collapse into a broken form rather than the correct “IŞIKLI TABELA”. Avoid CSS or template functions that uppercase titles; write the text in the form you want it.
Patterns That Work by E-commerce Page Type
Every page type serves a different search intent, so every page type needs a different pattern. Applying one template across the whole site is the most common mistake and the easiest to fix.
Product pages
The searcher knows the product and is deciding. The title needs the product name, a differentiating feature and the brand; in the description, trust factors work better than a price. The pattern: Product Name + Differentiator + Brand. In the description say something the same product's Trendyol or Hepsiburada listing does not: two-year warranty, same-day dispatch, free returns, authorised service. Consistent product names depend on disciplined catalogue naming, which ties directly to your SKU structure.
Category pages
The searcher has not chosen a product yet and is looking at options. A category name plus a promise of range or filtering works well: “Women's Trainers: Models and Prices”. In the description give information that makes choosing easier: how many brands, which size range, which filters exist. Category pages bring the most traffic of any page type in e-commerce, and in most stores they are the most neglected.
Blog and guide pages
The searcher wants information, not a purchase. Match the question exactly in the headline and summarise the answer in the description. The single biggest lift in click-through comes from writing the description as “here is the answer” rather than “in this article you will learn”. Do not be afraid of giving the answer; people still click for the detail.
Campaign and brand pages
Dates and validity are critical here. If you use year-bound phrasing like “2026 Sale”, assign someone to update it when the year turns. A stale year is the cheapest and most avoidable form of lost trust.
| Page type | Title pattern | Description focus | Avoid |
|---|---|---|---|
| Product | Product + feature + brand | Warranty, shipping, returns | Hard-coding the price |
| Category | Category + models/prices promise | Option count, filters | One-word headlines |
| Blog | The question itself | A summary of the answer | The “in this article” opener |
| Campaign | Campaign + date range | Terms and duration | An out-of-date year |
| Corporate | Topic + brand | Concrete proof, numbers | Generic self-praise |
Do not write the price into the text
Putting “1,299 TL” into a meta description looks tempting but creates two problems: when the price changes the text becomes a lie, and the search engine may keep showing the old text for weeks. Declare price and stock with structured data (schema.org Product and Offer) instead; the search engine already reflects that in the result and keeps it current.
Why Google Rewrites What You Wrote
It is normal for the description you wrote not to appear in the result. The search engine can pull a passage from the page body when it thinks it matches the query better. That is design, not malfunction: if one page surfaces for ten different queries, a single fixed description cannot suit all ten.
Reasons for rewriting
- Mismatch: the description does not answer the query and a better sentence exists on the page.
- Absence: the field is undefined, so the engine picks a passage itself.
- Repetition: the same description is used on hundreds of pages and does not distinguish them.
- Stuffing: text padded with repeated keywords is not judged readable.
- Length: a very short description cannot fill the space the result offers.
Reducing the rewrite rate
It cannot be prevented entirely, but it can be reduced. The three most effective steps: write a unique description for every page, spend the first sentence on the main question the page genuinely answers, and keep that same information on the page as a clear paragraph. When the engine cannot find a good sentence in the body, it tends to use yours.
When to leave it alone
On broad category pages that surface for many long-tail queries, letting the engine choose a description per query usually beats your fixed text. On those pages, spend your energy clarifying the page content rather than forcing the description.
Scaling Across Thousands of Pages: Templates, AI and Auditing
On a fifty-page site meta tags are written by hand. On a four-thousand-product catalogue that is impossible and you need templates and automation. The goal is not to generate hand-written quality automatically but to guarantee that what is generated automatically is acceptable.
Variable-driven templates
A template feeds off catalogue fields: {brand} {product_name} {key_feature} | {store}. The quality of this approach is bounded directly by your data quality. If the product name field contains values like “Black-2”, the template produces exactly that quality. So before building templates, clean the name, brand and key-feature fields in the catalogue.
Bulk generation with AI
Where templates fall short — especially on category and blog pages — bulk generation with AI works. The critical part is checking the output against acceptance criteria: is it inside the character limit, does the target keyword appear once, does it promise something the page does not deliver, is the brand name spelt correctly. Run that check automatically or the production speed turns into correction debt. We laid out the approach in our prompt engineering piece and the bulk side in our product description guide.
Hunting duplicate titles
Pages sharing a title tag signal to the search engine that they are copies of one another. Colour and size variants are the most frequent source. The fix: consolidate variants onto one product page, and if separate pages are genuinely needed, add the differentiator to the title. Audit this from Search Console's coverage report and a site crawl; monthly is enough.
| Page count | Method | Audit frequency | Owner |
|---|---|---|---|
| 0-100 | Entirely by hand | Quarterly | Content owner |
| 100-1,000 | Template + top 50 by hand | Monthly | Content + SEO |
| 1,000-10,000 | Template + AI + acceptance criteria | Monthly, automated | SEO owner |
| Over 10,000 | Automated generation + sampled audit | Weekly, automated | SEO + technical team |
Feeding meta fields automatically from the catalogue depends on product data being clean on the ERP side. Product names coming out of Logo, Mikro or Netsis are usually written for accounting rather than for customers; doing that translation in the integration layer is cheaper than fixing every page one at a time.
Measurement: What to Watch and When to Decide
Meta optimisation is one of the few SEO jobs whose result can actually be measured. Set the measurement up properly and you can see the payoff of the team's work within four weeks.
The right report in Search Console
Apply a page filter in the performance report and use comparison mode to put the 28 days before the change next to the 28 days after. The metric to watch is click-through rate, not click count: clicks also rise when impressions rise. Do not credit a CTR increase as a win without confirming that average position stayed flat.
How much data makes a difference meaningful
On a page with 200 impressions a month, a change in click-through is noise. A practical threshold is at least 1,000 impressions per page across the comparison period. Measure pages below that threshold as a group rather than individually; “the combined CTR of these 40 category pages” is a meaningful number. The same logic applies in advertising, which we covered in our A/B testing piece.
If rankings drop
A few days of movement after a title change is normal. When a drop lasts more than two weeks and exceeds one position, it is usually because a word you removed from the title blurred what the page is about. Restore the old title and make one change at a time.
| Metric | Source | Comparison | Decision threshold |
|---|---|---|---|
| Click-through rate | Search Console | 28 days before / after | 10% relative increase |
| Average position | Search Console | Same period | Drop over 1 place: roll back |
| Impressions | Search Console | Same period | Should stay flat |
| Revenue per session | Analytics | Same period | If it falls, the promise is wrong |
That last row matters. If click-through rose but revenue per session fell, the headline is promising something the page does not deliver. It is the one failure mode that looks good in the short run and costs you in the medium one.
Meta Optimisation in Thirty Days
The timeline below runs in the same order on a four-thousand-product catalogue and on a fifty-page corporate site; only the duration of each stage changes.
Days 1-7: inventory and prioritisation
Export the page-level report for the last 28 days from Search Console. Flag pages with high impressions and low click-through; that group pays back fastest. In the same week list the fields that are empty, duplicated or outside the length limits. No copy is written this week; only the list comes out.
Days 8-14: write the first fifty by hand
Write the titles and descriptions for the first fifty prioritised pages by hand. Doing it manually is essential here, because the template will be derived from these fifty. For each page ask one question: which single sentence separates us from the other nine results in the same position?
Days 15-21: build the template and apply it in bulk
Turn the pattern the fifty examples revealed into a template per page type and apply it to the rest of the catalogue. Preview a hundred random pages before applying; that is where you catch the edge cases the template breaks. Write the acceptance criteria and set up the automated check.
Days 22-30: measure and lock it in
Wait a week for the changes to be crawled, then open the comparison report. Fold the winning patterns back into the template and roll back the losers. By the end of the month you should have three things: an approved pattern per page type, a monthly automated audit, and one person who owns this work.
Here is the job for tomorrow morning: open the last 28 days of the page report in Search Console, sort for pages with more than 1,000 impressions and a click-through rate under 2 percent, and read the title and description of the top ten. Seeing how many of those ten have an empty or duplicated description will tell you where this month's work starts. On a platform where meta fields are templated by page type, that work collapses from editing pages one by one to fixing four templates.
Frequently Asked Questions
Is the meta keywords tag still used?
No. Google has not used it for ranking in years and filling it in gains nothing. Some on-site search engines and legacy systems read it; otherwise leaving it empty is fine.
Should the title tag match the H1?
It does not have to. The title is read in search results and the H1 inside the page; they do different jobs. The title has to carry the brand and a differentiator, while the H1 can be shorter and more natural. As long as they do not contradict each other, differing is not a problem.
How many times should the keyword appear?
Once in the title and once in the description is enough. Repetition hurts readability and raises the odds of the description being rewritten. If it appears twice in a natural sentence that is fine; do not force it.
Do emojis increase click-through?
It varies. In some categories they draw attention, in others they undercut credibility, and the search engine may strip them anyway. Safer to skip them for corporate and high-ticket products; if you want to try, measure by page group.
Is an empty description better than a weak one?
On pages that surface mostly for long-tail queries, leaving it empty can be reasonable, because the engine picks a matching passage. On main category pages and important product pages, always write one: there you need to control the promise.
How long before a change shows up?
Recrawling a page takes anywhere from a few days to a few weeks. Wait at least a week after the change before measuring, and four weeks before deciding. A report read earlier than that misleads.
Do colour variants of the same product need separate titles?
If the variants sit on separate pages, yes, otherwise you create duplicate titles. The better solution is to consolidate variants onto one page and offer colour as a choice; both SEO and the shopper gain.
Do Turkish characters cause problems in titles and URLs?
Not in titles, where correct encoding is enough. In URLs, simplifying Turkish characters (i for ı, s for ş) is still safer; it removes the risk of breakage when links are shared and in some systems.
Can I hand this job entirely to AI?
You can hand over the generation, not the checking. Bulk generation run without acceptance criteria makes the same mistake on thousands of pages simultaneously. Building a template without writing fifty examples by hand produces the same result.
Is this a priority for a small store?
Yes if you are getting impressions, no if you are not. Below roughly five thousand impressions a month, invest first in the content and technical work that gets pages visible at all; meta optimisation converts existing visibility into traffic, it does not create visibility.