[{"data":1,"prerenderedAt":30},["ShallowReactive",2],{"blog-body-meta-description-audit-en":3},{"content":4,"toc":5},"\u003Cp>Writing a good meta description takes fifteen minutes. Auditing two hundred of them is a different job, and most sites never do it. The descriptions on a two-year-old site were written by several people, at different moments, against whatever the page looked like then. Many of the pages have changed. Very few of the descriptions have.\u003C\u002Fp>\n\n\u003Cp>This is a triage problem, not a writing problem. The goal is to find the small number of descriptions that are actively costing you clicks, fix those, and leave the rest alone. Rewriting everything is slower, riskier, and mostly wasted.\u003C\u002Fp>\n\n\u003Ch2 id=\"why-old-descriptions-rot\">Why old descriptions rot\u003C\u002Fh2>\n\n\u003Cp>Three forces work on a description after it is written. The page changes and the description does not, so the promise and the content drift apart. Search engines rewrite more of them than they used to, which means an untouched description is often not the one shown — and you cannot tell which ones are being replaced without looking. And your own understanding of the topic improves, so last year's phrasing now reads as a missed opportunity.\u003C\u002Fp>\n\n\u003Cp>None of that is dramatic on any single page, which is why audits get postponed. But the effect is cumulative and it lands in the same place: a search results page where your listing looks generic next to three competitors who wrote a clearer sentence about the same thing.\u003C\u002Fp>\n\n\u003Ch2 id=\"four-failure-modes\">Four failure modes worth searching for\u003C\u002Fh2>\n\n\u003Cp>Almost every description that deserves attention falls into one of four buckets. Sorting by these buckets is faster than reading each one and judging it.\u003C\u002Fp>\n\n\u003Cul>\n  \u003Cli>\u003Cstrong>Duplicates and near-duplicates.\u003C\u002Fstrong> The same description on two pages, or one sentence with a word changed. Common on category pages and local pages.\u003C\u002Fli>\n  \u003Cli>\u003Cstrong>Truncated mid-thought.\u003C\u002Fstrong> Descriptions written before the current pixel width, cut off before the point lands.\u003C\u002Fli>\n  \u003Cli>\u003Cstrong>Restated titles.\u003C\u002Fstrong> A description that repeats the title with the word \"official\" attached. It occupies the space without adding information.\u003C\u002Fli>\n  \u003Cli>\u003Cstrong>Missing or auto-generated.\u003C\u002Fstrong> Empty, or filled with the first paragraph of body copy, which typically starts with a greeting instead of a reason to click.\u003C\u002Fli>\n\u003C\u002Ful>\n\n\u003Cp>Two more worth a quick pass: descriptions that reference something no longer true — an old price, a campaign that ended, a feature that was renamed — and descriptions whose call to action is a command (\"Click here to learn more\") rather than a reason.\u003C\u002Fp>\n\n\u003Ch2 id=\"triage-before-rewriting\">Triage before rewriting\u003C\u002Fh2>\n\n\u003Cp>Before touching anything, sort the pages into three groups by what the page is for.\u003C\u002Fp>\n\n\u003Cp>Pages that earn traffic and convert: leave them alone unless they are in a failure bucket. They are working; the risk of making them worse is real. Pages that get impressions but few clicks: these are the priority, because the description is the variable you can change without touching the page, and the gap between impressions and clicks is the measurement you need. Pages with neither impressions nor clicks: do not start here. A better description will not fix a page nobody searches for.\u003C\u002Fp>\n\n\u003Cp>That ordering usually reduces an audit of two hundred pages to a rewrite list of twenty. The discipline is to resist improving pages that do not need it, which is the temptation that turns a two-hour job into a two-week one that never finishes.\u003C\u002Fp>\n\n\u003Ch2 id=\"duplicates-and-near-duplicates\">Duplicates and near-duplicates\u003C\u002Fh2>\n\n\u003Cp>Duplicates matter more than any other failure mode because they can affect which of two pages is shown at all, and because they are mechanical to detect. Sort your descriptions alphabetically and read the sorted list; identical or near-identical entries end up adjacent, which makes them impossible to miss.\u003C\u002Fp>\n\n\u003Cp>The fix is not to vary the wording. It is to decide whether the two pages should both exist. If they serve genuinely different searches, write descriptions that state the difference in the reader's terms — what this one is for that the other is not. If they serve the same search, the duplication is a symptom of a page problem, and rewriting the descriptions just hides it.\u003C\u002Fp>\n\n\u003Ch2 id=\"the-truncation-question\">The truncation question\u003C\u002Fh2>\n\n\u003Cp>Search engines cut descriptions at a width measured in pixels, which varies with the characters used, so a fixed character count is a rough guide at best. The practical approach is to put the important part in the first half and treat everything after it as optional detail.\u003C\u002Fp>\n\n\u003Cp>Test your own listings by reading only the first eight or nine words of each description. If those words do not tell the reader what the page is and why it matters, the opening is the problem, not the length. The most common version of this failure is a description that begins with the company name, or with a phrase like \"In today's fast-paced world\" that carries no information at all.\u003C\u002Fp>\n\n\u003Cp>Do not pad to use the space. A short description that gets to the point beats a long one whose second half is filler, because the filler is where the reader decides you have nothing specific to say.\u003C\u002Fp>\n\n\u003Ch2 id=\"rewriting-in-batches\">Rewriting in batches\u003C\u002Fh2>\n\n\u003Cp>Rewrite in batches of ten to fifteen pages that share a template — a set of product pages, a set of articles in one category. Shared structure means you can make one decision about the pattern and apply it, instead of solving the same problem fifteen times.\u003C\u002Fp>\n\n\u003Cp>For each batch, write down the single thing this group of pages has in common that a reader cares about, and the single thing that differs. Then each description states the common benefit and the specific differentiator. That structure produces descriptions that are distinct from each other by construction, which is what prevents the duplicate problem from coming back on the next pass.\u003C\u002Fp>\n\n\u003Cp>Keep a note of the pattern and the reasoning next to the batch. When someone writes the next ten pages, the pattern is the thing that stops the same rot from accumulating again.\u003C\u002Fp>\n\n\u003Ch2 id=\"measuring-whether-it-worked\">Measuring whether it worked\u003C\u002Fh2>\n\n\u003Cp>Give a rewrite four to six weeks before judging it. Search results take time to settle, and shorter windows produce noise that looks like a result.\u003C\u002Fp>\n\n\u003Cp>Compare impressions and clicks for the pages you changed, against the pages you deliberately left alone. The untouched group is the important half of that comparison — without it, a seasonal shift looks like a win. Expect the effect to be modest: a description changes the click-through rate of a listing, not its position, so a few percentage points on a page with moderate impressions is a normal outcome and, multiplied across twenty pages, a real one.\u003C\u002Fp>\n\n\u003Cp>If a rewrite made things clearly worse, revert it. Descriptions are cheap to change and there is no reason to defend a change that did not work.\u003C\u002Fp>\n\n\u003Ch2 id=\"when-to-leave-one-alone\">When to leave one alone\u003C\u002Fh2>\n\n\u003Cp>Leave a description alone when the page is performing, when the query it targets is brand-driven — searchers looking for you by name do not need persuasion — and when you cannot state the change in one sentence that would clearly help a reader. That last test is the useful one. If you cannot articulate why the new version is better for a person rather than for a metric, you are rearranging words.\u003C\u002Fp>\n\n\u003Cp>Audit on a schedule tied to something real: when a batch of pages is added, when a product line changes, when a template is redesigned. An audit triggered by an event finds the pages that actually drifted. An audit triggered by guilt tends to start, produce a half-finished spreadsheet, and stop.\u003C\u002Fp>",[6,9,12,15,18,21,24,27],{"id":7,"text":8},"why-old-descriptions-rot","Why old descriptions rot",{"id":10,"text":11},"four-failure-modes","Four failure modes worth searching for",{"id":13,"text":14},"triage-before-rewriting","Triage before rewriting",{"id":16,"text":17},"duplicates-and-near-duplicates","Duplicates and near-duplicates",{"id":19,"text":20},"the-truncation-question","The truncation question",{"id":22,"text":23},"rewriting-in-batches","Rewriting in batches",{"id":25,"text":26},"measuring-whether-it-worked","Measuring whether it worked",{"id":28,"text":29},"when-to-leave-one-alone","When to leave one alone",1791511468486]