The most common problem in our audits was text nobody re-read.

The short answer

The most useful check in our audits was the least technical one: reading the page. On one site, an audit tool correctly reported that most pages had no meta description. The same tool had nothing to say about every page built from one template carrying the same image alt text, written for one page and wrong on all the others. The first problem is an empty field. The second is a wrong one, and wrong fields look fine to software.

Five kinds, and the fix for each

Across our site audits, the leftovers fell into five groups. The examples are described loosely on purpose, because the businesses are not named.

1. A name from somewhere else

What we saw: an About page whose title, and whose title for social sharing, carried the name of an unrelated business. Our guess is that the page was started from another site's template.

The fix: correct both titles, because they are set in different places. Then search every page and every template for that other name, since a page copied once is usually copied more than once.

2. Instructions left in

What we saw: a published write-up containing a paragraph that told the site owner what to write there.

The fix: replace it with the real account: the problem, what was done, who did it and what was delivered. If those facts are not available yet, take the page down until they are. A half-finished page does more harm than a missing one.

3. One field, every page

What we saw: the alt text above, and a second version on a different site: a set of category pages that each showed the same mixed list under their own label. In both cases the template was reading from the wrong place.

The fix: repair the template's connection to its data, not each page. One correction then fixes every page built from it. Give informative images a description of what is actually in them, and leave purely decorative images with an empty alt attribute.

4. Filler typed while testing

What we saw: a published description made of repeated keyboard filler, the kind typed to check whether a form saves. Lorem ipsum left in FAQ answers.

The fix: write the real text, then add a rule to whatever publishes your content: nothing goes live with an empty or placeholder field. Many content systems can make a field required or block publishing until it is filled.

5. Unfinished thoughts

What we saw: case studies with sentences that stopped partway, or that guessed at what a client might have achieved and wrote it as fact.

The fix: finish them with facts the client has confirmed, or cut the sentence. Where a result has not been measured, say so. A buyer reading "results pending" learns you are careful. A buyer catching a guess learns the opposite.

The same pattern turned up in lighter, outside-in reviews of other sites: work labelled with the wrong service, a sentence about a different line of business on a service page, and blocks repeated on a contact page.

Why tools pass it

Audit tools check that something is there. They cannot check that it is right.

• A title tag exists, so the "missing title" check passes, whatever company it names.
• An image has alt text, so the "missing alt text" check passes, whatever image it describes.
• A page loads and has plenty of words, so the "thin content" check passes, even when some of those words are instructions to the author.

The one automated signal that does help is duplication. A tool that reports duplicate titles, descriptions or alt text will often be pointing at a template fault, even though it calls it a duplicate problem rather than a wrong one.

Why it matters

It matters to search engines, and it matters more to the person reading.

Google says it uses alt text, together with computer vision and the content of the page, to understand what an image is about. An alt text describing a different page is telling it the wrong thing everywhere it appears. Google also says that identical or similar descriptions on every page are not helpful when individual pages appear in results, and that each description should describe its own page.

The bigger cost is the buyer. A business with another company's name in its About page title has told a careful buyer that nobody checked. The search problem is small. The trust problem is not.

How to find it on your site

About an hour for a small site, most of it reading.

1. Search for placeholder words. Look for "lorem", "ipsum", "your company", "insert", "sample", "TBD" and square brackets. Most crawlers have a custom text search that checks every page. A Google site: search works too, but only finds pages Google has indexed.
2. Export and sort. Put every page's title, description, main heading and image alt text into one sheet. Sort each column. Identical values on pages that should be different are where a template is reading the wrong field.
3. Read one page per template, then two more. If one page is wrong, check the same field on two others built from the same template. Template faults repeat. A typo does not.
4. Check the sharing title separately. The title social platforms show can be set in a different place from the page title.
5. Read the case studies as a stranger would. Anything that guesses at a result, stops partway or reads like a note to yourself should be finished with facts or taken down.

What the same search found on our own site

We ran steps 1 and 2 on getborvis.com on 3 October 2026, on every page of the site except the four published with this one.

The placeholder search is the instructive one. We publish prompt templates with words like [service] and [competitor] in brackets, and two of our pages use "placeholder" as an ordinary word. So the search found 19 things and none of them were mistakes. On a site that publishes templates or examples, expect noise, and read every match rather than counting them. The shared heading is a choice: each page of the list has its own title and description, so we left it.

What to fix first

1. Anything on a page that sells. Titles, testimonials, prices and FAQ answers on service and pricing pages.
2. Template fields. One fix to a template corrects every page built from it, so it is the best return on time.
3. Everything else, in order of how often the page is visited.

Then add the check to whatever you do before publishing. On most of these pages the layout had clearly been finished with care. The words had not had the same attention.

Limits

• The examples come from a small set of audits and lighter reviews. They show that this happens, not how often it happens across the web.
• Text extracted by a crawler can show repeats that are not on the rendered page. One repeated-block finding from a lighter review is unconfirmed for that reason and is not described above.
• For the free audits we do not know what was fixed afterwards, so nothing here describes those sites today.