On-page SEO
optimization in Chicago

Specifications, prices, comparisons and schedules are the parts of a page people came for. They are also the parts most often built as images, as a stack of divisions styled to look like rows, or as a paragraph pretending to be a list.

A table of data should be a table. That gives it headers, a relationship between a cell and what it describes, and a structure that a machine can read. Styled boxes look identical and mean nothing.

Headers are the part that gets left out. A row of bold text is not a header unless it is marked as one, and without that, somebody using a screen reader hears a stream of numbers with nothing attached to them.

Wide tables are the recurring problem on a phone. The usual fixes are to let the table scroll horizontally inside its own container, or to restructure it into stacked blocks. Both work. Shrinking the text until it fits does not.

Lists carry meaning too. An ordered list says the sequence matters, an unordered one says it does not, and a paragraph of items separated by commas says nothing at all.

Then the case that costs the most: a specification published as an image. The values inside it are invisible to search, unusable on a small screen, impossible to copy, and unreadable to anybody using assistive technology. It is the single most common way a page hides its most valuable content.

Captions and summaries are worth adding. A short line saying what the table shows, and what the units are, answers the question the reader would otherwise have to infer.

Where a table holds prices or specifications, structured markup lets search engines use the values directly. That is a separate layer from the visible table and it is only as good as the table underneath it.

The check is quick. Turn off the styles and read the page. If the data still makes sense in order, the markup is right. If it becomes a heap, the markup was decoration.

In Chicago, the on-page work we sell is a page-by-page audit and rewrite of what decides ranking and conversion: the title, the heading outline, the copy, links between pages and structured data. The entry tier, on-page fixes, starts from ~$550; full SEO optimization sits near $1,100. The schedule goes into the estimate once we have reviewed the site.

Open two pages built on the same template and read them side by side. Menu, footer, cookie notice, contact form, a testimonial slider, the list of other services. Then the body, often a few hundred words. On many sites the shared furniture outweighs the text that makes a page different.

Search engines separate a template from the main content, and they do it well. Trouble starts when the main content repeats. A block of benefits pasted onto thirty service pages. A FAQ copied with one word swapped. An intro that changes only the product name. All of it sits in the main content area and counts as the page's own text. And it is the same everywhere.

The cloned FAQ deserves its own mention. Questions and answers get written once for the root service page, then copied to each location or variant with the name of the place inserted twice. It reads as the most specific content on the page. It is the least specific.

Catalogues produce the same effect with variants. Twelve colours of one item on twelve URLs, each carrying the full shared description, are one page repeated twelve times.

Measured, it looks like this. Break each page into runs of five consecutive words. Count how many runs a page shares with its sisters. When a large part of a page also appears on every sister, the page owns very little. A search engine choosing which version to index then has no reason to keep more than one.

Search Console shows the symptom. Such pages sit under Crawled, currently not indexed, or under Duplicate without user-selected canonical, with a sister picked as the canonical. Nothing is technically broken. The pages are too alike.

The work starts with a crawl that stores the main content of each page, with navigation stripped. Every pair within a template is then compared. Out comes a table of pairs, sorted by overlap, with the shared passages printed next to each. That table sets the order of work.

Repeated blocks get one of three treatments. Some leave the page for a single shared home, a benefits page or a general FAQ, and the page links there. Some stay but shrink to one line. The rest are rewritten page by page around something that truly differs: the process, the inputs, the constraints, the decisions that belong to this service.

Synonym swapping fails. A paragraph with every fourth word replaced is the same paragraph to a reader and to a comparison. New text has to say something the sister page does not say.

Rewriting starts from the question each page answers. Sister pages about one service can split the ground between them. One covers how the work is scoped, another the order of checks, a third the ways it goes wrong. A reader searching for any of those lands on the page that treats it properly.

Writers get a brief that stops the drift back into shared text. It names the subtopic, lists the passages already used on sister pages and flags the phrases to avoid. Without it, rewrites converge on the same safe sentences within a few pages.

A canonical tag is no cure here. It tells a search engine which of two duplicates to keep. Neither page becomes worth indexing because of it. Noindex on the weaker sister is honest only when that page has no separate purpose.

Location pages are the hardest case. The service is identical in every city, so the honest material that differs is thin. Invented local colour is worse than repetition. It is untrue, and readers notice. The better route is depth on the service itself, divided by subtopic, so each page answers a different part of the question.

Thresholds are a judgement call, so we state ours. A page whose main content shares more than about a third with any sister gets flagged. Past half, it is rewritten or merged before anything else on the site is touched.

After the rewrite we measure again with the same method and log the overlap for each pair next to its index status. Indexing responds slowly, over weeks. Overlap is the early signal. Index status is the late one, and the one that pays.