How to Fact-Check an Agency's Published Content

Every content agency pitch includes results. A slide with a traffic chart. A logo wall. A line like "grew organic sessions 340%." None of that tells you whether the work was any good, because you are looking at a summary the agency built to persuade you. What you want is the raw material: the actual articles they published, on real client sites, and evidence those pages did something.

This is a different exercise from checking a proposal. A proposal is a promise. Published work is a record. You can pull it apart, read it, and check it against the live web without asking the agency for anything. Here is how to do that in an afternoon.

Start by getting three live client URLs, not a deck

Ask the agency for three client websites where they currently run the content program, plus the names of two or three specific articles they wrote for each. If they hesitate, that is information. Some clients sign NDAs, and that is legitimate, but no agency has every client under wrap. If they cannot name a single live page they produced, you are being sold a capability, not a track record.

What you are after is the URL of the actual article, not the homepage. "We work with Smith Dental" is worthless. "We wrote smithdental.com/blog/does-insurance-cover-invisalign" is auditable. Get to the page level or keep asking.

A quick way to find their work without asking

You can often skip the agency entirely. Many agencies list clients on their own site or in case studies. Take a client name, then search the client's blog for content and read the bylines and the writing. If several unrelated "clients" publish articles in an oddly identical voice and structure, you are looking at one agency's template applied everywhere. That is not automatically bad, but it tells you what you would actually be buying.

How do I check if the content they wrote actually ranks?

Open the exact article URL, pull the main question or topic it targets, and search that phrase in an incognito or private browser window so your own history does not skew results. Then look for that specific URL in the results. This is the single most useful check, and it takes about a minute per article.

Three outcomes:

• The page ranks on the first page for its target question. Good sign. The content earned a position for the thing it was written to answer.
• The page ranks somewhere on page two or three. Common and honest. Content takes time, and mid-page positions can still pull traffic. Not a red flag by itself.
• The page does not appear anywhere in the first several pages. Note it, but do not condemn it yet. The article may be recent, or targeting a term with brutal competition. What matters is the pattern across all the URLs you check.

Do this for all six to nine articles. If none of them rank for anything close to their topic, and the pages are more than six months old, the agency is producing content that sits unread. That is the outcome you are trying to avoid.

Add a free rank-and-traffic estimator

Free tiers of tools like Ubersuggest or the limited free lookups in Ahrefs and Semrush let you paste a URL and see estimated keywords and traffic for that page. The absolute numbers are rough. The useful signal is relative: does the page rank for anything, and are those keywords related to the client's business? A dental article that ranks for fifteen dental questions is working. One that ranks for two unrelated stray phrases is not.

Read the actual writing like a customer would

Traffic is only half of it. Pull up three of the articles and read them the way one of the client's customers would. Ask:

• Does this answer the question in the title, or circle it for 400 words first?
• Is there specific, local, or industry detail, or could you swap the business name and use it anywhere?
• Would you trust this business more after reading it, or does it read like filler written to hit a word count?
• Does the internal linking make sense, pointing to related articles and service pages, or is it a dead end?

Content that ranks but reads like nothing wins you positions and loses you the sale at the moment of decision. You want both. The read-through catches what the rank check cannot.

A worked example: auditing one agency in 40 minutes

Suppose you run an HVAC company in Tucson and an agency sends you a deck claiming strong results for a plumbing client. Here is the sequence:

1. Get URLs (5 min). You ask for live pages. They send tucsonplumbingco.com/blog/why-is-my-water-heater-leaking and two others.
2. Confirm the pages are real and dated (5 min). Open each. Check for a publish date and a real byline or client-voice writing. All three exist and were published eight to fourteen months ago. Old enough to have ranked if they were going to.
3. Rank check each (10 min). Incognito search "why is my water heater leaking" plus a Tucson modifier. The article shows up mid first page. The second article ranks page one for a long-tail question. The third does not rank at all.
4. Traffic estimate (5 min). Paste each URL into a free tool. The first two show a handful of ranking keywords, all plumbing-related. The third shows none.
5. Read-through (10 min). The two ranking articles are specific, answer the question in the first paragraph, and link to the client's service pages. The third is thin and generic, which explains why it went nowhere.
6. Verdict (5 min). Two of three articles rank and read well. One is a dud. That is a normal, honest hit rate. Compare it against the next agency you audit.

The point of the exercise is not to find perfection. It is to see the real distribution of an agency's output instead of the three best examples they chose to show you.

What published results should I be suspicious of?

Be suspicious of any result you cannot trace to a specific live URL, because unverifiable numbers are the easiest thing in the world to inflate. A "340% traffic increase" with no starting number could be four visits to eighteen. A screenshot of a ranking with no date could be from a term nobody searches. A logo wall proves someone paid an invoice, not that the work performed.

Other patterns worth a second look: results dated years ago with nothing recent, traffic charts that spike and never show what happened after, and "case studies" where the client is anonymous and no URL is provided. One anonymized case is fine. A whole portfolio of them means nothing is checkable.

Build a simple scorecard

Score each agency the same way so you can compare them side by side rather than on charm.

An agency that clears most of these is showing you a real record. When you run this against a few firms, the differences get obvious fast. A program built to publish, interlink, and rank on a steady cadence, which is the model ClearPath Content runs, should survive this kind of inspection without flinching. If an agency does not want you checking, that is your answer.

The takeaway: stop grading agencies on their decks. Get the URLs, check the ranks in an incognito window, read three articles like a customer, and run the scorecard. Forty minutes of your time separates the firms that actually publish results from the ones that just present them.