How to Verify an SEO Agency's Real Results

Every agency pitch includes a results slide. A traffic curve going up and to the right, a couple of keyword wins, maybe a logo wall. None of that tells you whether the work was any good, or whether the client would have grown anyway. Before you hand over $199 to $499 a month for a year, you can check the actual published work yourself. It takes about an hour and needs no special access.

The goal is simple: find the real pages an agency produced, confirm they exist and rank, and judge whether they read like something a human would trust. Here is the process I use when a business owner asks me to sanity-check a competitor's provider.

Get the raw material: real URLs, not screenshots

A traffic graph is not evidence. It is a picture of a graph. Ask for something specific instead: three to five live URLs the agency wrote and published in the last 12 months, with the client named. If they will not name clients, ask for the URLs anyway — you can often identify the client from the domain.

Common dodges and what they mean:

• "We can't share client work, it's confidential." Published web pages are, by definition, public. This usually means the work is thin or ghost-produced under a name they would rather you not check.
• "Here's a case study PDF." Fine as a starting point, but the PDF is marketing. Ask for the underlying URLs the case study refers to.
• "Results take time, we can't show much yet." Reasonable for a six-week-old client. Not reasonable for an agency that has been operating for years. They should have a back catalogue.

Once you have five URLs, you have everything you need.

Confirm the pages actually rank

A published article is not a result. A published article that pulls in searches is. For each URL, run these checks:

1. Paste the URL into a keyword tool (Ahrefs, Semrush, or the free Google Search Console export if it is your own site). Look at how many keywords the page ranks for and its estimated position. A page that ranks for nothing after 8+ months is a page that did not work.
2. Search the article's main topic in an incognito window. Does the page appear on page one or two? If it is buried on page five for its own headline topic, it is not earning anything.
3. Check the publish date against the ranking. A page published three weeks ago that ranks for nothing is normal. A page from 18 months ago that ranks for nothing is a pattern, especially if three of your five samples look the same.

You are not looking for perfection. You are looking for a hit rate. If four of five sample pages rank for real, relevant terms, that is a working program. If one ranks and it happens to be the one on the case study slide, you have found a cherry-picker.

Read the writing like a customer, not a marketer

Ranking is half the picture. The other half is whether the content would make a real buyer trust the business. Read each article the way one of your customers would and ask:

• Does it answer the question in the first paragraph, or stall for 300 words of preamble?
• Are there specific numbers, steps, and examples, or is it padded with phrases like "in the modern era"?
• Does it sound like the business owner, or like a template with the town name swapped in?
• Would you, reading it, believe this company knows what it is doing?

Copy a paragraph into a plagiarism checker and an AI-detection tool. Neither is perfect, and AI detectors especially throw false positives, so treat the result as a flag rather than a verdict. But if every sample comes back as heavily duplicated across other sites, the agency is spinning one article across many clients. That is worth knowing.

Is a rising traffic graph proof the agency caused it?

No — a rising traffic graph is not proof the agency caused the growth, because you cannot see the counterfactual on a single line chart. Traffic can climb for reasons that have nothing to do with the agency's work: a seasonal spike, a viral news mention, a competitor going out of business, a brand campaign running in parallel, or a Google update that happened to favour the site. To separate agency impact from noise, tie the growth to specific pages. Ask which URLs the agency produced, then check whether those exact pages are the ones gaining traffic in the tool. If the growth is concentrated on the homepage and the brand name while the agency's articles sit flat, the agency did not cause the curve. If the growth lines up with the publish dates and URLs of their work, that is a real signal.

How long should I wait before judging an agency's content?

Give new content roughly four to six months before you judge whether it ranks, but judge quality and cadence from day one. Rankings for anything competitive genuinely take months — a page needs time to be crawled, tested by the search engine, and to gather relevance signals, and this timeline varies a lot by industry and competition. What you should not wait on is whether the writing is accurate, in your voice, and published on the promised schedule. Those are visible in the first week. So the honest framework is two clocks: a fast clock for quality and consistency, and a slow clock for ranking outcomes. An agency that is missing its publish cadence in month one will not magically improve its ranking odds in month six.

A one-hour audit checklist

Run this before any contract. Suppose you are a dental practice evaluating a provider that pitched you last week.

1. Request five live URLs from clients in the last year. (10 minutes, mostly waiting on their reply.)
2. Run each URL through a keyword tool. Note keywords ranked and best position. (15 minutes.)
3. Incognito-search the main topic of each. Record where the page lands. (10 minutes.)
4. Read two articles fully. Score them: does it answer fast, sound human, include specifics? (15 minutes.)
5. Check publish dates against rankings. Old page + zero rankings = red flag. (5 minutes.)
6. Spot-check for duplication across the agency's other clients. (5 minutes.)

Tally it up. Four or five pages that rank and read well is a strong sign. Two working pages out of five, both old, is a program treading water. Zero, or a refusal to hand over URLs at all, is your answer.

What good looks like

A provider worth paying should be able to say: here are the pages, here are the terms they rank for, here is the publish schedule we held to, and here is how those pages connect to the rest of the site. They should be relaxed about you checking, because the work survives checking. When we scope engagements at ClearPath Content, we expect prospects to run exactly this kind of audit on us — it is the fastest way for a sceptical owner to separate real work from a good slide deck.

The practical takeaway: never buy from the results slide. Get five URLs, spend an hour with a keyword tool and your own reading eye, and let the published work argue for itself. An agency doing real work will pass in an afternoon. One that isn't will find a reason you can't see the pages — and that reason is the result.