# How to Tell if My SEO Agency Is Working: 7 Metrics to Check Monthly **URL:** https://mettevo.com/blog/article/how-to-tell-if-your-seo-agency-is-working **Published:** 2026-10-06 **Updated:** 2026-10-06 **Author:** Oleg Silin **Category:** Blog | Mettevo > Learn how to tell if your SEO agency is working by tracking rankings, organic traffic, leads, conversions, and other meaningful SEO performance metrics. ![How to Tell if My SEO Agency Is Working](https://stage.mettevo.com/wp-content/uploads/2026/10/Professionals_looking_at_laptop_…_20261006134936.jpg) --- Our decision rule: SEO is working when non-branded clicks to pages that bring leads or sales rise for three consecutive months and the monthly report explains every drop with data. Traffic and average position cannot show that. Seven metrics, illustrated here with data from nine Google Search Console properties we manage, make the check. A monthly SEO report test shows how to tell if SEO is working without relying on an agency's summary. It compares the figures an agency reports with the same figures from independent sources: Search Console, Chrome field data, a backlink index and analytics. A figure that does not reconcile is a question for the agency. Each metric below names its source, a pass rule and a fail rule. ## **Key Takeaways** - Traffic and average position cannot separate a shrinking market from a weak agency. mettevo.com's impressions fell 63% from February to August 2026 while its average position improved. - Track non-branded clicks. In the week of September 22 to 28, 2026, two brand queries produced 14 of mettevo.com's 53 clicks. - Links decay: Ahrefs found 66.5% of links to 2,062,173 sampled websites had rotted since January 2013. - Escalate when three or more of the seven checks still fail across three monthly reports. That threshold is our judgment, not a validated benchmark. ## **Why do traffic and rankings mislead in 2026?** Traffic is weaker proof of SEO because AI summaries cut clicks. Our own properties show a related gap between visibility and ranking. On [mettevo.com](http://mettevo.com), impressions fell from 390,704 in February 2026 to 146,119 in August, a 63% drop. Over the same months, average position improved from 26.75 to 23.41. Both figures come from Search Console data pulled through Ahrefs on September 10, 2026. A report built on traffic and average position cannot tell whether the market shrank, an AI answer took the click or the agency lost ground. The seven metrics below narrow those causes: they show whether rankings moved, whether brand demand changed, and where the agency's work can and cannot be isolated. ## **What are the 7 metrics in a monthly SEO report test?** Ask for all seven every month, with last month's column beside this month's. The table lists each number's source and what counts as a pass or a fail for the agency. Metric Source Pass Fail (red flag) 1\. Branded vs non-branded click split Search Console Non-brand clicks up three months running. Growth comes from brand queries. 2\. Position-mix shift Search Console Click share by position band, month on month. Only average position is shown. 3\. AI Overview impact Search Console Impressions, clicks and position side by side. Drops are blamed on the algorithm. 4\. Core Web Vitals delta Search Console report (field data) All three vitals good, none worse than last month. Lab score only. 5\. Backlink velocity vs decay Backlink index Net new referring domains, named. Authority score only. 6\. Conversion-adjacent events Google Analytics 4 Events per organic landing page. Sessions only. 7\. Report integrity Your own pull Figures reconcile and the pull date is stated. Click-through rate (CTR) does not equal clicks divided by impressions. ### **1\. Branded vs non-branded click split** Split clicks into branded queries, which contain your company or product name, and everything else. Brand clicks come from people who already know you, so they reflect existing awareness more than SEO work. Google added a [branded queries filter](https://developers.google.com/search/blog/2025/11/search-console-branded-filter) to Search Console in November 2025, but its [help page](https://support.google.com/webmasters/answer/17011259) says the filter is unavailable for sites with a low number of impressions and can misclassify queries. A manual list of brand terms is the fallback. In our data for mettevo.com, total clicks fell from 62 to 53, down 14.5%, between September 15 to 21 and September 22 to 28, 2026 (Search Console, Web search type). Two brand queries, "mettevo" and "mettevo digital agency", produced 14 clicks in both weeks. Treating every other query as non-brand, clicks fell from 48 to 39, down 19%. Brand queries made up 32 of the 22,650 impressions in the latest week, 0.14%, and 26% of its clicks. The query table attributes only 24 of the 53 clicks to a listed query, and Search Console leaves anonymized queries out of it, so brand is a floor. ### **2\. Position-mix shift** Average position blends every query a site appears for, so it can improve while clicks fall. On mettevo.com, average position on desktop improved from 20.37 to 18.35 week over week (September 22 to 28, 2026) while desktop clicks fell from 42 to 21. Ask instead for the share of keywords, impressions and clicks in each band, such as positions 1 to 3, 4 to 10 and 21 or lower, with last month beside it. One flower e-commerce site we manage shows why bands matter. In our data from August 1 to September 9, 2026, across 53,999 ranking keywords, positions 1 to 3 held 15.2% of keywords and 5.9% of impressions but produced 29.2% of clicks. Positions 4 to 10 held 17.1% of keywords and 54.0% of clicks. Positions 21 to 50 held 32.9% of keywords and 4.7% of clicks. So 32.3% of keywords produced 83.2% of clicks. It is one site and one 40-day window. ### **3\. AI Overview impact: the impressions-to-clicks gap** Put impressions, clicks and average position side by side for the same pages each month. If impressions fall while position holds, rankings are not the cause. Fewer searches or an AI answer on Google's own page are the candidates. Since August 31, 2026, Google's [Generative AI performance report](https://support.google.com/webmasters/answer/16984139) shows impressions inside its AI features by page, country, device and date. Impressions are its only metric. See also our [guide to AI Overviews and SEO in 2026](https://mettevo.com/blog/article/ai-overview-and-seo-in-2026-how-to-rank-optimize-and-track-visibility-in-googles-ai-results). Two properties show the gap. In our data, a neurology clinic's impressions fell 69.8%, from 21,501,993 in March 2026 to 6,494,585 in August, while its average position moved from 5.2 to 8.0. An AI interior-design product held impressions flat from July to August 2026, at 341,365 and 341,942, while clicks fell 23%, from 6,754 to 5,214, and average position slipped from 13.1 to 15.8. We cannot attribute either case to AI Overviews: Ahrefs Brand Radar had no configured reports for them, and our Ahrefs-connected Search Console feed has no AI surface. ### **4\. Core Web Vitals delta** Ask for the month-over-month change in field data, not a lab score such as a Lighthouse result. Google's good thresholds are Largest Contentful Paint (LCP) at 2.5 seconds or less, Interaction to Next Paint (INP) at 200 milliseconds or less and Cumulative Layout Shift (CLS) at 0.1 or less. A site passes a metric when at least 75% of page views meet the threshold. Search Console's [Core Web Vitals report](https://support.google.com/webmasters/answer/9205520) covers the last 28 days, so a fix takes most of a month to show. Our [Core Web Vitals guide](https://mettevo.com/blog/article/core-web-vitals-optimization) covers fix order. We did not pull field speed data for the nine properties, so we have no portfolio delta to show for this metric.  ### **5\. Backlink velocity vs decay** Each month, ask for new and lost referring domains (the unique sites that link to you) plus the names of the ten best new ones. Links decay on their own. [Ahrefs found](https://ahrefs.com/blog/link-rot-study/) that 66.5% of the links pointing to 2,062,173 sampled websites had rotted since January 2013, so a flat net count means new links only offset losses. Google lists buying or selling links for ranking purposes as [link spam](https://developers.google.com/search/docs/essentials/spam-policies), which is why domain names matter more than an authority score. We did not pull backlink data for the nine properties either, so this metric also lacks a portfolio example. Treat a count of links built, or a rising authority score without domain names, as a fail. ### **6\. Conversion-adjacent events** Leads and sales are the final proof, but a small site records too few of them to read each month. Ask for conversion-adjacent events by organic landing page in Google Analytics 4 (GA4): calls, form starts, quote requests and add-to-cart clicks. Google defines a key event as an event that measures an action particularly important to the success of your business. Volume decides which events you can read. A regional accounting firm in our portfolio recorded 53 clicks in August 2026. At an assumed 5% lead rate, that yields about three leads a month, too few to show a trend. Calls and form starts give more events to compare. ### **7\. Report integrity: do the figures reconcile?** Recompute one figure yourself. Google [defines CTR](https://support.google.com/webmasters/answer/7042828) as clicks divided by impressions, so any CTR column can be checked with a calculator. Ask for the pull date beside every month-over-month comparison, so you know how complete the latest month was. Our own export fails this check on one property. For an AI interior-design product, the July 2026 CTR of 1.98% equals 6,754 clicks divided by 341,365 impressions. For mettevo.com, the same export gives an August 2026 CTR of 14.37% on 146,119 impressions, which implies about 21,000 clicks. Search Console's Performance report for September 22 to 28, 2026 shows 53 clicks on 22,650 impressions, a CTR of 0.23%. Across different windows, impressions agree in scale, but the two rates differ by a factor of about 60. We have not traced the cause, so we use no monthly CTR from that export in this article. ## **What should you do if an agency fails 3 or more of the 7?** Speed fixes need developer access and event tracking needs the client's analytics set-up, so one or two failed checks may sit outside the agency's control. Three or more failures is our escalation line, a judgment rather than a validated benchmark. 1. Send the failed checks in writing and ask for the next report to include them, with data exports attached. 2. Ask for read-only access to Search Console, Google Analytics 4 and the backlink tool, then verify two figures yourself. 3. Wait for three monthly reports. Google's [SEO Starter Guide](https://developers.google.com/search/docs/fundamentals/seo-starter-guide) says some changes take effect in a few hours and others could take several months, and suggests waiting a few weeks to assess results. 4. If three or more checks still fail with access granted, change vendor or scope. Google's [hiring guidance](https://developers.google.com/search/docs/fundamentals/do-i-need-seo) says to be careful if a company is secretive or won't clearly explain what it intends to do. Before you switch, read our [small-business guide to SEO providers, pricing and red flags](https://mettevo.com/blog/article/best-seo-services-for-small-business-in-2026-top-companies-platforms-and-pricing-compared), which includes ten questions to ask before hiring an agency. To hold Mettevo to the same seven checks, start from [our SEO services](https://mettevo.com/seo). ## **What do these data not show?** Read the examples above as cases, not benchmarks. Client names are withheld and each client is described by sector. Mettevo's own figures are named. - Sample: nine Search Console properties in one agency's Ahrefs account, from e-commerce, healthcare, local services, professional services and software. It is not a random sample of the web. - Windows: monthly figures cover February to August 2026, pulled September 10, 2026. Position bands cover one site from August 1 to September 9, 2026. The mettevo.com weekly figures, pulled September 28, 2026, compare September 15 to 21 with September 22 to 28, and the second week falls inside Google's [September 2026 spam update](https://status.search.google.com/incidents/XhUDXP7A67iHCD2kmbVu), begun on September 24. - Not measured: field Core Web Vitals, backlinks and analytics events, so metrics 4 to 6 carry tests and public benchmarks, not our numbers. - Not proven: the AI Overview effect is inferred from the shape of the data, and the mettevo.com CTR discrepancy is unexplained. ## **FAQ** ### **Is SEO dead now with AI?** No. [Google states](https://developers.google.com/search/docs/appearance/ai-features) there are no additional requirements to appear in AI Overviews or AI Mode, and that SEO best practices remain relevant. AI summaries still cut clicks: clicks on traditional results in 8% of visits with an AI summary against 15% without (March 2025). Judge SEO on clicks and leads. ### **What is the 80/20 rule in SEO?** The 80/20 rule, or Pareto principle, says roughly 80% of outcomes come from 20% of causes. In SEO it shows as click concentration. In our data, one florist site's positions 1 to 10 held 32.3% of ranking keywords and produced 83.2% of clicks (August 1 to September 9, 2026). Check that your report shows this concentration. ### **Can ChatGPT do a SEO audit?** Partly. A chat model works on the data you give it, so it can review a crawl export or a Search Console table and flag arithmetic that does not add up. It has no live view of your field speed data or backlinks unless you export or connect them. Use it to recompute CTR, then confirm figures in the tools. ### **How long does it take for SEO to kick in?** Google's [SEO Starter Guide](https://developers.google.com/search/docs/fundamentals/seo-starter-guide) says some changes take effect in a few hours while others could take several months, and suggests waiting a few weeks to assess results. Judge an agency on three consecutive monthly reports. Core Web Vitals fixes need most of a month to show, because [Search Console's report](https://support.google.com/webmasters/answer/9205520) covers the last 28 days. ### **Is SEO still worth it in 2026?** Yes, where the return is measurable. In our data, five of nine properties lost 20% to 59% of their impressions from July to August 2026, while one regional accounting firm's clicks rose from 18 in April to 53 in August. That is a single firm, not a benchmark. Track non-branded clicks and leads, not impressions. ### **Is SEO difficult to learn?** The basics are learnable: Google publishes a free [SEO Starter Guide](https://developers.google.com/search/docs/fundamentals/seo-starter-guide), and the seven checks above need only Search Console, a speed report, a backlink export and Google Analytics 4. The hard part is interpretation, such as telling a shrinking market from a lost ranking. Compare impressions with average position before blaming anyone.