Sam Torres: Here's What to Do When Your Website Traffic Tanks

Sam Torres flew in from Atlanta last night to talk about SEO traffic drops at MKE DMC. She's the CDO at Gray Dot Company (a strategic and technical SEO agency), and has a rare combo of technical chops (former front-end dev) and agency experience (14+ years of this).

Sam treats traffic investigations like murder mysteries: limited suspects, work backwards from evidence, don't assume you know "who dunnit" before you start.

Stop Panicking, Start Investigating

The first thing Sam said: about 30% of her "emergency" investigation calls turn out to be broken Google Analytics. Not algorithm updates. Not penalties.

So, is there really a problem?

• Huge gaps between Search Console and your analytics
• Traffic down across all channels
• Zero conversions anywhere

These are signs that your tracking is broken, not your SEO.

She's seen compliance banners kill analytics, migrations where someone forgot tracking codes for months, Google Tag Manager tags fighting each other. "People do weird stuff," she said.

So first, verify it's not a tracking issue.

Once you confirm the traffic drop is real, then you slice the data. Not just "traffic is down" but where exactly. Which templates? Which topics? Which devices? She mentioned using SiteBulb to automatically identify page templates, then pulling performance data to see which ones are bleeding.

Sometimes you don't need to investigate at all.

She had a client panicking about traffic drops. Turns out their conversions were fine. Actually improved. They lost impressions on stuff that didn't matter while their actual leads got better. "Then why are we gonna do this?" she asked them.

Another story: a bank used to rank for the letter "T". Just the letter T. Got two clicks in 18 months, but a ton of impressions. An executive was upset when they lost these impressions after a migration, because they were never informed that total impressions shouldn't be a leading metric.

Usual Suspects

She walked through the standard culprits with character names.

Content problems are usually cannibalization (multiple pages fighting for the same keywords), thin content from years of "publish 40 blog posts a month" strategies, or broken internal linking. She called out Gus Pelogia's series on using Screaming Frog with ChatGPT's API to analyze internal links at scale.

Technical issues still matter despite what people say. She was emphatic about this - called it "lies and slander" that technical SEO doesn't matter anymore. Google can render JavaScript but doesn't always bother. Other LLMs don't render it at all. Critical stuff should be in the initial HTML response, not waiting for JavaScript to load.

Lost backlinks hurt more than people realize. Publishers change content constantly, so those links you earned years ago might be gone. Old PR campaigns age out. Links pointing to 404s don't help anyone. She's not looking for spammy links here, just what's outdated or broken.

Algorithm updates happen constantly now. Remember when we got two or three algorithm updates a year and USA Today would write about them? Now they're endless. She recommends Marie Haynes' tracker over Google's official log because Marie comments on unconfirmed updates too. Also the GSC Guardian Chrome extension that overlays algorithm dates on your Search Console data.

SERP changes are everywhere. Hard to track in Search Console alone, easier with tools that screenshot results. If you haven't claimed your knowledge panel, do it. You can submit to Wikidata even if editing Wikipedia directly is too hard.

Competitors stepping up happens more than you think. She once showed a mortgage company executive that they had 8 people on their marketing team while their target competitor had 400. "I'm gonna need more resources."

Users change constantly. Rankings stay the same but traffic drops because fewer people search for that term, or they're researching on TikTok instead. She uses Pinterest Trends, TikTok Trends, YouTube comment sentiment analysis. There's a tool called Gummy Search for Reddit analysis.

Sell the Fix, Not the Problem

When you do find problems, you need to sell the fix. She treats it like product management - what's the value, what's the risk, what happens if competitors fix this while you don't. Document the KPIs so you can track success months later when you've forgotten what you were doing.

Strategic alignment with company goals matters. Revenue impact. Level of effort. Urgency. Why now instead of later.

They give away a free product roadmap template that auto-prioritizes based on effort, risk, and impact. Most agencies would charge for that.

Real Talk

The most honest part was about resetting expectations. The traffic from 2019-2020, especially for online businesses, isn't coming back. SEO changes constantly and it's exhausting but also what makes it interesting.

She mentioned a client that creates pure educational content. They had a post that ranked #1 for years. Now it's in free fall because of AI overviews. Nothing they can reasonably do except have hard conversations about what's realistic.

Sometimes the investigation shows you don't have a problem. Sometimes it shows you have a problem you can't fix. Sometimes it shows your competitors are just better now and you need to step up.

You know which one it is. "Is the juice worth the squeeze?"