As I mentioned in my article, “Search Dog’s Approach to SEO in the Age of AI,” I’m in the camp that properly optimizing for people means you’re optimizing for search engines which means you’re optimizing for AI search. And part of that is making sure you’re actually optimizing for the correct KPIs. Before spending half your marketing budget on a tool promising real insight into your brand’s AI visibility, let’s make sure that you’re actually tracking the right foundational metrics.
Breakdown of KPIs discussed:
- Keyword Rankings
- Impressions
- Clicks
- Branded Search Volume/Traffic
- AI Visibility
- Referral Traffic
- Backlinks and Mentions
- Leads
- MQLs
- SQLs
- Sales
- Revenue
Traffic
Keyword Rankings — Freezing
Search results shift based on location, device, search history, and user intent, which means your position three for a given keyword may not exist at all for the person you’re actually trying to reach. The rank you’re tracking in a tool and the rank your audience sees are often two different things.
Real ranking doesn’t tell you much anymore. AI Overviews, featured snippets, local packs, and product carousels are absorbing the clicks that used to flow to organic results. A page can sit at position two and drive almost no traffic because everything above the fold has already answered the question. Rankings measure where you show up but they don’t say whether showing up did anything.
I once had a client in a healthcare-adjacent industry who was obsessed with tracking keyword rankings. For my contact, it was a KPI they reported on to their bosses so of course, they were KPIs I reported on to them. The issue was that rankings were improving but leads were falling. What I found was that, while SEMRush and Ahrefs were showing improvements, in reality the SERPs for the majority of the keywords had localized. So, while pages were still ranking really well, there were now local map packs and local brands being cited in AI Overviews. And, for any given search, a local company might appear in the number one organic spot over the larger national brand. I found where all the leads were going but if I had trusted the rankings, I wouldn’t have.
Rankings are still useful as a diagnostic tool. If a page that was performing well suddenly drops, that’s worth investigating. If a competitor is consistently outranking you for a specific term, that’s worth understanding. But using rankings as a primary KPI, as the number you report to leadership or optimize toward, is optimizing for something that has less and less to do with business outcomes. Track them in the background. Just don’t run your strategy around them.
Impressions — Freezing
Impressions measure how often your pages appear in search results. A page can rack up tens of thousands of impressions for queries it barely ranks for, pulling in numbers that look healthy in a report while contributing nothing to the business.
There’s also a trust problem with the data itself. Google acknowledged a logging error in Search Console that caused impressions to be over-reported starting May 2025, with fixes still rolling out as of April 2026. That’s nearly a year of inflated numbers that anyone using impressions as a primary KPI would have been reporting against with no way of knowing the data was off. That’s a hard case to make for a metric you’re supposed to be optimizing toward.
That said, there are moments where glancing at impressions is useful. If you’ve just launched a new site, published a batch of new pages, or made significant technical changes, a quick look at impression trends can tell you whether search engines are seeing the updates and movement is being made.
Clicks — Cold
Clicks feel like a logical measurement of success because more traffic should mean more opportunity. The problem is that zero-click searches have become the norm. Depending on the search topic, device, and location, somewhere between 60% and 65% of Google searches now end without anyone clicking on anything, and when AI Overviews appear that number jumps even higher.
This doesn’t mean your content isn’t working. It may mean your content is doing its job and Google is extracting the value before the user ever reaches your site. An interaction was still had, it just doesn’t show up in your analytics. Clicks still matter for mid and bottom funnel pages where conversion is the goal, but as a primary KPI they’re telling an increasingly incomplete story.
Brand Search Clicks/Volume — Toasty
When someone searches for your company by name, they already know you exist. That awareness had to come from somewhere, and branded search volume is where all of it eventually shows up as measurable intent. It’s also one of the more reliable indirect indicators of how your brand is performing in AI search. If LLMs are surfacing your brand in answers, some of those users will follow up with a branded search. A sustained increase over time is a reasonable signal that your visibility across AI platforms is improving.
Branded search growth is rarely the result of any single channel. It reflects the compounding effect of your whole marketing strategy working the way it should. That makes it difficult to attribute to SEO specifically, and in 2026 that’s probably the right way to think about it. You shouldn’t be treating any metric as belonging to a single channel anyway. Track it in Search Console, Google Trends, and your brand campaign data, establish a baseline, and watch for movement over time.
AI Search / AEO / GEO
AI Visibility — Cold
The same logic as organic impression applies here. AI visibility is worth paying attention to, but the tools and the data haven’t caught up to the hype yet. Research published by Rand Fishkin in early 2026 found that there’s less than a 1 in 100 chance of getting the same list of brand recommendations from an AI tool twice, which raises a pretty obvious question about what most AI tracking platforms are actually measuring. The methodology behind a lot of these tools isn’t transparent, and the price tags are steep.
My take is the same as it is on most things in SEO: if you’re doing the foundational work well, creating credible content, building real brand authority, and showing up consistently in the right contexts, AI visibility tends to follow. Tracking it is fine, but it shouldn’t be the thing you optimize for at the expense of everything that drives it. I’ll probably write more on this one specifically because there’s a lot to unpack.
Referral Traffic — Cold
AI referral traffic can give you a glimpse into whether your brand is showing up in LLM responses, and for a while it felt like a reliable way to measure that. The problem is that LLMs are increasingly designed to keep users inside the chat rather than send them elsewhere, and the referral data is getting less meaningful with every model update. Older versions of ChatGPT would surface ten or more source links at the bottom of a response. Newer versions often cite two or fewer, and OpenAI is making no secret of the fact that keeping users in chat is the goal.
The sessions that used to show up as ChatGPT referrals are now showing up as direct or unattributed traffic in GA4. The AI may have played a significant role in the discovery, but it’s not getting any credit for it in your analytics.
Backlinks and Mentions — Sweaty
Links remain a meaningful signal, but the conversation around them is shifting. LLMs don’t navigate the web the way search crawlers do, which means a brand mention in a credible context carries weight even without a clickable link attached to it. The goal here has always been the same: earn recognition from sources your audience and the algorithms trust. The bar for what counts as meaningful has just gotten broader. A link from a relevant industry publication still matters. So does a mention in a podcast transcript, a Reddit thread, or a trade newsletter that never links out.
Outcomes
Leads — Cold
Leads are closer to the right direction but not all leads are created equal. A raw lead count without any qualification attached to it can be just as misleading as a traffic number. If your contact form is being filled out by people who are nowhere near your target customer, optimizing for more of them is optimizing for the wrong thing. Lead volume matters, but only once you have enough visibility into lead quality to know what you’re actually counting.
Marketing Qualified Leads — Warm
MQLs are a step in the right direction because they represent leads that have cleared at least a basic threshold of fit or intent. For businesses that don’t have the infrastructure to track anything further down the funnel, MQLs are a reasonable place to anchor your SEO reporting. The limitation is that marketing’s definition of qualified and sales’s definition of qualified are often pretty different, and MQLs can still represent a lot of leads that never go anywhere. They’re a useful middle ground, not a destination.
Sales Qualified Leads — Hot
This is where things start to get meaningful. An SQL has been evaluated by sales and confirmed as a real opportunity worth pursuing. Tying SEO work to SQL volume requires the marketing and sales teams to actually talk to each other, which not every organization has figured out, but when you get there it changes how you make decisions. You stop optimizing for traffic that looks good and start optimizing for the traffic that actually enters the pipeline.
Sales — Hot
Closed deals are the clearest signal that your SEO is working. If organic search is contributing to sales, that’s the conversation every business owner actually wants to have. The challenge is attribution. Multi-touch sales cycles, offline conversations, and long decision timelines make it hard to give SEO full or even partial credit in a lot of cases. But that difficulty isn’t a reason to stop trying. Even data on which organic touchpoints show up in closed deal journeys is more useful than any ranking report.
Revenue — Flaming Hot
Revenue is the KPI everything else should be working toward. Not all sales are equal, and understanding which organic traffic sources are driving your highest-value customers is the kind of insight that shapes long-term strategy. Most small businesses aren’t set up to track this cleanly out of the gate, and that’s fine. The goal is to build toward it over time. If you can eventually draw a line from an organic search to a closed deal to a dollar amount, you have the foundation for making real resource allocation decisions.
