Use TAM to forecast SEO/AEO
Keyword-based SEO forecasting is basically fiction.
This week’s newsletter is sponsored by North Star Inbound and Peec AI
Keyword-based SEO forecasting is basically fiction. You take search volume numbers, apply a click-through rate you read somewhere, multiply by a conversion rate you’re guessing at, and then present this to leadership as an organic search revenue forecast. The compounding inaccuracy of three guesses stacked on top of each other is not a model. It is like predicting annual wheat harvests by looking at clouds from your bedroom window.
[Sponsored by Peec AI]
Peec AI now connects your AI search data directly to Claude, Cursor, Perplexity, and more through MCP and build custom workflows.
Content briefings, competitive snapshots, Monday morning digests, sentiment drop alerts - all of it, inside the tools your team already uses.
See how it works →
I remember when I first started at SurveyMonkey and had to present one of these forecasts, which I put together using the exact steps above. In the midst of my presentation, I suddenly realized I didn’t believe a word I was saying. I had zero confidence that my forecast, as I detailed it, could ever come true. I committed to finding a better way to do this before having to deliver another one of those presentations.
The problem isn’t just the methodology of the forecast; SEO has too many unknown dependencies to forecast cleanly from the bottom up. Seasonality, CTR adjustments, and even personalization can detonate a keyword-based forecast, as a tiny change can drive the end state far from any predicted reality. And, all of this is before you factor in what’s happening to search right now with the rise of AI.
LLMs are eating the top of the funnel, so queries that used to generate clicks are now being answered with an AI Overview and soon with default AI Mode. The user gets what they need and never visits your site, so even if the search volume forecast were accurate, the follow-on clicks would not be.
Start at the top
The answer is to start with a top-down forecast: begin with the end state and then build a model to reach it. The top is the Total Addressable Market, TAM.
Many teams use TAM for forecasting, but it’s rare to see SEO teams doing so because they decided they were different and invented their own broken process instead building on black box mystique and authoritative “it depends” answers.
Starting from the total addressable market and working downward means your model shares a common framework with every other function that allocates resources. When you present a forecast that your CFO or CEO immediately recognizes the structure of, the conversation shifts from “I don’t understand this” to “I disagree with this assumption,” and that’s a much easier conversation to have.
Book me on Intro, and I can walk through your own TAM forecast in 15 minutes
The calculation
Here’s how to build a TAM forecast for SEO. Start with the raw universe of people who could conceivably be your customer. If you sell clothing to Gen Z women in the United States, go find that number. It’s roughly 35 million people. That’s your ceiling. It can’t grow by 40% because you had a viral content pop. That ceiling is a demographic fact, and it shifts slowly, which is exactly what you want in a forecasting input.
(Paid subscribers can download this HTML file to be run locally or in Claude.)
From there, you narrow down: Not every Gen Z woman in the US is in-market for what you specifically sell. If your product is premium, you’re already working with a subset of customers. If luxury buyers represent about 35% of that audience, your working TAM is now roughly 12 million people. Then you ask how often they buy. If this is an annual purchase category, that’s your annual TAM. If it’s biennial, cut it in half. These modifiers exist in your category, whether you account for them or not, and the more you add, the more accurate you will be.
Search as a modifier
The most important SEO-specific metric is how many of these people use organic search to make this kind of purchase. This is where most people get lazy and assume “everyone searches,” which is not true, and it depends on the product. Your product might be something that 99% of people buy in a physical store or online via social media channels, and do not search for.
Use whatever data you have. If you have nothing, be conservative. Overstating this number inflates every projection downstream, and the objective is to forecast cleanly. The last step is/was to take the total amount of search traffic and apply an assumed CTR to your estimated market share. That cut used to be simple. Organic search meant clicks, clicks meant visits, and that math became your SEO forecast.
The AEO edge
With the introduction of AEO, you now need to start splitting “organic search” into two very different things. There’s a click-generating search: someone types a query, Google returns ten blue links, and they click one. Then there’s answer engine behavior: someone asks an LLM a question, or gets an AI Overview at the top of a results page, and the answer is delivered without a click ever happening.
These two behaviors have the same starting intent but produce completely different outcomes for your business. A user who gets your product recommended by an AI assistant but never visits your site is not the same as a user who lands on your product page and converts. The former type of user might have SEO behaviors, but you are never going to capture them in your reporting.
Therefore, when you’re building your target market cut for SEO, you need to apply a click potential modifier. This isn’t a CTR from an early 2010’s click curve. Not every organic search touchpoint in your category will drive a visit, and that percentage is shrinking. Informational queries are getting hollowed out fastest from a click standpoint, so if someone asks a general question about your product category, there’s a real chance the LLM answers it. Transactional queries, the ones closest to purchase, are more resilient, but this will completely depend on the product type.
The honest approach is to apply a click potential rate to your organic search segment is to ask: What percentage of organic search interactions in your category actually result in a click to a website? That’s your modifier on total search potential.
Find a baseline
In your GSC data, look at your impression-to-click ratio across non-branded queries as a rough proxy. If you don’t, assume it’s lower than you think and build that conservatism into the model. A target market of people who use organic search is not the same as a target market of people who click through from organic search, and conflating the two will make your forecast optimistic in exactly the way that destroys trust with stakeholders.
This distinction between buyers and searches is the core of why the model works. Keyword forecasting treats every search as its own unit of analysis, which means you’re constantly chasing a moving target that responds to seasonality, news cycles, algorithm changes, and competitor content. Buyer-based forecasting anchors to the people doing the searching instead.
One person might search five different queries before converting. Another might search for one. The specific queries are just expressions of the same underlying intent, and what you actually care about is capturing the person, not any individual query. That number of searches is just a journey to the same destination.
The number of people you are targeting is more durable than keyword volume. A trending query can spike 400% in a month and collapse just as fast. The population of Gen Z women who buy premium clothing through organic search doesn’t move like that.
Market penetration
With a search potential sliced from your defined target market and your TAM, you now need one key assumption: what percentage of it can you realistically capture in your forecast? This is your market penetration rate.
If you have an existing product and historical data, look at your current penetration and project a growth rate consistent with other channels' assumptions. You can use paid search share of voice as a proxy, or find another data point that indicates your current penetration rate. A brand share-of-voice survey and general brand data can be very helpful for this effort.
If you don’t have internal data, you can be aspirational. Market penetration is also a target for the end of a period, not a number you hit on day one. Whatever your endpoint assumption is, you need to ramp to it over the forecast window. A model that assumes you jump from 5% penetration to 10% penetration in month one is not a forecast; it’s a wish.
Once you have that penetration rate locked in, the rest of the model is just a matter of logic. Multiply penetration by your target search market to get a user count. Apply a conversion rate to get buyers. Apply the average order value to get revenue. If traffic is a metric your stakeholders care about, you can work backward from the user count by applying an assumption about search frequency per buyer per period.
The key is that each variable in this chain has a real-world backstory. You can defend each one, and when the forecast drifts, you can identify exactly which assumption needs to be updated rather than rebuilding the whole model from scratch.
Transparently, as with any forecast, this will still be a guess, but you can refine it into a realistic forecast that you can use for business planning. When you start using TAM forecasting, it does not make you automatically more accurate, but it makes you wrong in ways your CFO can actually work with and work towards.
Download my basic forecast tool and start forecasting this channel's value.
[Sponsored by North Star Inbound ]
What happens when you combine best-in-class SEO with conversion optimization?
BigRentz’s traffic increased by 186% (85k), added 1950 conversions in 12 months.
Self Financial’s traffic increased by 50k/month, and added 685 new customers.
Lastly, Secure Data added 1968 phone calls.
North Star Inbound’s SEO strategies earn leads, conversions, and revenue..
Book a call for a free content audit and 10% off any engagement.
Intrigued? [Learn more here]