COVID uncovers a lot of fragility in our lives. One of them: how companies reach customers. Many markets move online overnight amid the painful realization that not being visible online is a very fragile place to be.
The second-order result of this big wave of markets going online is that SEO is in more demand than ever because it’s a scalable and cost-efficient channel. However, its benefits don’t come without a price.
One of the most fundamental challenges of SEO is predicting returns. Investing without a guaranteed ROI is called gambling. But who is foolish enough to gamble with their business?
We only know a few parameters that move the SEO needle with 100% certainty, which makes predicting SEO traffic and revenue hard. Not knowing how to solve this problem puts you in a weak, reactive position. It undermines your credibility and strips you of the resources you need to get things done.
The solution is to “follow the money”, a.k.a. predict how much revenue your strategy will return. Win leadership over with a business case so solid, they can’t say no!.
As cherry on top, Similarweb was so kind as to provide conversion rate benchmarks for different industries.
The ROI of SEO
I asked my Twitter followers why they thought SEOs have such a hard time getting resources. The majority of responses said something along the lines of “SEO isn’t a priority”. While that’s true, the underlying problem is that SEOs don’t make good business cases that show the revenue potential.
In the guide to inhouse SEO, I wrote, “Don’t sell traffic, sell revenue. If you make traffic or rankings your goal, it’ll be harder to get resources than revenue or leads.”
This guide is a manual for selling revenue in two steps:
- Project traffic
- Tie traffic projections to revenue
How to think about SEO traffic projections
Uncertainty in SEO can’t be an excuse for not making a business case. You have to make one if you want to get stuff done because engineers, writers, or designers mostly prioritize roadmaps by revenue impact. If we can’t promise a specific outcome, we can at least narrow the impact down and attach outcome probabilities.
You can narrow traffic predictions down to five areas by varying degrees:
Hard – whole site
Traffic predictions to the whole site aren’t really narrowed down. Sometimes, it’s unavoidable. For example, page speed often impacts the whole site (not always).
Medium – page type
Assume you know an SEO recommendation will increase the traffic/rankings to a specific page type, say category or product pages. In that case, you can look at the traffic that all pages of that type get right now and estimate the incremental impact of an optimization.
Projecting on a page type basis commonly applies to areas like meta-data optimization (think meta-title patterns), internal linking, or content and is much easier than to the whole site. Classic cases are meta-data optimization, adding content, building links or improving internal linking.
Medium – keyword syntax
Queries follow a certain syntax, e.g., “best x in y”, especially when you have many pages with the same template. If that’s the case, you can project traffic from keywords with the same syntax based on the traffic you already get from them or with your custom click-curve (more about that in a bit).
Projecting by keywords is even easier than by page because you don’t have to understand what a page (type) would be eligible to rank for.
Easy – traffic to a single page
Projecting traffic impact to a single landing page is the easiest way, though not every business follows a centralized structure. All you need to do is look at the traffic the landing page gets from organic search and internal referral traffic and then connect that to your SEO recommendations.
Easy – Traffic from a keyword
Projecting traffic increases from a keyword is the easiest way, but most companies don’t focus on single keywords. The rationale is the same as before
- create a custom click-curve,
- multiply projected traffic by the CTR for a specific position,
- estimate the incremental positions you can achieve with specific recommendations.
Now that you understand how to narrow traffic projections, let’s look at connecting this with traffic increases.
The step-by-step process to projecting traffic
The step-by-step process outlined here is not set in stone, but more of a rough guideline. Adjust it to your needs.
1. Identify keywords
I assume you already have a set of relevant keywords. If not, run a competitor keyword gap analysis with Similarweb’s Keyword Comparison or look for low-hanging fruit keywords (ranking on position #11-20). The goal here is to have a list of strategic keywords you want to go after.
2. Calculate your “traffic TAM”
Traffic TAM (total addressable market) is the total amount of traffic you can get for your set of strategic keywords if everything works out. The basic idea is to look at the sum of clicks you can get from your set of keywords, which is not the same as search volume.
One easy way to do this is with Similarweb’s Search Visits, where you can upload a keyword list and get the combined potential traffic.
Calculate your own traffic TAM:
- Create a keyword list in SW (under “Workspace”)
- Go to Research > Keyword Analysis > Organic > select your list
- Search Visits = your SEO TAM
Alternatively, build a click-curve to understand how much traffic you would get at each keyword position. If you don’t know how to make your own, you can take one from Sistrix or Ahrefs.
- Go to Keyword Explorer
- Enter your keyword
- Go to Traffic Share by Pages
- Look at the traffic at each position
Mind you that the click-curve in Ahrefs only works for a single keyword.
The last step is to get your target keywords’ ranking positions to know how much growth potential you have. So, to get your traffic potential, subtract the total potential traffic from the traffic you get or look at the gap between your keyword rankings and position one.
Now it’s time to put some SEO recommendations together! Audit your site or specific page types to understand where you can drive organic traffic. The audit should come after identifying keywords that haven’t maximized their traffic potential (usually because they don’t rank in the top 3), so you know what pages to focus on.
One heuristic to help you find optimization opportunities is comparing the pages you want to rank higher against competitors to get a sense of Google’s requirements. I would personally always start with a comparison against the top results – either for one keyword or for a set of keywords – to find out what you have to improve. In some cases, this can be a site-wide thing like UX or backlinks, but it often comes down to a page or page template level. The only exception is when you compete against different types of sites for the same keyword.
4. Impact estimation
The next step is estimating the impact of your recommendations on keywords and traffic. Many SEOs shy away from settling on predictions because they can’t guarantee for them, but the shift move is to accept that it will be imperfect. Start somewhere, manage expectations, and iterate over time.
You can either project incremental rankings or traffic. This means either saying, “We expect an X% organic traffic increase from doing Y” or “we expect rankings for a specific set of keywords to increase by X positions and a traffic increase of Y as a result.” I suggest the latter option because the logic is easier to follow, especially when the audience is not SEO experts.
The best way to project the incremental traffic increase from a set of keywords is to build a custom click-curve. The gangster move is building a custom click-curve by page type! However, you need to ensure you have enough data. Otherwise, you won’t have enough data about your average CTR on each position and your curve won’t be accurate.
When creating your click-curve, make sure to segment brand from non-brand keywords. Brand keywords tend to distort click-curve because they naturally have high click-through rates (think: navigational user intent), so it’s best to filter them out.
The impact of an SEO recommendation is a question of:
- Your backlink profile
There are no cookie-cutter recommendations, but you can provide three different scenarios for your projections: conservative, realistic, optimistic. For conservative scenarios, tone your projection down to the minimal outcome; the realistic scenario is where you want to land; the optimistic one is the best-case scenario.
Another way to project traffic is by launching an optimization at small scale and extrapolate the results to all pages of the same type. Say you propose adding content to category pages. Instead of adding content to all category pages, roll it out on a few, first, and measure the results. Then, you can apply the relative traffic increase to its potential to all category pages.
I personally compare the page types I’m optimizing side-by-side with the top competitor and look for subtle differences in content, user experience, and links. I also pay attention to user intent, one of the fuzziest aspects of SEO. Then, abstract these differences to the page type I’m looking at, to develop tactics and strategies.
Make sure to explain your rationale behind each projection in written form and share it internally. That will allow other stakeholders to add comments, questions, or critique the approach to improve your projections better over time and create buy-in.
When you launch your optimizations, it’s crucial to closely monitor the impact and report it back to all involved stakeholders to create buy-in for the future. You will also understand the accuracy of your projections better and refine your models. This is one of the most powerful things you can do to update your mental models and understanding of SEO.
Before launching, measure baseline traffic (don’t forget Year-over-Year data to factor in seasonality). Optimally, forecast baseline traffic across the year to show how much you’d get without any incremental traffic. The baseline is flat traffic from the previous year, but if you project it across every month of the year, you can measure it against actuals and forecast your total incremental traffic.
SEO revenue formulas for different business models
Now, that the rationale and step-by-step traffic projection process are covered, let’s look at the connection between conversions and revenue. In this section, I provide formulas you can apply to your business model to calculate SEO revenue.
The reason I provide different formulas, instead of one is that there is no one-size-fits-all equation for SEO revenue. We need to customize our approach to the business model.
Another mistake is mistaking the CPC of a keyword with the value of its ranking in organic search. Repeat after me: the CPC of a keyword is not its value. The classic “just take the CPC sum of your keywords, and you have the ROI of SEO” is a fallacy. CPC and organic ROI are two different things.
Google Ads ranks results after the highest bidder. Organic search ranks results based on relevance and authority. Paid results deliver only value as long as you pay; organic results can deliver permanent value without marginal cost. Paid results can occur in many places on the SERP; organic results have a steady appearance (exception: SERP Features).
To find the true value of an organic ranking, we need to bridge the gap between traffic and the monetary value of a conversion. If you already know how to project and connect the two, this will be quick for you.
In the equations below, “traffic” is equal to “Keyword Search volume X CTR”. CTR is a value that you need to determine with your own click-curve and project to the position you want to achieve. For example, if you know that a keyword has 100 searches per month and you rank in position 5, your goal is to understand how many of the 100 searches you get and project how many you would get at position 3. That difference is what traffic projections are all about.
- The goal is to estimate how many incremental positions you can gain from an optimization for a set of keywords.
- Once you built your custom click-curve, as explained in step 4, you know your CTR per ranking position.
- Then, you multiply the expected CTR across all keywords with the keyword’s search volume.
A better alternative to Google’s Search Volume is Similarweb’s Search Visits, which I mentioned earlier. It gives you a much more accurate estimation of organic traffic for a list of keywords. Keep in mind that you still have to multiply such a metric with your custom click-curve.
E-commerce (Examples: Amazon, eBay, Costco, Target, etc.)
E-commerce businesses have the benefit that customers buy products right on the site, which simplifies revenue attribution. It can be as straightforward as going to GA and looking at Conversions > Multi-channel funnels > Assisted conversions and click on organic.
Alternatively, you can go to Conversions > Goals > Overview and segment goal completions by organic traffic and look at the goal value.
If you use a different analytics platform and can’t emulate the Google Analytics process I outlined here, use the following formula with your own data:
SEO revenue = traffic X avg. product page conversion rate X avg. product revenue margin
Marketplace (Examples: UBER, Doordash, Facebook, Airbnb, Spothero, Fundera)
Marketplaces in this context are not e-commerce marketplaces, but startups that connect offer with demand for services and other types of value.
SEO revenue is straightforward here:
SEO revenue = traffic X conversion rate X customer lifetime value (LTV)
B2C subscription or SaaS (Example: NY Times, Asana, Atlassian, Spotify, Shopify)
I split subscription businesses into B2C and B2B. The pricing structure between the two differs in so far that companies selling to companies typically have three different tiers that scale with their customers’ needs. Companies selling to consumers typically offer a single tier; sometimes, they charge annually instead of monthly.
The most simple way to calculate revenue from SEO is:
SEO revenue = traffic X conversion rate X average LTV
You can certainly make this more complex. I go with average customer lifetime value because taking upgrades from freemium to paid accounts or expanding into different tiers makes things much more complicated. The marginal information you get out of taking things to such complexity is often minimal.
Projecting SEO revenue for enterprise sales-driven companies is tricky because Marketing hands MQLs off to sales and doesn’t close the deal. So, they can’t control every step of the customer journey.
The formula here is:
SEO revenue = traffic X Visitor-to-MQL rate X MQL-to-close rate X average deal size
Once again, none of these equations are perfect. They’re starting points but need to evolve as your intelligence about SEO revenue improves over time.
Similarweb Industry benchmarks
To make revenue projection faster and simpler for you, Simlarweb was so kind as to provide exclusive conversion rate benchmarks for different industries.
The data comes from Similarweb’s Funnel Conversation module, which tracks checkouts on Thank You pages. This allows Similarweb to get a very accurate idea of how many people searched on Google, then clicked through a site and converted.
I analyzed four key metrics:
- average conversion rate per industry
- median conversation rate per industry (to account for outliers)
- the lowest and highest conversion rate in each industry
My goal was not just to get a realistic picture of conversion rates in several industries but to find outliers we can learn from.
Of course, you can and should plug the average conversion rates into the SEO revenue equation above.
As you can see from the data, the highest conversion rates are found in the fashion industry, followed by marketplaces and retailers. The biggest difference between average and median exists in the fashion retailers category, which means that a few outliers have considerably higher conversion rates.
SEO is not an exact science, but that can’t stop us from building basic models and iterating over time. When facing uncertainty, it’s best to start somewhere, learn fast, and move forward.
The biggest problem is often to get started, but you can make quick progress once you establish a model based on the tools I give you in this article.
As I mentioned earlier in the article, each equation can become much more sophisticated. You can split your projections up between different pricing tiers as SaaS business, for example. Complexity knows no limits.
Compare different revenue attribution models.
Fine-tune conversion rates by page type.
Project revenue years into the future.
There are lots of ways to make projections more sophisticated. Just get started.