SEO forecasting: a method you can defend, with ranges and a calculator
Most SEO forecasts multiply search volume by a click-through rate and assume the page ranks first. This is the method that survives a finance review: two kinds of forecast, rankings weighted by probability, a ramp, AI Overviews priced in, and a range instead of one number.
What is SEO forecasting?
SEO forecasting is estimating how much organic search traffic, and the conversions and revenue that come from it, a site will produce over a future period. A useful forecast states its assumptions, gives a range rather than a single number, and separates the traffic existing pages will keep producing from the extra traffic that planned work might add.
The calculator further down turns these rules into numbers you can change. If you need the payback and cost side of the same decision, the SEO ROI Calculator models revenue, churn and payback over the same timeline.
Why most SEO forecasts fail
Most SEO forecasts fail in one of five predictable ways, and every one of them makes the forecast too high. A finance team that has seen one inflated SEO forecast discounts the next one, so getting these right matters more than the precision of any single number.
| Failure | Why it inflates the forecast | What to do instead |
|---|---|---|
| Assuming position one | The worked examples in most guides multiply volume by the first result’s click-through rate | Weight each ranking band by its probability, using your own track record for similar pages |
| Trusting tool volume | Tool volumes are estimates, and Keyword Planner overstated Search Console impressions for 91.45% of keywords in Ahrefs’ study | Reduce tool volume by a stated percentage, or calibrate it against impressions for keywords you already rank for |
| Ignoring the ramp | Traffic is credited from the month a page is published | Spread each page’s traffic over a ramp, and put most new-page traffic in the second year |
| Ignoring AI Overviews | Click-through rates from before AI Overviews are applied to queries that now show them | Estimate the share of target keywords with an AI Overview and discount their click-through rate |
| One number | A single figure reads as a promise, and any miss reads as failure | Give low, expected and high cases, with the assumptions behind each |
The pattern is visible in the guides that rank for this keyword. Search Engine Land’s step-by-step projection, for example, is careful to call its ranking gains assumptions and recommends at least six months to a year as a timeframe, but the method itself has no probability of ranking and no ramp: it is a calculation of what happens if the ranking gains arrive. That is a useful ceiling. It is not a forecast.
Two forecasts, not one: baseline and opportunity
A defensible SEO forecast is two forecasts added together. The baseline forecast predicts what the pages you already have will produce if you change nothing. The opportunity forecast predicts the extra traffic from work you have not done yet: new pages, rewrites, and ranking improvements. They use different data and different methods, and keeping them apart is what lets someone check the forecast later.
- Built from your own Search Console history, so it starts from real clicks
- Can be tested by forecasting months you already have and comparing
- Captures your own seasonality and recent trend
- Cannot see the effect of work you have not done
- Assumes the recent past continues, so it misses algorithm updates and new competitors
- Covers new pages, rewrites and target keywords you do not rank for yet
- Makes the business case for new work explicit
- Rests on tool volumes, assumed click-through rates and an assumed chance of ranking
- Cannot be tested until the work ships, so it needs a wide range
The split also answers the question finance actually asks, which is how much of the forecast depends on the money being requested. If the baseline already covers last year’s traffic, the opportunity forecast is the return on the new budget, and that is the number to compare with its cost.
How to forecast the baseline from Search Console
The baseline forecast starts from your own monthly organic clicks, because that is the one series you can trust to reflect your site. The goal is not a sophisticated model. It is a method you can explain in a sentence and have tested against months you already know.
Get enough history
You need at least two full years of monthly data to see seasonality, and Search Console’s Performance report holds only 16 months, a limit it has had since 2018. Two fixes: set up the Search Console bulk data export to BigQuery, which sends a daily copy of your data into your own BigQuery project, where it stays for as long as you keep it, and in the meantime export the Performance report each month to a sheet. If you have less than two years, say so in the forecast and widen the range.
Remember what the series leaves out. Ahrefs found that anonymised queries, which Google withholds from query-level reports because few people searched them, accounted for 46.77% of Search Console clicks across roughly 887,000 properties in April 2025. Forecast total clicks by page or for the whole property, not the sum of the queries you can see, or the baseline will start about half too low.
Start with the seasonal naive benchmark
The simplest serious baseline is the seasonal naive forecast: next March will look like last March, adjusted for the trend since. It sounds too simple to use, but the evidence on forecasting methods is that simple benchmarks are hard to beat. In the M4 forecasting competition, twelve of the seventeen most accurate methods were combinations of mostly statistical methods, and none of the six pure machine learning methods beat the combination benchmark. Use the simple method first, and adopt anything more complex only if it beats the simple one on your own held-back months.
- 1ExportMonthly organic clicks for every complete month available, for the whole property.
- 2CleanMark months broken by tracking changes, migrations or outages, and replace them with the average of the same month in other years.
- 3IndexFor each calendar month, divide its clicks by the average month in that year, then average across years. That is the seasonal index.
- 4TrendCompare the last twelve months with the twelve before. That ratio is the trend you carry forward, capped at a level you can defend.
- 5ForecastAverage monthly clicks over the last twelve months, multiplied by the trend, multiplied by each month’s seasonal index.
- 6TestRepeat the whole method using data that stops a year earlier, forecast the year you held back, and compare.
Test the method before you trust it
Testing means pretending it is a year ago, building the forecast with only the data you would have had then, and comparing it with what actually happened. The International Journal of Forecasting review of forecasting theory and practice makes the same two points: it is extremely important to evaluate a forecasting method, and prediction intervals are what let you judge a point forecast. Record the average percentage miss across the twelve held-back months. That miss is the honest width of your baseline range.
If you want a statistical model rather than a spreadsheet method, Prophet, published by Taylor and Letham, was built for forecasting business series at scale, with interpretable parameters that analysts can adjust by hand. It is a reasonable next step. The same rule applies: keep it only if it beats the seasonal naive forecast on your held-back year.
How to forecast new pages and target keywords
The opportunity forecast covers traffic you do not have yet, so it has to be built from keywords. For each keyword or group, expected monthly clicks are its realistic search volume, multiplied by the probability of reaching each ranking band, multiplied by the click-through rate in that band, discounted for AI Overviews and spread over a ramp.
Realistic volume
Start from the volume your keyword tool reports, then reduce it. Keyword tools estimate volume from different sources and models, and the estimates run high: in Ahrefs’ study of 72,635 keywords, 91.45% of Google Keyword Planner volumes were above the Search Console impressions those keywords actually produced, and 54.28% were drastic overestimates. The best correction is your own: for keywords where you already rank on page one, compare the tool’s volume with your Search Console impressions, and apply that ratio to the new keywords. Without that data, state the reduction you used.
The chance of ranking
This is the input most forecasts leave out, and it matters most. New pages rarely reach the top ten quickly. Ahrefs’ 2026 update of its ranking study found that only 1.74% of 1 million random URLs first seen in September 2023 reached the top 10 within a year, and that 72.9% of pages in the top ten are more than three years old. Those are averages across the web, including pages nobody tried to rank, so a site with authority in its topic will do better. The point is that the probability is an input to estimate, not 100%.
Estimate it from your own record. Take the last ten pages you published for keywords of similar difficulty and count how many reached positions 1 to 3 and 4 to 10 within twelve months. Seven out of ten on page one and two in the top three gives you 50% for positions 4 to 10 and 20% for positions 1 to 3. If you have no record, use conservative figures and say so.
Click-through rate by band
Use your own click-through rates first. In Search Console’s Performance report, filter to pages of the same type as the ones you plan, and read the average click-through rate for queries where the average position is between 1 and 3, then between 4 and 10. Your rates reflect your titles, your brand and your query mix, which a published curve cannot.
If you have no history, published studies give a starting point, with caveats. Backlinko’s study of about 4 million results, using Search Console data from Semrush, put the first result’s average click-through rate at 27.6%. The Advanced Web Ranking CTR study is updated monthly from anonymised Search Console data and lets you filter by device and intent. Both measure averages across very different queries, so treat them as a ceiling for an unknown site rather than a prediction.
The AI Overview adjustment
Where Google shows an AI Overview, fewer people click a result. Ahrefs’ February 2026 update, comparing desktop click-through rates for 300,000 keywords in December 2023 and December 2025, estimated that an AI Overview correlates with a 58% lower average click-through rate for the page ranking first, with smaller effects lower on the page. Pew Research Center tracked 900 US adults in March 2025 and found that users clicked a traditional result in 8% of visits when an AI summary appeared, against 15% when one did not. The Advanced Web Ranking study puts the first result’s loss on the order of half.
So estimate what share of your target keywords show an AI Overview today, which you can see in most rank trackers or by checking the results yourself, and discount their click-through rate. A 40% to 60% reduction for those keywords is consistent with the studies above. In the Advanced Web Ranking data, informational and commercial queries lose the most clicks at the first position, so a forecast built on how-to, definition and comparison queries needs the larger discount.
| Input | Best source | Fallback if you have no data |
|---|---|---|
| Search volume | Tool volume calibrated against your own Search Console impressions | Tool volume reduced by a stated percentage |
| Chance of ranking | Your own record for similar pages and difficulty | Conservative estimates, stated as assumptions |
| Click-through rate | Your own Search Console CTR by position band | A published curve, treated as a ceiling |
| AI Overview share | Rank tracker data or a manual check of the results | Assume most informational queries show one |
| AI Overview discount | Your own CTR on keywords with and without one | 40% to 60%, from the studies cited above |
| Ramp | How long your past pages took to reach steady traffic | 12 months or longer for a new page |
Ramp: when the traffic arrives
A forecast that is right about the total and wrong about the timing still fails, because the budget review happens on a calendar. Most traffic from a new page arrives well after it is published, so spread each page’s mature traffic over a ramp instead of counting it from the first month.
The ranking study above is the reason. With 72.9% of top-ten pages more than three years old, a page published this quarter is competing with pages that have had years to collect links and engagement. On an established site the ramp is shorter; on a new site, or in a competitive space, it can run past two years.
The simplest way to apply a ramp is a straight line: traffic rises evenly from zero to the mature level over the ramp period, then holds. With a nine-month ramp, a page’s first year delivers about 62% of a full year at maturity. With an eighteen-month ramp, it delivers about a third. The calculator below uses that straight-line rule, which is the right level of precision for a decision about budget.
SEO forecast calculator
The calculator turns the opportunity formula into numbers. Enter the total monthly volume for the keywords a page or project targets, your realistic reduction, the chance of reaching each band and your click-through rates. It shows expected clicks at maturity, the range between settling on page one and reaching the top three everywhere, and what the first year delivers on a straight-line ramp.
Hypothetical defaults. Replace every input with your own data; the click-through rates especially should come from your Search Console.
Two checks keep the calculator honest. First, if the expected case is close to the upper reference, your probabilities are probably too high: reaching the top three for a whole keyword group is rare. Second, compare the expected clicks with what your best existing page of the same type earns. A new page forecast to beat your best page needs a reason in writing.
For a whole project, run the calculator for each page or keyword group and add the results, then add the baseline. Group keywords that one page will target together, using the same search intent, so you do not count two pages for the same query.
From traffic to conversions and revenue
Traffic forecasts only matter to a budget holder once they become conversions and money. Multiply forecast clicks by the conversion rate for the same type of page, then by the value of a conversion, and present the result with the same range as the traffic.
The forecast is the return side of a decision; the cost side is the budget. How to build that budget from the pages and fixes the plan needs, rather than from a percentage, is covered in the guide to setting an SEO budget.
Three rules prevent the usual overstatement. Use the conversion rate of comparable pages, not the site average: informational guides convert far below pricing and comparison pages, and a forecast built mostly on guides at the site’s average rate will be several times too high. Use gross profit rather than revenue where you can, because that is what pays back the cost. And account for the sales cycle: in B2B, a click this quarter may become revenue two or three quarters later, which is why the payback period, not the monthly figure, is what finance will judge.
| Line | Low case | Expected case | High case |
|---|---|---|---|
| Baseline clicks, next 12 months | 84,000 | 92,000 | 99,000 |
| Opportunity clicks, first 12 months | 3,000 | 9,000 | 18,000 |
| Total organic clicks | 87,000 | 101,000 | 117,000 |
| Conversions at 1.5% on opportunity pages | 45 | 135 | 270 |
| Value at $400 a conversion | $18,000 | $54,000 | $108,000 |
The SEO ROI Calculator takes the next step: it models the same ramp against monthly cost, adds churn for recurring revenue, and shows when the cumulative return overtakes the cumulative spend. For B2B SaaS, where a customer is worth several years of subscription, the economics are set out in B2B SaaS SEO: the economics, and the pages that carry it.
How to present an SEO forecast so it is trusted
A forecast is trusted when the reader can see what it rests on and what would change it. The format below fits on one page and answers the questions a finance or leadership review will ask.
- Three cases. Low, expected and high, for traffic, conversions and value, with the expected case as the plan.
- Baseline and opportunity on separate lines. So the reader can see how much depends on the new budget.
- An assumptions register. Every input, its value, its source and who owns it: volume reduction, ranking chances, click-through rates, AI Overview share, ramp, conversion rate and value.
- The test result. How far the baseline method missed when it forecast a year you already had.
- What would change it. The two or three assumptions the result is most sensitive to, and how you will know early if they are wrong.
- A review date. When forecast and actuals will next be compared, and what triggers a re-forecast.
| Assumption | Value used | Source | Owner |
|---|---|---|---|
| Keyword tool volume reduction | 30% | Tool volume against Search Console impressions for 40 keywords we rank for | SEO lead |
| Chance of positions 1 to 3 | 20% | Last ten guides on keywords of similar difficulty | SEO lead |
| CTR in positions 4 to 10 | 3% | Search Console, guide pages, last six months | SEO lead |
| Share with AI Overviews | 40% | Rank tracker, target keywords, this month | SEO lead |
| Visit to conversion rate | 1.5% | Analytics, guide pages, last twelve months | Marketing ops |
| Value per conversion | $400 | Finance, average gross profit per converted lead | Finance |
Write the forecast in the language of the business, not of SEO. ‘About 135 extra demo requests in the first year, between 45 and 270’ is a statement a budget holder can act on. ‘Visibility will improve by 40%’ is not.
Tracking the forecast against actuals
A forecast earns its value after it is published, when it is compared with what happened. Compare forecast and actual clicks monthly, by baseline and by opportunity page, and look at the gap in percentage terms.
Set two rules in advance. First, a re-forecast trigger: if actuals fall outside the low or high case for two consecutive months, rebuild the forecast with the new information rather than explaining the gap away. Second, a learning rule: each quarter, update the inputs that were wrong, especially the chance of ranking and the ramp, using the pages that have now had time to mature. After a year, your own inputs replace most of the published figures in this guide, and the forecasts get narrower.
When you need to know whether a specific change caused a change in traffic, a forecast comparison is not enough, because many things move at once. The causal impact method, a Bayesian structural time-series approach published by Brodersen and colleagues in 2015, builds a counterfactual from related series that were not affected, and is the standard tool for that question. The broader problem of measuring organic search when attribution undercounts it is covered in SEO measurement when attribution is already broken.
Forecast traffic is one input to the metrics that run the business. Which few of them belong on the front page, and how they constrain each other, is the subject of the guide to marketing metrics worth tracking and reporting.
Common SEO forecasting mistakes, and how to spot them
The mistakes below each push a forecast in a predictable direction. Most push it up, which is why an SEO forecast that has not been checked for them is usually too optimistic.
- Every keyword at position one. Check: is there a probability of ranking anywhere in the model?
- Tool volume taken at face value. Check: was it compared with Search Console impressions for keywords you already rank for?
- Published click-through curves for a site with its own data. Check: are the rates from your Search Console?
- No AI Overview discount. Check: what share of the target keywords show one today?
- Traffic from month one. Check: is there a ramp, and how long is it?
- Two pages counted for one query. Check: is each keyword assigned to exactly one page?
- Site-average conversion rate on informational pages. Check: does the rate come from the same page type?
- The baseline never tested. Check: what was the miss on a held-back year?
- One number. Check: are there low and high cases, and is the target the expected case?
Duplicate targeting is worth a second look, because it inflates the forecast and then splits the rankings when the pages go live. If two planned pages share most of their target keywords, the fix is in the guide to keyword cannibalization.
Forecast the pages you already have and the pages you plan to build separately, because they need different methods. For existing pages, start from your own Search Console history with a seasonal baseline and test the method on months you held back. For new pages, weight each ranking by how likely it is, apply your own click-through rates, discount for AI Overviews, and spread the result over a ramp of a year or more. Present a range with every assumption written down, and compare it with actuals every month.
When I review a forecast, the first question is what share of it comes from pages that do not exist yet. If that share is most of the number, the forecast is a bet on execution, and the range should say so.
The second is whether the method has ever been tested. A team that can show its baseline missed by 8% on last year is far more credible than one with a more sophisticated model it has never checked.
Frequently asked
What is SEO forecasting?
How do you forecast SEO traffic?
How accurate are SEO forecasts?
What click-through rate should I use in an SEO forecast?
How do AI Overviews affect SEO forecasts?
How long does it take for SEO to show results in a forecast?
Should an SEO forecast be one number or a range?
What tools do you need for SEO forecasting?
I am not taking on new work right now, but I am always glad to talk ideas. If this raised a question about your own situation, send it over.
Say hello