SEO Forecasting: Predict Booked Calls, Not Traffic

Every SEO forecast you have been shown works the same way. Take search volume, multiply by a clickthrough-rate curve, hand you a traffic number. Ahrefs states the formula plainly in its own SEO forecasting guide: 'Search volume x average click-through rate (CTR).' It then adds the warning most agencies leave off the slide — 'you also run into the issue of using generic CTR curve models which may not be representative of your site.'

That forecast was built for an in-house marketer who needs a CFO to sign off. You are the CFO. You do not buy traffic. You buy booked calls, and you pay for them out of the same account that pays for trucks and payroll.

So here is the chain that actually gets you to a call, every haircut it takes on the way, and the honest reason most keywords do not survive it.

Why is a traffic forecast the wrong output?

A traffic forecast is the wrong output because traffic has no price and no denominator. '4,000 visits a month by month twelve' does not tell you whether that is 4 booked calls or 40, and at a $2,500/mo retainer the difference between those two numbers is the difference between a great channel and a fired agency.

The trap is that a traffic number is unfalsifiable in a useful way. An agency can hit the traffic forecast and still deliver nothing you can bank, because the traffic came from a blog post about 'how a heat pump works' read by people who already have one.

A forecast has one legitimate job: to decide, before money is committed, whether a keyword is worth targeting. Not to sell you a retainer. If a forecast has never killed a keyword, it is a sales asset wearing a spreadsheet.

What is the actual chain from search volume to a booked call?

There are six multiplications and one division between a search-volume number and a booked call, and every one of them shrinks the answer. Skip any of them and your forecast is out by an order of magnitude, not a rounding error.

Most published forecasts run the first three steps and stop. The last four are where a service business either makes money or does not, and they are the four steps only you have the data for.

How much of the volume survives the AI Overview and the map pack?

Far more than the headlines suggest — if your keywords have commercial intent. Ahrefs analyzed 146 million SERPs and found AI Overviews trigger on just 4.3% of commercial-intent and 2.1% of transactional-intent keywords, versus 21.4% of informational keywords (Ahrefs, September 2025 data). Local searches trigger one only 7.9% of the time.

Read that carefully, because it cuts both ways. 'Emergency plumber Tulsa' rarely gets an AI Overview. 'Why is my water heater leaking' very often does — question queries trigger an AI Overview 57.9% of the time in the same dataset.

Where an AI Overview does appear, the haircut is severe. Ahrefs compared 300,000 keywords in Google Search Console and found position 1 loses 58% of its clicks when an AI Overview is present, position 3 loses 46.4%, and position 10 loses 19.4% (Ahrefs, December 2025 data). Pew Research Center tracked 900 US adults across 68,879 real Google searches and found that when an AI summary appeared, users clicked a search result on just 8% of visits — versus 15% when no AI summary was present (Pew Research Center, March 2025 data).

So the forecast rule is not 'apply a 30% AI haircut to everything.' It is: check whether your specific keyword actually triggers an AI Overview, and apply the penalty only to the ones that do. Half the keywords in a service-business plan need no haircut at all. The informational ones need a large one.

What CTR should you assume at each position?

Assume a range, never a point — because the two most-cited CTR figures for position 1 differ by a factor of ten, and both are real. The curve everyone pastes into forecasts (39.8% for position 1, 18.7% for position 2) comes from First Page Sage, which describes itself on the page as 'a meta-analysis, combining research on click-through rates' — it is an aggregation of other people's older numbers, not a primary dataset.

The measured reality from real Search Console data is far lower. In Ahrefs' 300,000-keyword study, the average position-1 CTR on informational keywords was 3.9% in December 2025, down from 7.6% in December 2023. On informational keywords where an AI Overview appeared, position 1 got 1.6%.

Nobody has published a large-sample, current CTR curve for commercial-intent local service queries. That is the number your forecast needs, and it does not exist. So do not fake it:

• If you have Search Console history, build your own curve from your own data. This is the only trustworthy input, and it takes an afternoon.
• If you are a fresh domain with no history, forecast a band — a pessimistic case and an optimistic case — and make the go/no-go decision on the pessimistic one.
• If the keyword only clears your bar in the optimistic case, it does not clear your bar.
• Never accept a single-point CTR assumption in an agency deck without asking which dataset it came from and what year.

How do you convert clicks into leads and leads into revenue?

With your numbers, not benchmarks — and if you do not know your lead-to-close rate and average job value, stop forecasting and go get them, because they swing the answer more than anything Google does. A 20% close rate on $9,000 roofs and a 20% close rate on $400 drain cleans produce forecasts that are not in the same universe.

The one benchmark worth knowing is how badly form benchmarks lie. Zuko's form database, covering over 93 million tracked sessions, puts the average form completion rate at 51.71% (Zuko, 2025 data). That is the share of sessions that started a form and finished it — not the share of visitors who became leads. Substituting the first for the second inflates a forecast by roughly a factor of ten. Agencies do this constantly, usually by accident.

One more input belongs in the chain and never makes it in: response time. In a 2011 Harvard Business Review study, firms that contacted an online lead within an hour were nearly seven times as likely to qualify that lead as firms that waited just one hour longer (HBR, 2011). If your front desk returns web leads the next morning, the forecast you signed off on already assumed a lead-to-close rate you are not going to hit. That is an ops problem, not an SEO problem, and no ranking will fix it.

This is the forward-looking twin of measuring SEO ROI after the fact. Same chain, opposite direction. If you cannot measure it backwards, you cannot forecast it forwards.

How long until a fresh domain reaches the position you assumed?

Longer than any forecast assumes, and the base rate is brutal: Ahrefs tracked 1 million URLs first seen in September 2023 and found only 1.74% of newly published pages ranked in Google's top 10 within a year (Ahrefs, published May 2025). Filtering to non-empty English content across 2 million URLs, Ahrefs says 6.11% made it — which they call the fairer number to use.

The same study found 72.9% of pages sitting in Google's top 10 are more than three years old, and the average #1-ranking page is 5 years old. It also found that when a page did break into the top 10, 40.82% of the time it happened within the first month — so if a page is nowhere after six months, the odds of it climbing later are poor without a rewrite.

For a forecast, that means one thing: divide. If your keyword needs position 3 and your realistic path to position 3 is ten months, then a 'year one' revenue figure is really two or three months of revenue, not twelve. An agency that projects month-twelve revenue and quietly bills it as an annual number is off by 4-6x. This is the single most common lie in an SEO pitch deck, and it is why how long SEO takes is a forecasting input, not a FAQ.

Why do most keywords fail the forecast — and why is that the point?

Because the chain multiplies six fractions together, and six fractions multiplied is a very small number. That is not a bug in the method. That is the method telling you the truth that a traffic chart was hiding.

Run it yourself with your own inputs. Take a 1,900/mo head term where you would realistically land at position 5 in year two. Apply a CTR band, apply your form conversion rate, apply your close rate, apply your job value, then divide the annual figure by the months you actually held that position. Now run the identical chain on a 90/mo bottom-funnel term — the one with a city name and a service in it — where you could plausibly hold position 2 by month five.

The small keyword usually wins, and it usually is not close. Higher intent, higher conversion, no AI Overview, and it starts earning eight months sooner. This is why we plan around precise long-tail terms rather than the head terms with the impressive volume, and why what SEO actually costs only makes sense next to what the keywords can actually return.

A forecast that approves everything is broken. The forecast is a filter. Its output is a shorter keyword list, and a shorter list is the deliverable.

What forecast should you refuse to accept from an agency?

Refuse any forecast that is a single number with no band, no denominator, and no time discount — and refuse any forecast that comes with a ranking guarantee attached, because nobody can guarantee a ranking and anyone who does is telling you something they cannot know. Ahrefs, whose data most of these decks are built from, calls forecasts 'educated guesses based on data.' That is the correct posture.

The verdict: the only forecast worth signing off on is one where the pessimistic case still returns more than it costs, the ramp is shown month by month, and the keyword list got shorter after the math was run. We publish our pricing so you can put a real cost on the left side of that equation before anyone builds you a deck, we work month-to-month with no minimum, and we cut a channel that produces no qualified leads in 90 days rather than defend it with a traffic chart.

If you want the chain run on your actual keywords, your actual close rate, and your actual job value — including the keywords we would tell you to drop — start with our pricing and then Get my free audit. You will get the forecast that kills keywords, not the one that sells retainers.