Size the opportunity before you build

Content has a real cost and an uncertain return, which is the combination that justifies doing sums first. A forecast does not have to be right to be useful. It has to be honest about its own inputs, and it has to let you compare one cluster against another without arguing about taste.

What follows is the arithmetic, the four numbers it rests on, and the two habits that make most forecasts worthless.

Forecast a range, then plan around the middle

Four estimates multiplied together do not produce a fact. Volume is modeled, click curves are averaged across results pages that share nothing, the position you will reach is a bet about competitors, and your conversion rate is a past figure applied to a future page.

Reporting one number hides all of that. Report three cases instead: a pessimistic one at a modest position with your weakest recent conversion rate, an optimistic one at a strong position with your best, and a middle case built from the position you actually expect.

The gap between the low and high cases is not an embarrassment. It answers the question that decides the plan: how wrong do the assumptions have to be before this stops being worth a quarter of the roadmap.

Size the cluster, not the keyword

A page collects searches in dozens, so forecasting a single term forecasts a fraction of what the page will actually earn from. Group the keywords first, then forecast the group. Clustering gives you the unit that matches the thing you are about to pay for.

Four numbers per cluster:

1. Total monthly volume for everything in the group, which caps what the page could collect.
2. The difficulty of what currently ranks, which decides how long the work takes and whether it finishes inside a horizon you care about.
3. The position you can realistically reach, read off the pages already there rather than chosen to make the case work.
4. What a conversion is worth to you, from your own numbers.

Then multiply: volume, by the click rate for your expected position, by your conversion rate, by the value of a conversion. Run it at three positions and you have your three cases.

Number two carries more weight than most plans allow it. Difficulty is the only one of the four that speaks about time, and time is what kills content cases. A large keyword behind three established domains is a three-year project described as a quarterly one. A modest keyword behind a forum thread and a thin affiliate page is a quarter's work with the same ceiling on a shorter fuse. Both may be worth doing. Only one of them is worth promising.

Volume sorts a keyword list badly

Order a list by search volume and you have ordered it by how many people are curious, with no reference to whether you can win or what a visit is worth. It is a sort that reliably promotes definitional terms answered a thousand times over, and reliably buries the terms typed by someone about to spend money.

It also has a legitimate job. Within one intent and one difficulty band, volume is a fair way to break ties. The error is promoting it to the primary sort across a whole list, where it answers a question nobody asked.

A forecast is what corrects this, because difficulty and value both point away from volume. If your forecast's ranking always matches the volume ranking, the forecast is decoration.

Convert sessions into the currency the meeting uses

Sessions are the number SEO reports and the number nobody else is measured on, so they get discounted on arrival. The same case carried through conversion rate and value per conversion lands as a figure that competes with everything else asking for the budget.

Two things make the translation credible. Name the rate you applied and where it came from. And subtract the cost of producing the page, plus whatever links it needs, because payback is a comparison against cost rather than a revenue figure standing alone.

What you are left with is a payback window. This cluster covers its production cost in this many months in the middle case, and roughly twice that in the pessimistic one. That sentence survives a finance review. A projected session count does not.

Run the same math in reverse

The calculation also works backwards, and backwards is often where the better return sits. Take a page you already own, look up where it ranks, and apply the click rate for a position above it. The difference is demand the page is already adjacent to and not collecting.

This costs almost nothing to check, and it competes well against new work: the page is indexed, it has history, and the job is editing rather than building. The Search Console guide in this library locates them for you, and it is usually the fastest way to fill a quarter without commissioning anything.

Do it with an agent

Connect the Doverity MCP and the sizing reads live difficulty and volume rather than an export from last quarter. This prompt runs the whole case; the keyword research skill follows the same sequence.

Use the Doverity MCP. Run research_keywords on "[my topic]" and group
what comes back by the intent each keyword carries.

A. For every group, give me total monthly volume, the middle
difficulty score across its keywords, and the three domains holding
the strongest positions. Mark any group where all three are domains I
have no realistic path past.

B. Take the five groups with the best volume-to-difficulty ratio and
model sessions at positions 3, 5 and 8 on a standard click curve, so
each has a low, middle and high case.

C. My conversion rate is [rate] and each order is worth [value] to me.
Rank the groups by middle-case revenue divided by difficulty.

D. Against [cost] to produce and maintain one page, give me the
break-even month for each group in its middle case.

Step C is where the plan changes. Ordering by revenue against difficulty regularly lifts a group that volume had buried, and that reordering is the reason to do the forecast at all.

SEO forecasting FAQ

How reliable is an SEO forecast?

Reliable enough to rank options against each other, and not reliable enough to commit to a figure. The inputs are estimates, the click rates are averages, and the schedule depends on competitors who are also publishing. Hold the middle case loosely, watch what the spread implies about risk, and replace the assumptions with measured positions as soon as the page has been live long enough to produce them.

What counts as a good return on SEO?

Nothing universal. It depends on what a conversion is worth to you and how long you can wait, so the useful comparison is internal: the cost of the pages and links against the revenue they bring in over their life, measured against whatever else the same money could have done. SEO tends to look unimpressive early and stronger with time, which means the period you choose to measure over often decides the answer before the math does.

What should an SEO forecast report?

The two figures a business treats as real: leads and revenue. Rankings, sessions, and impressions explain movement in those two and make poor targets in themselves, since none of them can be spent. Lead with the money figures when asking for budget, and bring the diagnostics when you need to explain why the money figures moved.

What is the formula for keyword traffic potential?

Total volume for the group, times the click rate at the position you expect to hold, gives sessions. Sessions, times conversion rate, times value per conversion, gives revenue. Run it at several positions rather than one, because a single result built from four estimates carries an authority the estimates never had.