BROKEN INTERNET CLUB / FIELD GUIDE
Look up a term when you need it. The six-step method lives on the Range Model page.
See the six stepsRange Model: Baseline
A reasoned estimate of what may happen, built from evidence, assumptions and delivery conditions.
Why it matters: It should help someone decide what to do next without pretending the future is settled.
Illustrative example: The fictional shoe shop forecasts additional sales over the next 12 months if a defined piece of SEO work happens.
Do not confuse it with: A promise, guarantee or target.
Range Model: Baseline
A target states what the business wants. A forecast estimates what current evidence and conditions support.
Why it matters: Putting the two side by side reveals the gap that needs a decision.
Illustrative example: The target is 250 additional sales. The best-supported forecast is 140, leaving a 110-sale gap.
Range Model: Baseline
The estimate for the same period if the proposed SEO work does not happen.
Why it matters: Incremental impact must sit above what would probably happen anyway, not above zero.
Illustrative example: The shoe shop estimates its 12-month visibility, clicks and sales without the new work before adding any uplift.
Range Model: Baseline
The period covered by the forecast, such as the next 12 months.
Why it matters: Targets, scenarios, delivery timing and actuals need the same time boundary.
Illustrative example: The 140-sales case and the delayed 110-sales revision both refer to the same 12-month period.
Range Model: Baseline
A baseline that allows for recurring patterns rather than freezing today in place.
Why it matters: Doing nothing does not mean traffic stays flat.
Illustrative example: Use comparable past periods, adjusted for known changes, rather than repeating the latest month twelve times.
Range Model: Group
Queries or pages that behave similarly enough to share assumptions.
Why it matters: One whole-site CTR hides real differences, while thousands of keyword forecasts create noise.
Illustrative example: Commercial road-shoe queries stay together. Brand searches, guides and a shoe-size tool use separate evidence.
Range Model: Group
Click-through rate: clicks divided by impressions.
Why it matters: CTR describes how much visibility becomes traffic, so the query group and result environment matter.
Illustrative example: 100,000 additional impressions at 5.0% CTR produce 5,000 additional visits.
Do not confuse it with: Ranking position. The same rank can produce different CTRs.
Range Model: Group
Visibility means appearing. Traffic requires somebody to click.
Why it matters: AI answers and feature-heavy results can increase presence without producing the same click opportunity.
Illustrative example: A page keeps the same ranking but receives fewer clicks after the result becomes more crowded.
Range Model: Group
The share of visits that create the chosen outcome, such as a sale.
Why it matters: Use the finest reliable evidence. Query-level conversion often cannot be measured cleanly.
Illustrative example: The visibility and CTR model uses a commercial query group, then applies a reliable category-page conversion rate.
Deeper note: A hybrid model can use query or page groups for impressions and CTR, then a page or page-type conversion rate.
Range Model: Group
Prevent several groups or projects from claiming the same search opportunity.
Why it matters: One impression, click or sale cannot be incremental twice.
Illustrative example: A template project and an internal-linking project do not both claim every road-shoe query.
Range Model: Evidence
The data, test, benchmark or judgement behind an assumption, with its period and limits.
Why it matters: An assumption should not put on a fake moustache and return as a fact.
Illustrative example: The 5.0% CTR comes from comparable non-brand category queries in Search Console.
Range Model: Evidence
Evidence from the organisation's own search, analytics, conversion or delivery history.
Why it matters: It is usually more relevant than a broad benchmark when definitions and samples are stable.
Illustrative example: Use the retailer's own category-page CTR history before borrowing an industry curve.
Range Model: Evidence
An outside benchmark used provisionally when first-party evidence is thin.
Why it matters: Labelling borrowed evidence stops it being mistaken for an observation about this site.
Illustrative example: An external CTR study is recorded as a benchmark and given a wider range until local data improves.
Range Model: Evidence
How directly, recently and reliably an assumption is supported.
Why it matters: Stronger evidence can justify a tighter range. Confidence in the meeting cannot.
Illustrative example: A recent test on comparable pages supports a tighter CTR range than an old industry average.
Do not confuse it with: Delivery certainty.
Range Model: Evidence
A defined presence in AI answers, such as a mention, citation, linked citation or appearance in a monitored answer set.
Why it matters: The visibility measure must be explicit before it can be forecast or reviewed.
Illustrative example: Forecast linked citations across 50 defined commercial prompts, then measure the result each month.
Do not confuse it with: Revenue. A mention does not automatically have a pound value.
Range Model: Evidence
A measurable AI referral visit can feed a visits × conversion × value model. Non-click visibility should remain separate unless evidence supports the commercial link.
Why it matters: This avoids invented attribution while still allowing measurable visits to be valued.
Illustrative example: Track linked AI visits directly. For unclicked mentions, watch later branded search or assisted behaviour without claiming causation.
Range Model: Scenarios
The overall spread of believable outcomes.
Why it matters: It keeps uncertainty visible without pretending a hand-built range is a statistical confidence interval.
Illustrative example: The shoe-shop range runs from 100 to 180 additional sales.
Range Model: Scenarios
A named case inside the range, built from assumptions that make sense together.
Why it matters: Scenarios explain what has to be true rather than decorating one number with plus or minus 20%.
Illustrative example: Low, Best-supported and Upside each combine evidence, response and delivery assumptions.
Do not confuse it with: A probability or confidence interval.
Range Model: Scenarios
The case most supported by the evidence and assumptions available today.
Why it matters: It gives the decision a useful centre without claiming more certainty than the evidence earns.
Illustrative example: 100,000 additional impressions × 5.0% CTR × 2.8% conversion produces about 140 additional sales.
Do not confuse it with: A promise or automatically the statistically most likely outcome.
Range Model: Scenarios
Change one assumption at a time and observe how much the result moves.
Why it matters: The assumption that can change the decision deserves the closest attention.
Illustrative example: Moving CTR from 5.0% to 4.57% changes the simplified result from about 140 sales to about 128.
Deeper note: Sensitivity is a test inside the six-step model, not a seventh step.
Range Model: Scenarios
The number of additional conversions needed for the forecast contribution to cover the project cost.
Why it matters: A result can look impressive until the cost test makes the decision visible.
Illustrative example: £8,000 divided by £60 contribution per sale is 133.33, rounded up to about 134 sales.
Range Model: Scenarios
The amount each additional outcome contributes after the agreed variable costs.
Why it matters: Contribution is often more useful than revenue when testing whether work is worth its cost.
Illustrative example: At £60 contribution per sale, 140 additional sales contribute about £8,400 before the project cost.
Do not confuse it with: Topline revenue per sale.
Range Model: Scenarios
A probabilistic range with a stated coverage claim, calibrated against comparable outcomes.
Why it matters: A hand-built decision range has not earned an 80% or 90% statistical label by default.
Illustrative example: An 80% prediction interval should contain roughly 80% of comparable future outcomes when properly calibrated.
Do not confuse it with: Low / Best-supported / Upside planning scenarios.
Range Model: Delivery
Something that must happen for the forecasted work or benefit to happen.
Why it matters: If it has to happen, it belongs in the forecast.
Illustrative example: Developer capacity, approval and a release slot determine whether the work can go live in March.
Range Model: Delivery
How strongly the current plan supports shipping the defined work in the planned window.
Why it matters: A ticket in the backlog and a confirmed release slot should not receive the same timing assumption.
Illustrative example: A named owner and confirmed March release provide stronger delivery evidence than an unassigned idea.
Do not confuse it with: Evidence that CTR or conversion will respond as forecast.
Range Model: Delivery
How much of the forecast period remains after delivery and how quickly the response builds.
Why it matters: The organisation's clock and search's clock are different.
Illustrative example: A June delay revises the same-period shoe-shop forecast from about 140 to about 110 sales.
Deeper note: The 110 is an illustrative revised forecast, not a universal rule that three months always removes 30 sales.
Range Model: Review
Compare the saved forecast with actual evidence, then update the forecast for the period that remains.
Why it matters: Changing the forecast when evidence changes is part of forecasting.
Illustrative example: Keep the original 140-sales case, record actual delivery and performance, then revise the remaining months.
Range Model: Review
Forecast a historical period using only information available beforehand, then reveal what happened.
Why it matters: It tests the method without the benefit of hindsight.
Illustrative example: Build a baseline for last year's 12-month period using only older data, then compare it with the actual result.
Deeper note: Backtesting is an advanced Review practice, not a competing core step.
Range Model: Review
Weighted Absolute Percentage Error: total absolute forecast error divided by total actual volume.
Why it matters: It answers how far away the forecast was overall across comparable periods.
Illustrative example: If absolute monthly errors total 12 sales and actual sales total 120, WAPE is 10%.
Do not confuse it with: An explanation of why the forecast missed.
Range Model: Review
A check for whether forecasts repeatedly land too high or too low.
Why it matters: A model can have an acceptable average error while leaning in one direction.
Illustrative example: Repeated over-forecasting points to optimistic baseline, response or delivery assumptions.
Range Model: Review
A set of genuinely comparable past interventions used to inform a new forecast.
Why it matters: Comparable work provides an outside view of delivery and response.
Illustrative example: Use earlier category-page changes with similar teams, markets and scope.
THE POINT
A range is not a confidence interval. Best-supported is not a promise. Sensitivity, backtesting and WAPE are useful practices inside the six-step method.
Build a decision range