A niche article can take a full day to research, write, and review. Before committing that time, calculate what it would need to earn back its cost. Then check how much evidence supports the assumptions.
This worksheet helps a small AI or SaaS team evaluate low-volume keyword ROI through a bounded experiment. You’ll define the reader’s task, estimate the work, and test how changes in reach or conversion affect the decision. The result is a budget decision with visible uncertainty.
Separate unknown demand from a small estimate
Record the exact query, target country, language, provider, and metric update date before interpreting a volume number. Keep these states separate:
- A positive estimate means the provider reports some search activity.
- A reported zero means the provider estimates zero under its measurement method.
- A missing or unavailable value means demand is unknown.
A missing value gives you no basis for substituting the volume of a broader phrase. Nor should you add similar keyword estimates without checking whether they overlap.
Google says Keyword Planner’s average monthly searches include close variants and depend on the selected location, network, and date range. Its search-volume statistics are rounded. Its competition metric describes advertisers, so low paid competition does not establish easy organic rankings. See Google’s definitions of keyword metrics.
Keep the provider’s estimate in an evidence column and any number you choose for modeling in an assumption column. Label an unsupported input “assumption only” so a convenient guess cannot pass for a measurement.
Confirm the reader’s task before calculating value
Search the exact phrase and a few natural alternatives with US results in view. Read the leading results and record whether they offer tutorials, product comparisons, troubleshooting, definitions, or another kind of answer.
Then describe the reader’s task in one sentence. For example: “An operations manager needs to remove duplicate invoices before importing them into accounting software.”
Use that sentence to check whether your product supports the task, whether the reader could reasonably become a customer, and whether the answer needs an article at all. A label such as “high intent” leaves those questions unresolved.
Collect evidence from relevant customer conversations, support questions, or public discussions. Preserve the original wording separately from your interpretation. A repeated problem can justify a small content experiment, but it does not establish a search-volume estimate or a conversion rate. The FindVex guide to building a pain-point evidence sheet shows how to organize that evidence.
Check your existing pages, too. If one already answers most of the question, adding a worked example may be cheaper and more useful than creating another article. Evaluate the additional cost and benefit of that update.
Build a model with visible assumptions
Choose a fixed evaluation horizon, such as six months from publication. Include research, writing, technical review, implementation, distribution, and maintenance in the content cost. Assign founder time an explicit hourly value; otherwise, a labor-intensive article can appear nearly free. Use the related guide to including founder time in acquisition cost if that input needs work.
For a simple acquisition model, use these calculations:
| Output | Calculation |
|---|---|
| Search visits | Monthly searches × effective traffic months × captured visit share |
| Customers | Search visits × visit-to-paid conversion rate |
| Contribution | Customers × average contribution per customer within the evaluation horizon |
| Scenario ROI | (Contribution − content cost) ÷ content cost |
Captured visit share is the assumed fraction of searches that produce a visit to your page. It combines uncertainty about visibility and clicks. Search Console CTR uses a different denominator: your recorded impressions. See Google’s Performance report definitions.
Effective traffic months account for a slow start. Four effective months within a six-month horizon means you are modeling the equivalent of four months at the assumed traffic level. This simplification does not predict when rankings will appear.
Contribution per customer means revenue remaining after the variable costs you choose to include, such as inference, payment processing, and incremental support. Use a consistent definition. Count only contribution earned within the evaluation horizon, allowing for customers acquired late in the period. Full lifetime value does not belong in a six-month payback calculation.
If your business has a sales process, split conversion into visit-to-qualified-lead and qualified-lead-to-customer rates. Define what makes a lead qualified before counting it.
You can also work backward:
- Customers required to break even = content cost ÷ contribution per customer.
- Visits required = customers required ÷ visit-to-paid rate.
- Qualified leads required = customers required ÷ qualified-lead-to-customer rate.
Use positive, supported inputs for these thresholds and round required counts up to whole people. If conversion is unknown, stop at the customer threshold. If contribution per customer is also unknown, leave financial payback unresolved and choose a bounded learning budget. An empty cell is more useful than an invented rate.
Worked hypothetical example: an invoice software article
Suppose a small SaaS team is considering an article about finding duplicate invoices before import. All inputs below are invented to demonstrate the worksheet. They are not keyword measurements, benchmarks, or expected results.
The team sets a six-month horizon and budgets $600: eight hours at $60 per hour, plus $120 for review and updates. It assumes four effective traffic months and an average of $150 in contribution per acquired customer within the horizon.
It tests three combinations:
| Input | Lower case | Middle case | Higher case |
|---|---|---|---|
| Monthly searches | 40 | 80 | 120 |
| Captured visit share | 15% | 20% | 25% |
| Visit-to-paid rate | 2% | 3% | 4% |
The resulting arithmetic is:
| Output | Lower case | Middle case | Higher case |
|---|---|---|---|
| Search visits | 24 | 64 | 120 |
| Modeled customers | 0.48 | 1.92 | 4.8 |
| Contribution | $72 | $288 | $720 |
| Scenario ROI | −88% | −52% | 20% |
Fractional customers are model outputs; actual customers arrive in whole numbers. These cases are neither probability estimates nor confidence intervals. The higher case combines several favorable assumptions, so it deserves particular scrutiny.
At $150 contribution per customer, the $600 article needs four customers to break even. At a 3% visit-to-paid rate, the model requires 134 visits after rounding up, compared with 64 in the middle case. That is a conditional threshold, not a promise that 134 visits will produce four customers.
To isolate one assumption, hold everything else fixed. Doubling the middle-case conversion rate to 6% produces $576 in contribution, still below the $600 cost. Reducing the content budget to $288 would bring the middle case to break-even, but only if its reach, conversion, and contribution assumptions still held.
The team can now decide whether a smaller scope could answer the reader’s question, whether it needs better conversion evidence, or whether to defer the article.
An article may also help sales or support. Record those uses separately. Do not assign them invented dollar values to make the acquisition model positive, and do not count the same customer twice across channels.
Set a review rule before spending the full budget
Write down a spending limit, an observation window, and the decision you’ll make at review. For example, cap the first version at four hours, review discovery and traffic monthly, and reassess its economics at six months. These are planning choices, not universal SEO deadlines.
Google says changes can take hours to several months to appear in search results, and some changes produce no noticeable effect. The worksheet therefore needs room for uncertain timing. See the SEO Starter Guide’s timing guidance.
Use Search Console to inspect page and query performance, with the United States country filter when appropriate. Its Performance report provides clicks, impressions, CTR, and average position, with dimensions including pages, queries, and countries. See Google’s Performance report documentation.
Keep that search evidence alongside your own records of qualified leads, paid customers, and contribution. Tracked customer activity supports an attribution judgment; it does not prove the article caused every purchase.
Diagnose the stage before changing the article:
- Little visibility: inspect discovery, indexing, and query relevance before judging conversion.
- Visibility without visits: inspect the result presentation and whether the page matches the searcher’s task.
- Visits without qualified action: inspect audience fit, the answer, and the next step.
- Qualified activity without payback: revisit costs, customer economics, and elapsed time.
With small counts, report the counts alongside any rate. One customer from ten visits is too little evidence to confidently budget the next ten articles at a 10% conversion rate.
Copy this decision worksheet
Record the query, US audience, reader task, and search-result check date. Add the customer evidence and any existing page you could improve. Then copy this table and fill in only what you can support:
| Input | Evidence, source, and date | Lower assumption | Middle assumption | Higher assumption |
|---|---|---|---|---|
| Monthly searches | Include reported, zero, or unavailable status | ___ | ___ | ___ |
| Effective traffic months | State the evaluation horizon | ___ | ___ | ___ |
| Captured visit share | Explain the basis, or mark assumption only | ___ | ___ | ___ |
| Visit-to-paid rate | Record counts and measurement period, if available | ___ | ___ | ___ |
| Contribution per customer | Define included costs and earning period | ___ | ___ | ___ |
| Total content cost | Include founder hours and cash expenses | ___ | ___ | ___ |
For a sales-led business, replace the visit-to-paid row with separate visit-to-qualified-lead and qualified-lead-to-customer rows. Calculate visits, customers, contribution, and ROI for each case. Then change one input at a time to see which assumption changes the decision.
Finish with this decision record:
- Customers, visits, or qualified leads required to break even: ___
- Separate sales, support, or learning purpose: ___
- First-version spending cap: ___
- Review dates, owner, and tracking method: ___
- Evidence needed before spending more: ___
- Decision: draft, reduce scope, gather evidence, or defer: ___
Choose one task before commissioning the article
Pick one candidate keyword and complete the sheet. Identify the assumption that most changes your decision, then assign a task to investigate it before committing the full budget. If conversion determines whether the article pays back, review comparable traffic and customer records. If the answer is needed for support regardless of search demand, set a support budget and judge the work against that purpose.
End the exercise with a named owner, a spending cap, and a review date.
