Budget research, content and technical work by the decisions each team has to make. Give research enough time to establish the problem, reserve content effort for facts the business can approve, and price technical changes against the templates or systems they affect. Hold a small reserve until someone can say which of those three needs it and what the extra work will deliver.
A useful budget names the work, the person whose capacity it uses, what has to happen first, and the evidence that will close it. A single allowance for “AI optimisation” makes it hard to see whether the project is waiting for a product decision, an editor or a release slot. Separating the work also helps you redirect effort when the original estimate turns out to be wrong.
AI Visibility is published by Lunasol, which offers GEO and AEO services. The public examples were checked on 5 November 2026. The hours and business scenarios in this article are illustrative. Lunasol service overview
Pay research to resolve a decision
Start by stating what the team needs to learn before it can commission the next piece of work. A retailer might need to know which product sizes customers ask about, whether the answer exists in approved product information and which pages share the same template. Those questions can be investigated with a narrow sample. A catalogue-wide review may only make sense once the team knows whether the issue repeats.
Specify how many enquiries or pages will be examined, which team can confirm the facts and what the research must hand over. Ask for a short record of the question, evidence, uncertainty and recommended next action. The action can be to edit a page, repair a source, expand the sample or leave the page alone. Each disposition should be clear enough for someone else to take it forward.
A research allowance is easier to estimate if access and preparation are separate. An analyst can compare supplied records while the service owner checks what can be shared and confirms the meaning of abbreviations or product codes. If the project relies on internal records, reserve that owner’s time. A research plan that assumes immediate access can consume its allowance waiting for context that nobody was assigned to supply.
Set a stopping rule. For instance, a retailer could inspect a sample from one product family and stop once it has identified which issues are isolated, which recur and which remain unresolved. If the sample produces an open question about a shared template, fund a short investigation into that template before extending the research to the rest of the catalogue.
Reserve content effort for approved facts and buyer guidance
Content time covers interpreting the approved facts for a reader, drafting or revising the page, checking the draft and addressing review comments. Estimate those activities separately when the people involved have different schedules. A short edit may need more time from a product specialist than a writer if the business has yet to approve the underlying rule.
Decathlon UAE’s public page for the black MH100 walking boots shows a selected size, other size options and a size guide. Decathlon Middle East’s official Instagram reel from 29 October 2026 promotes staff guidance on choosing hiking shoes and points readers to its range online or in stores. These are different pieces of material. One presents product-level information and the other introduces a category decision. A retailer might budget for both if each serves a customer need. Decathlon UAE product page and Decathlon Middle East’s hiking shoe reel
For a comparable store, editing a category guide could involve describing the kinds of products a shopper can compare, then linking to the appropriate range. Updating individual product pages may require variant-specific facts and different approvals. Give the editor an approved source for any measurements or compatibility claims. Schedule a second review if multiple people must agree on the same terminology.
Also decide what format the buyer needs. A product table, a category explanation and an Instagram reel can each require different production and review time. Add translation, local review or media work only for the deliverables in scope. If an existing asset already helps the buyer make the intended decision, the content budget may only need to cover linking and maintenance.
Estimate technical work by what changes together
A technical estimate should identify the component that changes, the test cases and the release owner. If one page has an error, the work may be a single correction. If pages share a rule that maps product data into the wrong field, the fix may belong in the template or the data import. Estimate investigation, implementation, validation and release separately when the cause is still uncertain.
Google’s product variant documentation illustrates single-page and multipage setups. Those approaches have different URL and markup structures. Google also documents merchant listing eligibility for pages where shoppers can purchase, with product rich results focused on a product or its variants. Ask the developer which architecture your own site uses before budgeting a change. A category guide and a product page may need different treatment. Google product variant documentation and merchant listing guidance
Technical work might include adjusting how a selection maps to its product details, checking that affected variants display the intended facts and verifying the changed pages after release. For the fictional retailer, the team could check one common variant, one edge case and one product outside the target family to make sure the shared change behaves as intended. Those are example test cases. The implementation owner should choose the actual cases based on the site.
Keep the content and technical estimates connected when they depend on the same source. If a merchant has yet to approve which size label belongs to a variant, publishing and template verification both depend on that fact. Assign one owner to resolve it and allocate their effort once. The editor and developer can then work from the same decision record.
Build a small budget with a reserve
The example below allocates 60 hours for a fictional retailer improving one product family. The hours are planning assumptions. They do not represent a quote, an industry ratio or Decathlon’s budget.
| Workstream | Initial hours | Output that justifies the allocation |
|---|---|---|
| Research | 12 | Dated buyer question sample, product fact review and a decision on where issues arise |
| Content | 18 | Approved revisions to the selected guide or product copy, with reviewer time included |
| Technical work | 22 | Agreed template or data change, test cases and release verification |
| Unallocated reserve | 8 | Released against a named issue and an owner able to complete it |
| Total | 60 | One bounded project with a review after the first release |
These hours are a portfolio view. The project lead still needs to see which roles can do the work. If the developer has ten hours before the release window, an allowance of 22 technical hours cannot make all 22 available this week. Split the work into the tasks that can ship and the work that needs another slot.
Use the same unit within each planning view. Hours help compare capacity. If you also need a cash estimate, convert hours by the appropriate role and record fixed supplier charges separately. A reallocation of four writer hours to four developer hours can preserve the capacity total while changing the budget in dirhams. Confirm both before treating a proposed transfer as funded.
Release extra effort when the evidence changes
Suppose the research finds approved information is already available for most of the sampled items, but the fictional retailer’s shared template fails to display one field consistently. The team can reduce the writing allocation by four hours and transfer those hours to technical work. If it also releases four reserve hours for regression checks, the updated allocation is 12 hours for research, 14 for content, 30 for technical work and four held in reserve. The project still has a 60-hour ceiling.
The transfer depends on two decisions. First, the content owner confirms which writing can be dropped without leaving the buyer’s question unanswered. Second, the technical owner confirms the extra work and its test cases fit a release slot. If either decision is pending, leave the proposed transfer open in the budget. A moved figure is only useful when the receiving team can act on it.
The findings might point in the other direction. If the template already displays the intended fields, but the business lacks an approved explanation of how two ranges differ, move the next effort toward fact approval and writing. The technical team may still need time to validate the page after publication. Keep the required verification visible even if there is no template change to build.
Record the trigger, previous allocation, new allocation and reason. If the required work exceeds the reserve, bring the decision back to the project owner. Options include reducing the page set, sequencing work over another release or approving extra effort. State which buyer question would wait under each option so the choice has a commercial context.
Track time spent and time committed separately. Suppose research used nine of its twelve hours and needs the remaining three for a scheduled approval. Those three hours are spoken for even though they have yet to be used. If the approval is cancelled because the team receives a complete product record, the project lead can release the hours. Apply the same rule to reserved release and editorial review slots so the budget reflects the work the team can still finish.
Give recurring work its own allowance
A one-time correction can create a continuing responsibility. Someone may need to update product facts when a supplier changes a size label, review a category guide when the range changes or rerun a defined visibility check. Budget those tasks over the review period and name the person who owns the next cycle.
Avoid hiding that time inside the implementation estimate. Release verification answers whether the agreed change shipped. Recurring review answers whether the evidence and customer route still work after the business or the platform changes. A handover can assign the responsibility and preserve the checks, but the receiving team still needs time to perform them.
At the first review, compare planned and actual effort by workstream. Ask which assumptions were wrong and how that affects the next phase. If research keeps reopening because a source changes frequently, assign ongoing ownership for that source. If each release requires more testing than expected, update the technical estimate before adding more pages.
Use the budget to choose what to do next
For each line, keep the owner, quantity, definition of done and reason for the estimate. Revisit the allocation when research closes, when the scope of the technical fix is known and after the first release. These checkpoints let the project lead fund the work that the evidence supports and stop carrying estimates for work that has fallen away.
The existing pricing guide explains why changes to scope alter a provider’s quote. The in-house and agency guide helps assign who does the work. Use this allocation after those decisions to reserve capacity and decide what to fund when the project uncovers a different problem. GEO pricing and scope and GEO in house or with an agency



