A useful export SEO workflow connects a buyer question to a maintained page and a measurable next step. A spreadsheet containing thousands of phrases does not make that connection by itself. This tutorial uses a fictional felt desk-mat manufacturer that sells standard sizes and accepts corporate custom orders. An office buyer needs sample conditions, quantities and production confirmation; an individual shopper needs dimensions and care instructions. The workflow keeps those needs distinct while covering keyword research, on-page improvements and earned references. All company details and work schedules below are illustrative, with no promised traffic or revenue result.
Establish product evidence and choose a bounded skill set
Choose one target market and language before expanding the plan. Gather approved material descriptions, thickness, dimensional tolerances, available colours, customisation limits and current lead-time conditions. Mark missing certifications or performance claims as unresolved. An assistant must not invent waterproof ratings, environmental credentials or import-compliance conclusions to make a listing persuasive. Assign an owner to each product fact and a reviewer to each page. Define a qualified business enquiry in operational terms: for example, it includes a quantity range, requested dimensions and destination. Keep that definition separate from all form submissions. Arrange a staging website, a page inventory and authorised access to aggregate analytics before asking any tool to evaluate performance. Start with SkillPI Skill Finder, then inspect the target package, source revision, licence, dependencies and permissions. Two verified catalog entries have narrow roles here. Writing Guidelines reviews technical explanations in the sizing guide during the on-page stage. It does not supply keyword volumes, product evidence or ranking forecasts. Webapp Testing checks a running local page and a simulated enquiry journey. It needs the relevant Python and browser dependencies, a server command and test data. Its evidence concerns browser behaviour, not real email delivery, indexing or organic rankings. No SEO-specific SkillPI package was verified for this workflow. Keyword interpretation and outreach decisions therefore remain explicit human tasks. A missing specialised skill is a reason to keep a manual step, not to attach a package with an unrelated purpose.
Scenario one: turn keyword research into a page map
Build seed phrases from the product, the buying action and the user's practical problem. For the desk-mat example, these include felt desk mat, custom size desk mat, wholesale desk mats and how to measure a desk mat. Label each phrase with market, language, buyer type, intent and the date of the evidence. Transactional modifiers such as wholesale or custom do not remove the need to examine what buyers actually expect. Use an authorised research tool to inspect demand and competition. Record the tool, geography, period and definition of every metric. If volume is unavailable, preserve that unknown instead of asking an assistant to estimate an impressive number. Compare difficulty scores within a consistent tool and context. Inspect a small, relevant sample of visible results to determine whether the task is served by product pages, directories, tutorials or local suppliers. Do not automatically scrape a search engine or run an unapproved rank-query service. Separate the work into core commercial phrases, supporting product attributes and informational questions. Assign a corporate custom-order page to business purchasing needs, product pages to specific standard sizes and a measurement guide to sizing questions. Question-style AEO research can reveal useful explanations, but adding a FAQ is not evidence that an answer engine will cite the page. The deliverable is a page map, not a loose keyword dump. Each row contains the primary phrase, related expressions, destination page, intent, buyer question, existing evidence and next action. A row for custom felt desk mat wholesale might require the sample process and information needed for a quotation. A row for desk mat size guide might require a diagram showing keyboard placement and monitor-base clearance. Prioritise a small set that the team can actually maintain. If two existing pages address the same purchasing task, compare their content, links and conversion paths before deciding whether to merge or distinguish them. A change of URL needs a separate redirect and internal-link plan. Avoid changing a working public address merely to fit a prettier keyword. The page map should explain why each destination deserves to exist.
Scenario two: improve the page and the buyer journey together
Begin with basic accessibility to readers and site navigation. Can a visitor reach the page from a relevant category? Is its title distinctive? Is the main content readable on a narrow screen? Have the site owner or developer review index controls, canonical addresses, internal links and sitemap handling. Do not accidentally make a private staging site indexable during an optimisation exercise. A corporate custom-order page can be organised around applications, available specifications, samples, production confirmation, a quotation checklist and real questions. Its introduction should identify the intended buyer and scope. The specifications section must distinguish approved options from requests that need engineering review. The sample section should explain what must be confirmed before an order advances. Repeating a commercial phrase in every paragraph does not substitute for those details. For the sizing guide, describe a reproducible procedure. Measure the usable desktop, account for the monitor base and cable space, then compare the remaining area with the buyer's actual keyboard arrangement. State the units and what each measurement represents. Label diagrams as illustrations and keep product photography distinguishable from explanatory artwork. A drawing that looks plausible is not evidence of a manufactured tolerance. Use Writing Guidelines at this point to review clarity, missing conditions and ambiguous instructions. The product owner still verifies the facts. An example title is “Custom felt desk mats: sizes and sample process”. The description should summarise what the page actually contains; it must not promise immediate shipping while the page says production follows a quotation. Record the old wording, new wording, fact reference, owner and affected page so later reviewers can understand each change. Run Webapp Testing against the local or staging copy with synthetic data. Check narrow-screen layout, navigation, units, required fields and the route from the guide to the appropriate enquiry form. Save screenshots and a reproducible failure description, then recheck the corrected path. Do not contact a customer, place a live order or treat a successful browser submission as proof that an operational team received it. The mature open-source Lighthouse project can inform a performance audit, although it is not a linked SkillPI skill here. The Google page-experience guidance explains that good metrics do not guarantee top rankings. Prioritise issues that obstruct reading or enquiry completion, such as oversized images, layout movement or controls covered by overlays. A perfect score alone is not the business objective. Recheck the important journey after an optimisation changes layout or loading behaviour.
Scenario three: create a resource worth citing
Off-page work begins with something useful outside your own sales message. Produce an original desktop measurement diagram and a specification checklist that office planners can use. Record its author, version, permitted reuse and the evidence behind its measurements. Remove promotional claims that would make a resource page misleading. Give a reviewer enough information to confirm where an illustration ends and a verified product statement begins. Research publications and resource pages that genuinely serve the intended audience: office-design editors, relevant trade associations and procurement guides are possible categories, not automatically qualified targets. Review each candidate's actual topic, readers, existing references and public submission rules. A third-party authority score does not establish relevance or editorial quality. Maintain a candidate register with the page, reason for relevance, distinct resource offered, public contact-source provenance, review state and outcome. In this example, the assistant prepares a reviewable pitch draft only; it sends nothing automatically. The draft explains the measurement method and its limitations. Place a proposed reference where it helps a reader check a dimension, instead of demanding an exact-match commercial anchor. A brand mention without a link may have communication value, but do not count it as a measured ranking gain. The Google link-spam policy identifies links created primarily to manipulate rankings as problematic, including paid links that pass ranking credit. Avoid bulk-link purchases and manufactured exchange networks. Qualify advertising and sponsorship links under the applicable rules rather than presenting them as independent recommendations. Keep permission and publication records when a legitimate editor accepts material. The stage output consists of the resource, the reasons each candidate fits and any authorised publication evidence. It is not a guarantee of a particular number of high-authority backlinks. Review referral usefulness and audience relevance alongside any search observations. Remove or correct outdated resource claims when product specifications change.
Run a small weekly cycle with inspectable outputs
A practical first cycle might assign product-fact approval on Monday, a three-intent page map on Tuesday, one custom-order page and one sizing-guide draft on Wednesday, staging checks on Thursday and resource review on Friday. This is an example allocation, not a universal timetable. Split it further if one person owns several roles. Resolve factual gaps before polishing copy, and approve the source page before repeating its information elsewhere. For a concrete handoff, the researcher gives the writer the page task, supporting phrases and unanswered buyer questions. The writer returns a draft with unresolved facts clearly marked. The product reviewer approves or rejects each claim. The developer receives the approved page plus layout and form requirements. The tester returns browser evidence. The outreach reviewer receives the finished resource and a candidate rationale, not an instruction to send unsolicited messages. Observe queries and landing pages, the completeness of qualified enquiries and traffic to mistaken destinations. Group observations by market, language, page and change date. Note advertising changes, seasonality and product launches that could explain a fluctuation. Use a consistent observation window appropriate to the site's traffic and the decision being made. Do not attribute every change to the last SEO edit or promise a fixed six-month uplift.
Troubleshooting and frequently asked questions
What if impressions rise but useful enquiries do not? Inspect intent, destination coverage and the quotation path. Separate retail sizing questions from business purchasing requests and review anonymised examples. The solution might be clearer eligibility or delivery information rather than more keywords. What if several pages look almost identical? Return to the page map and define the distinct task each page serves. Replacing adjectives or country names does not establish a new purpose. Expand regional pages only when there are real differences and someone can maintain them. What if performance scores improve but orders stay flat? Check product fit, price, availability and enquiry handling. Performance is one variable in a longer purchasing process. Browser evidence can identify a broken control, but it cannot prove product-market demand. Should every skill be installed at once? No. Install for the current stage after inspecting source and prerequisites. Can uninstalling guarantee no effects? No: a package may have generated files or changed a project. Back up relevant work, inspect changes and follow the host's removal process. Does publishing a large amount of similar content accelerate success? Volume without useful, maintained differences creates review and quality problems; begin with the pages whose buyer task and evidence are clear. Acceptance should focus on three concrete outputs: an evidence-backed keyword-to-page map, accurate pages with a working enquiry journey and an original resource that helps the intended reader. Together they make the team's actions and remaining unknowns visible enough to guide the next cycle.