This page walks every DomiDo user flow step by step from the perspective of the least confident persona. The method uses four questions at every action: will the user try the right action, will they notice it is available, will they associate it with the desired effect, and will they see progress after they take it. Whenever any answer is "no" or "uncertain", the design carries a deliberate response — a label change, a contextual help, a different layout, or a new screen. Customer flows are evaluated through Garden-Proud Geoff, the sixty-to-seventy-something iPad user who represents the least-tech-savvy consumer persona. Geoff is comfortable with technology but not an early adopter; he wants to design a curved raised planter bed for his garden. Admin flows are evaluated through the operations team, technically proficient on desktops. The flow-by-flow walkthroughs below record every place where the design steps in to keep Geoff from getting stuck.
A user with their own three-dimensional file finds Create on Home or in the tab bar, picks the Mode Selector, and chooses the upload option labelled in plain language ("Upload Your File", not "Upload 3D Model"). The file picker accepts the supported formats. Format-validation errors arrive inline before any upload begins, with a clear list of accepted formats and a suggestion ("Try simplifying your model in your three-dimensional software") if the file is too large. The processing screen — Geoff's strongest comfort zone — shows the twelve-stage stepper with named stages, so he sees forward progress at every moment. The Design Review screen places the price and the sticky "Add to Cart — [price]" call-to-action above the fold; on mobile the three-dimensional viewer is forty per cent of the viewport, leaving the price visible. A first-use gesture overlay teaches the viewer controls. After tapping Add to Cart, the cart and Stripe checkout flow are straightforward and the order confirmation completes the loop.
The cognitive load of the AI-mode flow peaks at prompt writing — translating a visual mental image into a description that an AI system can interpret is a skill none of the target personas have practised. The design's response is layered. On the Mode Selector, the AI-mode card states the total credit cost (around thirteen) so Geoff knows up front whether his balance covers the full pipeline. The card carries an amber warning when the balance is low but does not block him — he can proceed with informed consent.
The AI Prompt screen provides example prompt chips ("Curved planter wall, two metres long, in cream"), a brief instructional banner, and an optional reference-image upload. Tapping Generate triggers the staged-progress animation with elapsed time and "fun facts" content so Geoff knows the system is still working even during a ninety-second wait. The Generate button enters a Submitted state on tap so a nervous double-press cannot duplicate the request. The frontal-view preview screen labels the iteration text input as "What would you change?" and explains in microcopy that the AI will refine the image based on the new instructions — making it clear that Regenerate is not a one-shot lottery. A version-history strip shows previous attempts so Geoff sees his progress.
The projection workflow opens with a brief explanation: "We need eight views of your design to build an accurate three-dimensional model. Approve each view, or let us auto-generate the rest." Unusual angles (top, bottom) include visual reference examples that show what a correct bird's-eye view of a planter looks like, so Geoff does not waste time regenerating correct-but-unfamiliar projections. The "Auto-generate remaining views" accelerator appears above the fold after the first manual approval — saving up to thirty minutes for the user who would otherwise approve all eight projections by hand. Three-dimensional model generation uses outcome-based labels (High Quality recommended, Fast, Detailed) so Geoff makes an informed choice from quality, speed, and cost — not from infrastructure names. A countdown timer with stage descriptions shows progress. Credits are refunded on failure. The three-dimensional preview screen opens with the gesture-guide overlay so Geoff knows how to inspect the model. The reset-camera control returns the view to default in one tap if he loses his bearings. Design Review places the price and the sticky purchase call-to-action above the fold. The secondary "Start Assembly" call-to-action is replaced by a tertiary one further down so it cannot be confused with the purchase action; the language of the primary call-to-action is consistently "Add to Cart — [price]" everywhere it appears.
The catalogue surface shows the use-case categories the platform enables. A user enters from Home or the Catalogue tab, taps a category card, and sees Product Detail with a three-dimensional preview and parametric size sliders. The gesture-guide overlay covers the same affordances as elsewhere. The sticky purchase call-to-action and the consistent label leave no ambiguity about how to buy.
Gallery Browse pairs three-mode search (text, voice, image) with curated tabs and a masonry grid. The "Add to Cart — [price]" call-to-action on Gallery Design Detail sits at the top of the viewport — sticky, prominent, and using the same label as everywhere else. Social actions (like, comment, share, save to collection) sit below the purchase action so a user buying does not have to scroll past commenting first. The publish path is signposted from Design Review with a dedicated call-to-action and the moderation status is explicit.
Order Detail surfaces the Request Return call-to-action with a return-window countdown so Geoff sees how many days he has. The Return Request screen offers a clear reason dropdown with plain-language options and explains the grading process briefly so the user understands what determines the refund. Return Status uses a six-step timeline so progress is always visible.
The operations team sees a reorder alert on the Manufacturing screen with a pre-populated Purchase Order form. The Purchase Order Detail timeline shows every lifecycle step. Payment milestones surface on the same screen and on the payment calendar so the team never has to remember a date.
The pre-order flow's largest cognitive risk is the user confusing a pre-order with an order. The design's response is visual and explicit. The pre-order call-to-action carries a distinct visual treatment from the standard purchase call-to-action: a different colour and a "Pre-order: card verified, not charged until ready" badge — the critical "card not charged" fact appears before commitment, not after. Pre-order Confirmation reiterates the same statement at peak prominence. Pre-order Management cards carry the "Card Verified — Not Charged" status badge persistently. The pre-order card-expiry notification gives a seven-day countdown and a direct "Update Payment Method" call-to-action so a user never silently loses their pre-order.
Examination identified nineteen cognitive-walkthrough failures across the seven flows. The dominant category — fifty-eight per cent — was the third question: users could see the action but could not associate it with the intended effect. This is a systemic labelling, explanation, and context problem. The AI-mode flow carried forty-two per cent of failures by itself, driven by the prompt-writing barrier, the projection concept, provider names, three-dimensional viewer gestures, and the buried purchase call-to-action. The current design addresses each through prompt-writing guidance, an explanatory projection-workflow banner, visual references for unusual angles, the auto-generate accelerator above the fold, outcome-based three-dimensional generation labels, the gesture-guide overlay on every viewer, and a sticky purchase call-to-action at the top of every revenue-critical screen. Every cognitive-walkthrough failure is satisfied at the specification level.