A fast-loading checkout still responds slowly to clicks
Use interaction-latency evidence for the reported click, tap or keypress; a fast initial-render result does not answer that complaint. Largest Contentful Paint measures when the largest visible content element renders relative to navigation. Interaction to Next Paint observes latency from an interaction until the next paint. For a shipping change, also record when the requested total or selection finishes updating: the next painted frame may be feedback that work has started, not the completed result.
For: A research-only merchant comparing a fast page-load result with real complaints about sluggish checkout controls.
If the complaint begins after the checkout is already visible, a measurement starting at navigation answers a different question. Record the action in ordinary terms: choosing a listed shipping method, opening an existing control or typing into a field. Identify what the buyer expected to change and which part appeared to wait.
Chrome’s Largest Contentful Paint documentation defines LCP as the render time of the largest visible image, text block or video relative to navigation. It is evidence about initial content rendering. A favorable LCP measurement does not establish that a later shipping selection or text entry responds promptly.
Chrome’s Interaction to Next Paint documentation describes observing latency after clicks, taps and keyboard interactions through the next paint. That makes interaction measurement relevant to delayed visual feedback from those inputs. It does not make LCP useless; it means the two measurements answer separate questions and should remain separately labeled.
Separate visible feedback from the completed update
For the reported action, identify three observable points where available: when the input occurred, when the page next displayed feedback, and when the requested result finished appearing. In a typing complaint, note whether entered characters appear late. In a shipping-selection complaint, distinguish delayed feedback on the selection from a later wait for the total to finish updating. Record only what occurred in the real session.
The distinction follows from INP’s endpoint: it ends at the next paint, not at a definition of completed shipping work. If the next visible change is a waiting indicator, that change does not establish that the final total is ready. Preserve a separate start-to-result observation when the complaint concerns that later wait.
Conversely, a total that eventually updates does not show that the page responded promptly to the initial click. A slow interaction and a slow result can occur together. Keep both observations if the evidence shows both; do not relabel all waiting time as INP or assume every shipping delay is a browser-response problem.
Keep the measurement attached to its session
Ask for evidence that includes the actual action, device, browser, page state and time. A number from a different URL or a visit in which nobody used the affected control cannot identify the reported interaction. Where the timing was not measured, keep the complaint as an observation and mark the measurement unavailable.
Give the implementer the control name and the visible state before and after the action, together with any existing trace that measures it. A page-level metric can be useful context, but an unattributed value does not identify the component responsible for the reported wait. The investigation needs the action associated with the measurement.
Observe the live checkout only up to actions that do not submit an order or payment. Use existing evidence for complaints after payment submission. Do not ask a buyer to repeat Pay to obtain a timing, and remove personal input, payment data and private URL tokens from evidence shared for the investigation.
Require a proposed fix to address the measured interval
Before approving a page-speed proposal, ask which recorded interval it targets and what evidence will show improvement on the affected control. If the only result offered is a better initial-render score, the typing or click-response question remains unanswered. If feedback is prompt but the final shipping result is slow, request investigation of the component that produces that result rather than assuming an initial-load change will resolve it.
Assign the frontend maintainer the delayed-feedback evidence and identify the owner of any later update only when the component is known. A control’s label does not prove whether a theme, extension or external service caused the wait. Unknown ownership is a concrete item to resolve, not a reason to blame the payment provider.
After a targeted change, compare the affected action under the recorded conditions and state whether feedback and result completion were measured. Keep any unmeasured part open. For a Prism consultation, describe the checkout and the action-specific evidence; confirm investigation or implementation scope, responsibilities, fees and terms before work. A performance observation establishes neither processing eligibility nor future sales outcomes.
Performance question map
Use one reported action from one real session. Keep the initial-render measurement, interaction latency and time to the requested result in separate entries. Record units and measurement source; an unavailable measurement stays unavailable rather than becoming a score inferred from how the page felt.
Worksheet entries are not submitted by Prism’s worksheet and are not saved by the site. Use record types, availability, anonymized observations, or match/mismatch results. Do not enter government identifiers, customer names or addresses, customer messages, receipt-access links, card or bank details, passwords, or keys. Send sensitive documents only through the provider’s verified secure channel.
Performance question map. The last column is for temporary notes.
Performance field
Evidence to record
Interpretation for the work request
Your observation
Reported action
Evidence to recordExact control or field, input method, checkout state and expected visible response.
Interpretation for the work requestDefines the task the proposed fix must address.
Initial render timing
Evidence to recordLCP value if available, with URL, session context and measurement source.
Interpretation for the work requestDescribes largest-content rendering relative to navigation; it does not settle the later input complaint.
Interaction latency
Evidence to recordMeasurement tied to the actual click, tap or keypress through the next paint, or unavailable.
Interpretation for the work requestSupports investigation of delayed feedback; do not substitute the page-load timer.
Requested result completion
Evidence to recordObserved start and completion of the shipping selection, total update or other requested result.
Interpretation for the work requestKeeps a later wait separate when the next paint only showed feedback that work had begun.
Affected component
Evidence to recordVisible control and known theme, extension or service owner; mark unidentified components.
Interpretation for the work requestRoutes the evidence without inferring technical cause from the control’s name.
Session conditions
Evidence to recordDevice, browser version, time and page state for the measurement.
Interpretation for the work requestMakes the reported action comparable to the affected check after a change.
Investigation owner and acceptance
Evidence to recordNamed owner, interval the proposed change targets, and evidence required from the same action.
Interpretation for the work requestPrevents acceptance of a load-only result as resolution of an interaction complaint.
These are temporary notes. Leaving or reloading this page may clear them. Worksheet entries are not sent automatically. If you copy notes into the consultation message and submit the form, Prism receives them as part of your request.
Limits
The source definitions support selecting the relevant measurement. They do not diagnose this store’s cause or set a contractual performance target.
A next paint does not establish that an asynchronous update or payment has completed.
Do not submit or repeat payment to measure responsiveness, and do not include customer input or payment credentials in shared traces.
Chrome Largest Contentful Paint — checked 2026-09-21. Largest Contentful Paint records the render time of the largest visible image, text block or video relative to navigation. It supports the initial-render side of this comparison.
Chrome Interaction to Next Paint — checked 2026-09-21. Interaction to Next Paint observes latency from click, tap and keyboard interactions until the next paint. That endpoint supports separating interaction feedback from the later completion of a requested result.