Examine Inclusion and Access Across Channels

A service problem is incompletely framed when a team describes only its digital users. A person may start on the website and finish by phone or at a desk. Define the outcome sought, then list the channels the service actually offers and the handoff between them. Do not turn a channel into a trait of the person. Using a phone once does not prove a lack of digital ability, and an unfinished form does not prove poor skill. Ask what happened in the task, what alternative existed, and what evidence shows the person reached the outcome.

Use an access matrix with four columns: entry channel, observed obstacle, operating alternative, and task outcome. Each row needs a checkable fact, such as an upload that failed in a mobile attempt or a call that ended before a request was recorded. Do not write that phone is easier for everyone. For a failed upload, inspect file type and size guidance, connection state and system error before assigning one cause. If calling is the fallback, check whether phone staff can resume the same request without making the person repeat every detail or disclose data the task does not require.

Compare channels using their own denominators. If three of five digital attempts and four of four calls complete, the descriptions are 3÷5=60% and 4÷4=100% in those samples. The figures do not prove that phone is causally better. Tasks, staff assistance and case selection may differ. A useful comparison records task type, prior information, failure point and final outcome under a case ID that does not identify the person. If attempts cannot be linked, describe channel results separately rather than pretending they form a single journey.

Inclusion also means knowing what happens after a transfer. Someone who starts online and then calls needs the same request state or a safe way to resume it. 'Call us' is not enough if the line is unavailable or the request disappears between channels. Keep three checks separate: could the person reach a channel, understand the next step, and obtain a correct recorded outcome? Page wording may improve while record handoff stays broken. A bounded trial therefore checks the upload message, then transfer to phone, then linked results rather than incomparable groups.

Research can remain brief and privacy conscious. Use training cases or de-identified data and ask about a particular attempt rather than 'Is this channel easy?' Obtain consent for notes, and do not request a real document image when the research question is message comprehension. Record any channel whose users could not be recruited. Their absence is an evidence gap, not silent approval to close that channel. The lesson output is a decision with an observed obstacle, a testable alternative, an explicit denominator and a limit on what is unknown.