Client portal module

Messages: The Module You Should Sometimes Not Use

A chat window in a client portal is a trap as much as a feature. Because it accepts anything, it collects everything — revision feedback, task requests, meeting asks, brand assets — and all of it then has to be manually rehoused by someone on the agency side. This guide covers how Messages works in an Arpixa portal, and, just as usefully, which questions belong somewhere else.

The Messages screen in the Arpixa client portal showing a conversation with an agency, a topic selector and quick starter prompts

One thread per agency, and what follows from that

The structural decision to understand first: a client has one conversation with each agency. Not one per project, not one per topic. Opening Messages finds that conversation or starts it, and from then on it is a single running history.

The upside is that nothing is ever in the wrong place. Anyone who has used a portal with per-project threads knows the failure mode: six conversations, a question that spans two of them, and a client who picks one at random. A single thread cannot be filed incorrectly.

The downside is that the thread has no idea what any message is about. There is no project attached to a message, no deliverable, no invoice. Which is exactly why the topic field exists, and why the routing table below matters more here than in any other module.

Everything in the thread is live. Messages arrive without a refresh, typing is visible, and read state updates as the other side opens it. For a client working with several agencies, each conversation is separate and scoped to that agency — there is no shared channel and no chance of a message reaching the wrong supplier.

Which module for which question

This is the part of the page worth bookmarking. Nearly every request a client sends in a chat window has a better home, and using it costs the client nothing while saving the agency a transcription step.

Where each kind of client request belongs
If you want to say… Use Because
"Can you change the headline on page two?" Deliverables It is revision feedback on a specific item. Sent there it arrives attached to the file, with a priority, and it stops the deliverable being marked as approved by accident.
"Can we also add a favicon?" My Projects It is a task request. Raised from inside the project it lands already attached to the right work, rather than needing someone to transcribe it.
"Can we get thirty minutes on Thursday?" Calendar A date request with a type and a description, which the agency can approve or decline. A message about a time slot needs a human to convert it into a calendar entry.
"Here are the brand assets you asked for." Files Forty logos in a chat thread are unfindable in a fortnight. Uploaded to Files they are badged, organised into folders, and permanently addressable.
"I want to start a new piece of work." Brief Builder A structured brief captures budget, timeline, audience, and goals in fields. A paragraph in a message captures roughly a third of that and invites four follow-up questions.
"When is the next milestone?" Timeline Already answered. Asking costs a round trip; looking costs five seconds.
"I do not think we agreed to this scope." Messages Genuinely a conversation. Anything requiring negotiation, judgement, or tone belongs here.
"This invoice line does not make sense to me." Messages Also a conversation. There is even a built-in topic for it, because querying a bill needs explanation rather than a form.

The pattern behind the table is simple. If the request has a structure — a file to approve, a date to book, a task to add, an asset to store — the module that owns that structure will capture it better than a sentence can. Messages is for the things that genuinely are conversation: disagreement, negotiation, explanation, reassurance, tone.

Agencies get more out of this than clients do, so it is worth teaching explicitly at kick-off. A client who learns in week one that revision feedback goes in Deliverables saves a year of copy-and-paste. The wider habit is covered in how to keep clients updated.

The topic field, and why it earns its click

Above the message box sits a topic selector with five labels plus the option to leave it blank: Project Update, Invoice, Urgent, Deliverables, and General.

Since the thread carries no project or invoice reference, the topic is the only structured signal a message has. In a small agency that is the difference between a billing question landing with whoever does billing and it sitting in a general queue until someone reads all of it.

A word on Urgent. It works exactly as well as it is used sparingly, and it stops working the first week everything is urgent. The honest version: use it when a delay has a consequence you can name, and use Project Update when you would simply like an answer today.

Leaving the topic blank is allowed and is a small missed opportunity. It costs one click.

Four quick starters

On an empty conversation, four one-tap openers appear beneath an invitation to get in touch about a project or an invoice. Each one fills in both the topic and a reasonable opening sentence:

  • Asking for the latest update on a current project.
  • Asking for help understanding an invoice item.
  • Flagging something urgent that needs a quick response.
  • Asking for the expected timeline on the next deliverable.

These solve a real problem rather than being decoration. An empty text box with a blinking cursor is one of the most reliable ways to stop someone getting in touch at all, particularly a client who is not sure whether their question is worth asking. Four prefilled openers say, in effect, that these are normal things to ask about.

Note what all four have in common: they are questions, not instructions. That is a fair reflection of what the thread is best at.

Attachments, and the 5MB line

Files can be attached to a message, by picking them or dragging them in, with a limit of 5MB per file. Anything larger is refused at the point of selection rather than failing after an upload.

That ceiling is a design statement as much as a technical one. 5MB comfortably covers a screenshot, a marked-up PDF, a single document — the things that only make sense next to the sentence explaining them. It does not cover a folder of raw photography, and it is not meant to.

The rule of thumb: attach it to a message if the sentence is the point and the file is the evidence. Put it in Files if the file is the point. A brand pack buried forty messages back in a chat thread is effectively lost; the same pack in a folder is still there in a year.

Read state, typing, and availability

Three live signals run in the thread, and together they answer the question that makes clients anxious: has anyone seen this?

  • Read state on your messages, so you know the difference between unseen and unanswered.
  • A typing indicator when someone on the agency side is composing, which appears while they type and clears shortly after they stop.
  • An availability marker on the agency, so you can tell whether silence means they are heads-down or simply not there.

The first of those does the most work. “Seen and not yet answered” is uncomfortable for an agency and enormously reassuring for a client, and it removes the most common cause of a duplicate message being sent an hour later.

Keep client conversation next to the work

Start free in minutes, or log in to your Arpixa workspace. See pricing for plan details.

The client thread is part of the workspace, not a communications add-on. Arpixa plans run from a free tier with a single client, through Starter at $12 a month with twelve, to Pro at $29 and Advanced at $89 with no client limit. See pricing.

Clearing and deleting

Two housekeeping actions exist. A client can delete an individual message they sent, and can clear the conversation.

Be clear about what these are for. They are tidying, not retraction. A message that has already been read has already been read, and deleting it afterwards does not unsee it. The genuinely useful case is the duplicate, the half-typed line sent by accident, the attachment added to the wrong message.

Clearing the conversation is the heavier of the two and worth pausing over, because a running thread with an agency is also a record of what was agreed informally. That is often the only place a small decision lives.

Six habits that get faster replies

None of this is enforced by the software. All of it changes response times.

  1. Set the topic before you type. It is one click and it changes how the message is triaged.
  2. Put the ask in the first sentence. Context is useful; context first is not.
  3. One subject per message. Three unrelated questions in one paragraph get one answer.
  4. Name the thing. "The homepage hero" rather than "the top bit".
  5. Say when you need it by, and say why if the date is real.
  6. If it belongs in another module, send it there instead. The routing table above is the whole trick.

The second one is the highest-leverage. An agency reading a message on a phone between meetings will act on a first sentence that contains the ask and will defer one that opens with three sentences of background. That deferral is where most of the delay in client communication actually comes from.

Where this sits relative to the agency inbox

Worth stating for anyone comparing portals: this thread is client-facing, and it is not the same surface as the agency’s internal Inbox.

The agency’s internal inbox collects workspace updates, activity, and team coordination, and clients never see it. Internal discussion about a client — scoping, pricing, frustration, strategy — stays there. The client thread carries only what the agency chooses to say to the client.

That separation is not a limitation, it is the point. A single channel shared by internal and external conversation is how an agency accidentally sends a client something they were never meant to read. The broader case is in how to keep internal notes private from clients.

How to send a message that gets answered

  1. Open Messages in the portal and pick the agency if you work with more than one.
  2. Choose a topic from the dropdown, or tap a quick starter to have both topic and text filled in.
  3. Lead with the question or the decision you need, then add context underneath.
  4. Attach anything relevant, keeping each file under 5MB, or use Files for large asset sets.
  5. Send, and check the read marker rather than resending.

The last step is worth naming as a rule: check the read marker rather than resending. A resent message does not make anyone reply faster and it does make the thread harder to read later.

Frequently asked questions

How does messaging work in the Arpixa client portal?

Each client has one continuous conversation per agency. Opening Messages either finds that conversation or creates it, and everything is live: new messages arrive without a refresh, typing is shown, and read state updates as the other side opens the thread. There are no separate per-project threads, so the conversation is a single running history with that agency.

Are there separate message threads per project?

No. The model is deliberately one thread per agency rather than a thread per project, which keeps a client from having to guess which of six conversations a question belongs in. The trade-off is that project-specific asks are better raised in the project itself, where they arrive attached to the work.

What is the topic field for?

It labels the message so it can be triaged. Options are Project Update, Invoice, Urgent, Deliverables, and General, plus leaving it blank. It costs one click and it is the difference between a billing query reaching whoever handles billing and it sitting in a general queue.

What are the quick starters?

Four one-tap openers on an empty conversation, covering asking for a project update, querying an invoice item, flagging something urgent, and asking for the timeline on the next deliverable. Each fills in both the topic and a sensible opening sentence, which solves the blank-box problem that stops people asking in the first place.

Can a client attach files to a message?

Yes, with a limit of 5MB per file, and drag-and-drop works. It is the right tool for a screenshot or a single document that only makes sense next to the sentence explaining it. For asset sets, brand packs, or anything anyone will need to find again, Files is the better destination.

Can I tell whether my agency has read the message?

Yes. Read state is shown on your messages, and a typing indicator appears when someone on the agency side is composing a reply. The agency also carries an availability marker, so you can see whether they are around before deciding whether silence means anything.

Will my agency get an email when I send a message?

Treat the thread as an in-portal channel rather than an email trigger. Messages arrive live in the agency workspace, and the read marker tells you when they have been seen. If something is genuinely time-critical, use the Urgent topic and consider a second channel rather than relying on the thread alone.

Can a client delete messages?

A client can delete their own messages, and can clear the conversation from their side. Clearing is a housekeeping action rather than a way to retract something that has already been read, so it is best thought of as tidying rather than undoing.