Why clients ghost you on documents, and the workflow that fixes it

Ghosting is a design problem

Every broker has the same story. The client was engaged in the meeting, agreed on a strategy, said they would send everything through that night, and then went quiet for eleven days. When they finally responded they sent four of the nine documents, two of them photographs of a screen.

It is tempting to read this as a client problem. It is almost always a request problem. The client did not decide against you. They opened your email, could not work out what to do first, felt the whole task in one go, and deferred it. Then the email moved down the inbox and the task disappeared with it.

That is worth taking seriously because document collection is not a customer service inconvenience. It is the point where the deal is most likely to stall, and under the best interests duty it is also where you are gathering the evidence your file will need. ASIC's Regulatory Guide 273 expects your records to include the responsible lending assessment or the information you would use to prepare it. The documents are not overhead. They are the file.

The request is usually the thing that is broken

Look at a typical first document request and you will usually find three faults at once.

It is batched. Nine items arrive together, including some the client will find in two minutes and some that require ringing an accountant. The client reads the hardest item, estimates the whole task from it, and defers all nine. Batching makes an easy task feel like a hard one.

It is vague. Asking for recent payslips does not tell the client how many, from when, or whether a screenshot is acceptable. Every ambiguity is a decision they have to make, and any decision they are unsure about is a reason to leave it until later. Later does not arrive.

It has no date and no consequence. A request without a deadline is a request without priority, and it competes with everything else in the client's week from a standing start.

Fix those three and a large share of ghosting stops without any technology at all. Ask for two things first. Name each one exactly, including the period covered and the acceptable format. Give a date, and say what it affects, such as the pre-approval expiry or the finance clause.

Stop collecting documents over email

Email is the default channel because it is the path of least resistance, and it is the wrong one on two counts.

The operational count is that email has no state. There is no list of what is outstanding, no record of what has been received, and no way for the client to see progress. Every follow-up requires you to reconstruct the position by scrolling. Multiply that by a live pipeline and you have created a job whose entire content is remembering things a system should remember.

The privacy count is more serious. You are asking clients to send identity documents, payslips and bank statements as attachments. NextGen.Net's Tony Carn has made the point publicly that many applicants would be alarmed to learn documents are still being emailed, and he is right. Under the Privacy Act 1988, Australian Privacy Principle 11 requires you to take reasonable steps to protect personal information from misuse, interference, loss and unauthorised access.

The OAIC's data breach statistics for January to June 2025 recorded 532 notifications, with malicious or criminal attacks accounting for 59 per cent and human error 37 per cent. The finance sector reported the second greatest number of breaches at 14 per cent of the total. Email is where human error lives. A portal with per-client access is not a luxury purchase, it is the reasonable step.

Open banking removes the largest single ask

The heaviest item on most document lists is transaction history. It is also the one clients most resent, because gathering months of statements across several accounts is genuinely tedious and feels invasive.

The Consumer Data Right lets an accredited party receive that data directly from the bank with the consumer's consent, which removes the collection task entirely. Adoption is moving. Frollo's State of Open Banking 2025 reported that more than 32,000 Australians had shared data with their broker over the year, that nearly 3,000 brokers had been onboarded, representing more than 13 per cent of the national broker population, and that consent requests had risen sevenfold to more than 6,200 in July 2025.

The data is also better than the screen-scraped alternative. Frollo has published that merchant names are available in 52.7 per cent of relevant open banking transactions compared with 31.7 per cent of screen-scraped ones. Cleaner categorisation means less manual work reconciling a client's living expenses, which is a second time saving on top of not collecting statements at all.

The direction of travel is fairly clear. MFAA chief executive Anja Pannek has said publicly that a ban on screen scraping is a question of when rather than if. If your current process depends on a screen-scraping tool, that is worth planning around now rather than reacting to later.

Open banking has its own failure rate, and you have to design for it

This is where most write-ups of open banking stop, and it is exactly where a broker needs the next paragraph. Consent is not frictionless, and if you deploy it as though it is you will simply move your ghosting problem to a new screen.

The same Frollo research reported that close to 30 per cent of consent authorisations fail during bank authentication, that 88 per cent of those failures come down to login issues, one-time passcode problems or technical errors, and that 67 per cent of its users had an unsuccessful attempt at authorising consent the first time they tried.

Read that last figure again in the context of your own pipeline. Two clients in three will hit a problem on their first attempt. If your process is to send the consent link and wait, most of your clients will fail once, and a meaningful share of those will not try again unprompted. That is not a client who ghosted you. That is a client who tried, got an error, and had nowhere to go.

The design response is straightforward. Tell the client before they start that they will need their internet banking login and their phone for a one-time code. Send the link when they are likely to be somewhere they can complete it, not at 9pm. Watch for the failure rather than the success, and follow up on a failed authorisation within a day, by phone, offering to stay on the line. The failure rate is the thing you build the workflow around.

Identity is a separate bottleneck with a separate fix

Verification of identity behaves differently from financial documents. It usually cannot be satisfied by open banking, it often has lender-specific requirements, and it produces rework at the lender's end when it is done imperfectly.

It is worth handling it with a purpose-built tool rather than folding it into the general document request. NextGen publishes that applications using its NextGenID digital verification service can result in a 25 per cent reduction in time to approval and up to four days faster processing, and that the service reduces missing information requests issued by the lender. Those are the vendor's published figures rather than an independent measurement, and Finance OS, which publishes this article and also builds broker software, has not tested the product.

The general principle holds regardless of which tool you use. Missing information requests are the expensive kind of delay, because they arrive after submission, when the client has mentally finished and is hardest to re-engage. Anything that moves a document problem from post-submission to pre-submission is worth more than its face value.

The workflow that actually fixes it

Put the pieces together and the working version looks like this.

One channel, established in the first meeting. Tell the client where documents go, show them the portal on your screen before they leave, and never accept an attachment by email afterwards, because the first exception becomes the new default.

Consent first, while you still have their attention. Initiate open banking during or immediately after the appointment rather than in a follow-up email, and walk the client through the first authorisation while you are with them. Given that most first attempts fail, doing it in front of you converts your highest-friction step into your easiest one.

Then two documents, not nine. Ask for the smallest, most certain items first, name them exactly, and give a date. Completing a small task creates the momentum for the next one, and a client who has successfully uploaded something once will do it again.

Then release the rest in a sequence, with the item that requires a third party, such as an accountant's letter or a trust deed, flagged at the very start even though it is asked for later. Third-party items have lead times that your reminders cannot compress, so they need to be started early even when they are not needed early.

Reminders should be automatic and specific. A reminder that names the one outstanding document and its deadline works. A reminder that says the broker is still waiting on documents does not, because it puts the work of remembering back on the client. If a person on your team is manually chasing, the system has failed and you are paying a salary to cover for it.

And the client should be able to see the position without asking. A visible checklist showing what is received and what is outstanding removes most status enquiries and, more usefully, converts an ambiguous obligation into a finite list with an end.

Measure the right thing

Most brokers measure document collection by how it feels, which means they remember the two files that dragged and forget the twenty that did not.

Three numbers are worth tracking. Time from first request to complete file, measured per deal. The proportion of open banking consents that fail on first attempt, so you know whether your pre-briefing is working. And the count of missing information requests you receive after submission, which is the cleanest measure of whether your pre-submission checking is real.

If your average time to a complete file is a fortnight and your open banking consent succeeds first time for a third of clients, you do not have a client engagement problem. You have a workflow with two identifiable faults and a clear order to fix them in.