Field note · Mortgage broking
Half of what brokers send lenders is wrong first time. Here is where the hours go.
Five hours of administration on the broker-to-lender leg alone. Fifteen to twenty rounds of messages. Five people. And a one-in-three chance the case comes back for corrections anyway.
The short answer
Research published in May 2026 with more than 70 UK specialist lenders and brokers found that only half of submitted information is right first time. A typical secured loan case takes 15–20 rounds of messages, involves more than five people, and consumes upwards of five hours of admin on the broker-to-lender leg alone. The time is not in the advice. It is in the chasing.
Ask a broker where their week goes and they will not say "advice". They will say the phrase every broker says, which is chasing.
Until recently that was a feeling rather than a number. In May 2026 the messaging firm Nivo published research drawn from in-depth conversations with more than seventy UK specialist lenders and brokers, and put figures against it.
Why does a case take 15 rounds of messages?
Because the first submission is a guess about what this lender wants this month.
Each lender has its own pack, its own tolerances, its own unwritten preferences held in one underwriter's head. A broker working across twenty lenders is not running one process twenty times. They are running twenty processes badly, from memory.
So the first pack goes in incomplete. The lender asks for a thing. The broker asks the client. The client answers three days later, partially. Round four begins.
Where do the five hours actually go?
Not evenly. The Nivo figure covers only the broker-to-lender leg — the packaging, the sending, the answering of queries. It excludes the client-facing half entirely.
| The work | What it actually is | Can an agent hold it? |
|---|---|---|
| Re-keying | Moving the same fact between a fact-find, a portal and an email | Yes — this is the easy half, and it is most of the volume |
| Completeness checks | Knowing that this lender wants six months of statements, not three | Yes, if the rule is written down once |
| Chasing the client | Four polite messages over nine days for one document | Yes, and it never gets bored on message three |
| Chasing the lender | Knowing a case has gone quiet before the client notices | Yes — a timer is not a judgement call |
| Judging a marginal case | Whether this income shape will fly with this underwriter | No. This is the job. Do not hand it over |
Four of those five are clerical. One is the reason the client is paying you. Most firms automate none of them, because the pitch they get is "AI for mortgage brokers" rather than "the completeness check, for these nine lenders, by Thursday".
What does fixing it actually look like?
- Write down one lender's pack requirements. One. The one that generates the most rework. This is an afternoon with a spreadsheet, not a project.
- Put a completeness check in front of submission. It refuses to send an incomplete pack and names what is missing. No model needs to be clever for this.
- Automate the chase, not the decision. The client chase and the lender chase are timers with manners. They are the cheapest hours to buy back.
- Leave the marginal-case judgement alone. Every hour you free up should go there.
- Measure the correction rate before and after. If it does not move off 35%, the thing you bought did not work, and you should be able to see that in a month.
Why not just buy an "AI mortgage platform"?
You can. But notice what the Nivo numbers actually say: between 20% and 50% of cases stall or fail to complete. That is not a software gap, it is a coordination gap, and software that assumes a clean process will inherit the mess rather than fix it.
The cheapest first move is smaller and duller than a platform. Pick the single lender that causes the most rework, write its pack rules down, and refuse to submit without them. Then count.
Questions people actually ask
How many hours does mortgage case admin take?
Nivo’s May 2026 research with 70+ UK specialist lenders and brokers found upwards of five hours of administration per case on the broker-to-lender leg alone, excluding client-facing work.
What percentage of mortgage cases need corrections?
The same research found 35% of all cases require corrections or repeats, and that only half of submitted information is right first time.
Should a broker automate the advice itself?
No. The clerical work — re-keying, completeness checks, chasing clients and lenders — is where the hours are. Judging whether a marginal case will fly with a particular underwriter is the job the client is paying for, and it should keep a human decision point.
Sources
- Nivo, research with 70+ specialist lenders and brokers, reported by Mortgage Solutions, 7 May 2026. Figures: 50% right first time, 15–20 message rounds, 5+ people, 5+ hours broker-to-lender admin, 35% corrections, 20–50% stalling. — vendor research, named method and sample re-checked every 6 months
Checked 8 September 2026. Next scheduled check 7 March 2027. Numbers that move — leaderboards, live indices — are re-checked every 30 days; annual datasets and rules in force every six months; dated research once a year. If something here has gone stale before we got to it, tell us and we will correct it and say what changed.
Want the completeness check built for your top three lenders?
That is a scoped build with a fixed fee, not a platform migration. Tell us which lender causes the most rework and we will tell you what it would take.
Talk to us about this