exxos's Atari, Amiga & retro forum

A while back I posted about wasting a load of time on accounting software, ending up on Clear Books, and then finding two months later that the Lloyds bank feed still did not work. This is the follow up to that. I have gone back and looked at pretty much everything on the market for a sole trader with two businesses, and this is the full write up: what I looked at, what each one cannot do, what it costs, and where I ended up.

It is long, because the answers are scattered across a dozen vendor sites and half of them are wrong. Writing it up mainly so I do not have to work it all out twice. If anyone else is a sole trader staring down MTD, this might save you an evening or several.

Why I'm looking at this now

I run two separate sole trades. Under Making Tax Digital for Income Tax, that means separate digital records and a separate quarterly update to HMRC for each one, so eight submissions a year instead of four, plus one final declaration at the end that replaces the old SA100.

I probably have until April 2028 before this actually applies to me, so there is no rush. I would rather sort it out quietly in the background now than discover the problem the week it starts.

The reason I care at all is that I do not want to be doing this by hand every three months. At the moment I export bank and credit card statements, run them through a script that splits everything between the two businesses, and untangle the marketplace transactions manually because the card statement just shows a transaction number with nothing to say what it was for. Doing that once a year in January is tolerable. Doing it four times a year, remembering to download everything, upload it, run the script and tick the boxes, is exactly the sort of recurring faff I want automated away.

What I actually want is simple. One bank feed that pulls itself in, one list where I click which business each transaction belongs to, rules that handle the predictable stuff on their own, and a button that files the quarterly updates. That is it. No invoicing, no VAT, no payroll.

Turns out that "one list, tick which business" is the hard part.

The rules, for anyone who has not read up on it yet

A few things worth knowing before you look at any software at all.

Qualifying income is gross turnover before expenses, not profit, and it combines every sole trade plus any property income. Two businesses at £15,000 each is £30,000 for threshold purposes, not two lots of £15,000.

The thresholds are over £50,000 from April 2026, over £30,000 from April 2027, and over £20,000 from April 2028. HMRC decides your start date from the tax return two years earlier, not from what you are earning now. So income going up next year does not drag you in next year.

VAT registration has nothing to do with it. They are separate regimes entirely, which took me a moment to get straight.

Quarterly updates are cumulative year to date, not four separate snapshots. Each one supersedes the last, so anything you get wrong or miss in one quarter gets swept up automatically in the next. HMRC's own line is that errors are corrected in the next update rather than amended. Only the final declaration has to be right. There are also no penalty points for late quarterly updates in your first mandated year.

Spreadsheets are still allowed. You keep records in a sheet and use bridging software to file, as long as there is a digital link and you are not retyping figures into a web form by hand.

Software I looked at and why most of it does not work

The thing that kills nearly everything is two sole trades. Almost every product is built for one trade, or for one trade plus a rental property, which is the common case. Two trades is the awkward one.

Sage Sole Trader. Free, HMRC recognised. Sage's own support say it is designed for one income stream and a second trade needs a second subscription. Out.

FreeAgent. Free if you bank with NatWest, RBS, Ulster or Mettle. Their own documentation says multiple self-employments need a separate FreeAgent account each. Out.

Clear Books. This was the near miss. The free plan is genuinely free forever, HMRC recognised, covers quarterly updates and the year end return with no upgrade wall at filing time, and includes bank feeds. You can run multiple businesses under one login. The problem is that each business is a completely separate set of books. With one bank account feeding both, their support's own answer is to link the same feed into both businesses, explain the relevant transactions in each, and mark the other business's transactions as personal so the balances still reconcile. That is going down the same transaction list twice. Free, but double the clicking, which is the precise thing I am trying to avoid. Also worth noting their auto-categorisation is a paid feature, not free.

I also have history here. I signed up to Clear Books the first time round and the Lloyds bank feed never worked. Two months on it was still broken, and their own community thread about the Lloyds outage was still sitting there unfixed. That is what sent me back to square one and started this whole second round of investigation.

It also raises the thing that actually worries me about all of this. Picture explaining to HMRC that your submissions are late because the software could not talk to your bank. It is not the software company that gets the penalty, it is me. Handing a legal obligation to a third party's flaky integration and then carrying the risk personally is a rotten arrangement, and nobody involved seems especially bothered about it.

For what it is worth, I have since learned that the fallback saves you here. Every one of these products accepts a CSV statement import, so a dead feed means you go back to exporting by hand for a quarter rather than missing a deadline. Combine that with the cumulative nature of quarterly updates and the no penalty points grace period in your first mandated year, and it is less catastrophic than it looks. Still not a reason to pick software with a feed that does not work.

QuickFile. Free under 1,000 ledger entries a year, £60 plus VAT after that. The MTD Income Tax module needs a Power User subscription at £60 plus VAT regardless of size, and bank feeds are £15 plus VAT a year. Lloyds is listed as supported for current accounts, savings and credit cards, and they publish a live open banking coverage dashboard and a system status page, which is more honesty than most manage. The killer is that their nominal mapping screen only offers self-employment and UK property, and their staff confirmed one account handles self employment plus property. Nothing suggests two sole trades in one account, which would mean two accounts and roughly £150 plus VAT.

FreshBooks. Has a genuine multi-business feature under one login, but their own help says each new business gets its own 30 day trial, which points at separate billing. I also could not confirm it files MTD for Income Tax itself, and the route being marketed is exporting a CSV into a separate bridging tool. Two products and two subscriptions is the worst of both worlds.

Coconut. Bridging tier at £49.99 a year, but the year end declaration is another £49.99 on top, so about £100 for full compliance with no bank feed. Full filing tier is £129.99 a year with feeds and categorisation. It advertises multiple businesses and income sources, which may well mean what I want, but I could never confirm whether that means two self-employments or the usual trade plus property, nor whether the transaction screen actually offers a per-business tick.

Bank provided tools. Starling launched a free MTD for Income Tax tool built into its sole trader accounts in March. Tide has a free integrated one. NatWest and friends give away FreeAgent. Zempler gives two years of Coconut. All free, all one trade per account, so switching banks solves nothing unless you open two accounts. Lloyds, incidentally, offers nothing of the sort. Their call centre reportedly tells people they only support QuickBooks and Sage, which is nonsense, since open banking access is a regulatory obligation and not a partnership list.

Do it yourself. Worth mentioning because it was the closest thing to a proper solution. GoCardless Bank Account Data, formerly Nordigen, has a genuinely free tier: 50 requisitions a month, 720 days of history, refresh up to four times a day, and it is a licensed AISP. It returns clean JSON per transaction, so you could feed it straight into your own script and then use a cheap bridging tool to file. Actual Budget documents the GoCardless connection officially if you would rather not write the connector. Total cost around £100 a year and you keep your own pipeline. Very tempting for about ten minutes, until I remembered that the whole point was to stop maintaining a pipeline.

The one that fits

Xero, on the Simple plan, at £7 a month excluding VAT. About £101 a year. More than free, which grates, but it is the only thing I found that does the job in one pass instead of two.

The mechanism is tracking categories. One organisation, one bank feed, one reconcile list, and each transaction gets tagged to a business. Xero publish a guide specifically called "Managing multiple MTD for Income Tax businesses or taxpayers with one Xero subscription", which opens by saying it is for people running more than one sole trade. Their support confirmed up to five businesses in one organisation, and that Xero then files separate quarterly updates per business.

Better still, bank rules can set a tracking category automatically, and a rule can be scoped to a single bank account. One of my accounts is used solely for one of the businesses, so that entire account can be tagged by one rule with no conditions at all and never touched again. Personal Lloyds current accounts and personal credit cards are both supported for automatic feeds, at no extra charge, which was the last practical worry.

The caveats, and there are some.

Assigning a tracking category is not enforced, so an untagged transaction silently goes nowhere. Given quarterly updates are cumulative it self-corrects next quarter, but your in-year tax estimate will be optimistic until you tidy up. A filter for unassigned items before submitting deals with it in a minute.

Marketplace purchases still have to be sorted by hand. Rules match on the merchant description and every one of those lines looks identical, so no product on earth solves that. It goes in a review pile, which is what I do now anyway.

Xero has form for putting prices up. Between two versions of the pricing page during this very investigation, Ignite went from £16 to £18, Grow from £37 to £39, Comprehensive from £50 to £55 and Ultimate from £65 to £70. Simple held at £7, for now. Reviews also flag rigid billing and refused refunds, so be deliberate about cancelling if you ever leave.

Reviews overall are 4.4 out of 5 with customer service at 4.0. Ease of use, bank feeds and transaction matching get consistent praise. The rebuilt invoicing system gets consistently slated, though I do not invoice so I do not care.

The support experience, which deserves its own section

This was the genuinely infuriating part.

Xero has no inbound phone number. They dropped email support in January 2019, and emailing now just gets an auto reply telling you to log in. Their contact page has one published address and it is for job applicants. Every route for a prospective customer says the same thing: register for a login first, then raise a case.

So you cannot ask a pre-sales question without becoming a user. You cannot find out what you are buying without buying it, or at least handing over your details. Xero Central discussions are behind the login too, so you cannot even read what other people have asked.

There is no "create account" link either. Every button says Buy now or Get one month free, so a page that is actually a free signup reads like a checkout. I only found the signup URL by digging through the page source.

And when you finally get in, you land at an AI chatbot.

Which then gave me confidently wrong information, twice.

It told me the Simple plan is "formerly the Starter plan". It is not. Starter was retired in September 2024 and replaced by Ignite. Simple is a separate, newer plan added below Ignite.

Having invented that, it then told me the Simple plan has a limit of 20 bank transactions a month for reconciliation, and suggested I would need to upgrade to Grow or Comprehensive. That is three layers of wrong. The 20 transaction limit belonged to the old Starter plan, Xero removed it on 14 September 2020 as announced by a Xero admin on their own community board, and Starter has not existed since 2024.

If I had taken that at face value I would either have walked away from the only product that fits, or been upsold from £7 a month to £39. Asked the same question directly a few messages later, it said the documentation does not specify any limit, which is the correct answer. It also, to be fair, declined to bluff on a couple of questions it could not answer, which is more than some humans manage.

The lesson is to check anything a support bot tells you about pricing tiers against the vendor's own pricing page and their community boards, because the boards are where the actual answers live. Xero's Product Ideas board is public without a login and was where several of the useful confirmations came from. If people are still requesting a feature there, it generally is not shipped yet.

Getting escalated to a human was possible in the end. You just have to ask for one directly, then send numbered questions, because a ticket reply will typically answer point one and ignore the rest.

The bit that annoys me most

I now have to pay to submit a tax return that was free before.

That is the part I cannot get past. There are free options, and one of them would probably do the job. But free means the quarterly farce: download PDF statements, get them onto the server, convert them, convert them again to CSV, work through everything, upload the results. Once a year that was tolerable. Four times a year, times two businesses, it is not.

Here is what I already have. Scripts on my own server that take my downloaded statements and give me a list where I tick which business each item belongs to. It works out the totals, and those totals go onto the paper return. That system replaced something considerably worse. Years back, working through the lot took me and a family member the best part of two weeks. Getting AI to do the heavy lifting brought it down to about two days.

And now the whole thing gets ripped up again because of digital tax.

Eight quarterly submissions, plus two year end returns. For a one man band. That is not modernisation, that is just more work handed to the person who was already doing it.

So my choice is: pay roughly £100 a year to automate it properly, or keep the money and accept an ongoing manual grind four times a year forever. If I am going to be forced to redesign how I do my tax anyway, I want to come out of it with something easier than I had, not harder. That means paying. It is not a great deal but it is the least bad one.

What really takes the biscuit is that the threshold runs on gross income before expenses. My money in and out is fairly high, my actual profit is not much at all. So I get dragged into all of this on the strength of turnover that mostly goes straight back out again. In other words, I am paying money out for the privilege of telling HMRC how little I made.

More cost, more time, more hoops, no benefit to me whatsoever. Whoever signed this off wants a good slap round the back of the head. It is time and effort that could be spent on almost anything more useful, which at this point includes staring at a wall.

And while I am at it

This is not really about accounting software any more, but it is all part of the same picture.

I am just about sick to death of trying to run a business in 2026.

It is not one thing, it is all of it at once. Cost of living has everyone watching every penny, which drives sales down before anything else even happens. Then there was the EU regulatory nonsense, which cost me months of my life and a pile of money to deal with, and the upshot is I lost the EU market anyway. Store sales have been more or less at rock bottom since, with the odd sporadic spike that is nice while it lasts and then gone again.

On top of that there are ongoing problems with PCB suppliers, parts that are difficult or impossible to get hold of, and the whole tariff circus between the US and China rippling through everything I try to buy. Every direction I look there is another obstacle that did not exist a few years ago.

And now this. More work and more money, purely to tell HMRC what has been going in and out of my bank account four times a year instead of once.

None of it makes anything better. None of it sells a single extra board. It is all just overhead piled on overhead.

Frankly, if I had known years ago that this is where it was all heading, I am not sure I would have started the Atari business at all. It is always the same pattern. Hassle up, expense up, paperwork up. Motivation down, profit down. At some point you look at the graph and wonder what exactly you are still doing it for.

Where I have landed

Xero Simple, about £101 a year, pending one thing I cannot settle by asking anyone: whether the bank connection actually holds. My last attempt at this with different software never would connect, so that is the first thing I will test and everything else waits behind it.

For anyone with one trade and no property, ignore all of the above. Clear Books free will do you perfectly well and costs nothing. For anyone with a trade plus a rental, most of them handle that. It is specifically two sole trades that turns a five minute decision into a fortnight.

Second bit of advice: check the free option before you spend anything, and check it against your actual bank rather than the vendor's list of supported institutions. A feed that will not connect is worse than no feed, because you have paid for it and you are still exporting statements by hand.

Anyway, filed here so that in eighteen months when the deadline is actually looming I can read this instead of doing the whole thing again from scratch. Which, given the state of my filing system, is optimistic of me.

One last thought before I commit to anything

Having written all that down, I am not quite ready to hand over money yet.

Clear Books is one feature away from being the free answer. It already runs two businesses under one login, it is HMRC recognised, and it covers quarterly updates and the year end return at no cost. The only thing wrong with it is that the two businesses are separate sets of books, so with one bank account feeding both you go through the same transaction list twice. All that is really missing is a tag on the transaction saying which business it belongs to, which is precisely what Xero does with tracking categories. Hardly rocket science.

So before I spend anything, I think it is worth giving Clear Books another go. Two reasons.

First, the Lloyds feed. It never worked for me last time, but that was a while ago and open banking connections get rebuilt fairly regularly. It might just work now. If it does not, I have lost an evening rather than a subscription.

Second, and honestly the bigger one for me: Clear Books has an actual support forum where real people answer, and where I can see what everyone else has asked. Xero has an AI chatbot that told me two things that were flatly untrue, including one that would have cost me an extra £32 a month if I had believed it. I cannot stand dealing with these things. Being able to post a question somewhere public and get a human answer, and read the answers other people got, is worth a fair bit on its own.

So the plan is: try Clear Books again first, ask on their forum about the two business setup and whether anything has changed, and see whether the bank feed behaves. If it works and the double handling turns out to be less painful than it sounds, that is nothing a year rather than a hundred. If it does not, Xero Simple is waiting and at least I now know exactly what I would be buying.

Either way I will report back.

Following on from my last post about Making Tax Digital software, I went back and gave Clear Books another proper look. It was the one I picked originally, it was the one I wanted to work, and I had written it up as the near miss on price. Turns out it is off the list entirely, and I think I finally understand what went wrong the first time round.

Posting this separately because it is a decent chunk of new information and because I suspect a fair few people are going to walk into the same trap.

The short version

The Clear Books free plan does not include automated bank feeds. If you want your bank and card transactions pulled in automatically, you need the Sole Trader plan at £5 a month plus VAT. That is roughly £72 a year including VAT, and it is per business. I have two sole trades, so that is about £144 a year.

Xero Simple is about £101 a year and covers both businesses in one subscription. So Clear Books now costs about £43 a year more, for more work. That is the end of it as a contender.

How I found out

I logged into my old account and went to add a bank account. Got this:

Bank accounts are only available on our paid plans. Click here to see your upgrade options.

Which was a surprise, because I had bank feeds working when I tried this before and I am fairly certain I never paid anything.

So I clicked through to the upgrade page to see whether feeds could be bought as a cheap add-on. One comparison site had claimed they were available for about £15 plus VAT a year, which would have made Clear Books the cheapest option going even with its other faults.

They cannot. The page offers exactly three plans and nothing else:

Sole Trader Free. Record income and expenses, MTD for Income Tax, basic reports. No automated bank feeds, no custom bank automation, no receipt capture from photos or PDFs, no invoicing, no journals, no balance sheet. Support is the community forum only.

Sole Trader, £5 a month plus VAT. Adds automated entry from bank feeds, custom bank automation, auto expenses from PDFs and photos, invoicing, quotes, journals, balance sheet, audit log, unlimited users, and dedicated email support.

Sole Trader + VAT, £15 a month plus VAT. As above plus MTD VAT reporting and dedicated phone support.

No add-ons. No bank feed as a separate line item. The £15 a year figure appears to be somebody confusing Clear Books with QuickFile, which genuinely does charge £15 plus VAT a year for feeds on top of everything else.

What I now think happened the first time

This is the bit that has been nagging at me.

When I first signed up I had bank feeds. Then they stopped working, and at the same time there was a live Lloyds outage thread on their community forum that stayed unresolved for months. I put two and two together, concluded the software was broken, gave up and went looking elsewhere.

But signing up to Clear Books puts you on a 30 day trial with the full paid feature set. Bank feeds included. When the trial ends you drop to the free plan, and bank accounts become unavailable.

So there is a very plausible version of events where the trial quietly expired, the feed stopped because the feature was gone rather than because the bank was down, and I spent the following weeks blaming an outage that may well have been fixed by then. Nothing told me my plan had changed. Nothing said "bank feeds are no longer part of your plan". It just stopped.

I cannot prove that is what happened at this distance and neither can anyone else. It may genuinely have been the outage all along. But it fits, and it is worth knowing that the two failures look identical from where you are sitting.

Either way, the practical lesson stands. If your bank feed dies in any of these products, check your plan before you check the bank.

Whether it changed or was always like this

I am not certain. I had it in my head that the free plan used to include feeds, and my own experience of having them without paying seemed to back that up. But that experience is explained just as well by the trial.

It is also possible they restructured the tiers when they built the current free MTD plan. That plan is being marketed hard at the April 2027 threshold change and they describe it as built for the new rules rather than bolted on, which does hint at a rework.

Either way the answer today is the same. Free means typing your transactions in by hand.

Why that kills it for me specifically

The whole point of this exercise was to stop doing quarterly bookkeeping by hand. Manual entry four times a year across two businesses is worse than what I do now, not better. So the free plan was never actually a contender, whatever it costs.

And the paid plan does not save it either, because Clear Books treats each business as a completely separate set of books. With one bank account feeding both trades, their own support's advice is to link the same feed into both businesses, explain the relevant transactions in each, and mark the other business's transactions as personal so the balances reconcile. That is going down the same transaction list twice, every quarter, forever.

So the choice Clear Books offers me is: pay nothing and type everything in twice, or pay £144 a year and click through everything twice. Against Xero at £101 for both businesses in a single pass, there is no argument left to have.

The road not taken: writing my own

Since I already run my own server and have scripts that do most of this work, the obvious thought was to skip the accounting software entirely and build it myself. Pull the bank data in one end, tag it with my existing tools, push the numbers at HMRC out the other.

Important caveat before anyone acts on this section: everything below came from AI research and I have not verified any of it myself. It looked plausible enough to talk about, and I am writing it down because it might be useful to someone, but treat it as a starting point for your own investigation rather than fact. I did not get far enough to test a single line of it.

On the bank side, the option that looked most promising was GoCardless Bank Account Data, which used to be called Nordigen. Reportedly a genuinely free tier: fifty bank connections a month, 720 days of transaction history, refresh up to four times a day, and it is supposed to be a properly licensed account information provider rather than screen scraping. It returns plain JSON per transaction with date, amount, merchant name and description, which would drop straight into what I already have. The reported catches are that the authorisation flow is fiddly the first time and access tokens are short lived, so you need refresh handling, plus the standard open banking ninety day reapproval.

If you did not fancy writing the connector at all, Actual Budget apparently documents that same connection officially and is self hosted, so you could let it do the pulling and export CSV from there.

On the HMRC side, the MTD for Income Tax API is publicly documented. The bit I could not get an answer on is whether an individual can realistically get production credentials to file their own return, or whether the approval process is built purely for commercial software vendors. That is the question that would decide the whole thing and I never got it settled.

In the end I talked myself out of it anyway. The entire point of this exercise was to stop maintaining a bookkeeping pipeline, not to build a more interesting one. Doing it myself would mean I own every breakage, at quarterly deadlines, forever, with HMRC penalties attached. For about a hundred quid a year somebody else can own that instead. If the API access question ever gets answered clearly in favour of individuals, though, it would be a genuinely fun weekend.

The bit that really gets me

Step back and look at what is actually happening here.

I am now expected to pay a private company every month, for the privilege of telling HMRC what has already gone in and out of my bank account. Something I have been doing perfectly well for nothing, on paper, for years.

And it does not stop there. All that subscription money going to Xero, Sage, Clear Books, Intuit and the rest is revenue that did not exist before this was mandated. Revenue that then gets taxed, presumably. I have not checked that and I am no accountant, so take it as my assumption rather than fact, but it certainly looks like the government has invented an entire industry out of thin air by legislating that everybody has to buy its product, and then takes a cut of the proceeds.

Nobody asked for this. It does not make my books more accurate. It does not make me pay a penny more or less in actual tax. It just moves money from thousands of small traders into a handful of software companies, four times a year, forever.

If it really is about the money, I would honestly take the deal directly. Charge me £20 a year to stay on paper and leave me alone. I would pay it tomorrow, HMRC would be up on the arrangement compared to what they get from taxing my £101 subscription, and I would get my evenings back. Everybody wins except the software vendors, which may be precisely why it will never be offered.

Credit where it is due

Clear Books is not a bad product and I want to be fair about it.

The free plan is genuinely free forever with no transaction limits, no time limit and no revenue threshold, and it does include MTD quarterly updates and the year end return with no upgrade wall at filing time. That is a real thing to offer for nothing, and for a sole trader with one trade who is happy entering transactions manually it is a perfectly good answer.

They also have actual humans. Email support on the £5 plan, phone support on the £15 plan, and a public community forum where staff answer and you can read what everyone else asked. After spending an evening arguing with Xero's AI chatbot, which confidently told me two things that were flatly untrue, I do not take that for granted.

It is one feature away from being the free answer for people like me. All that is missing is a tag on each transaction saying which business it belongs to, instead of two separate ledgers. Xero does exactly that with tracking categories and it is hardly rocket science. If Clear Books ever adds it, they would own this corner of the market outright.

Where that leaves me

Back to Xero Simple, about £101 a year, both trades, one pass. The only thing left to prove is whether the Lloyds feed actually connects, which is the same question that started all of this and the one nobody can answer for me. First month is free, so that is the test and it costs nothing to run.

Two things to take away if you are looking at any of this yourself.

One: check whether bank feeds are on the free plan or only the trial before you build a workflow around them. It is not obvious and it will not be explained to you when it changes.

Two: if a feed stops working, check your subscription before you start filing bug reports. It might not be broken. It might just be gone.

Right, one more dead end to write up while it is fresh, so nobody else wastes an evening on it.

Since I already run my own server and have scripts that parse my bank statements, the obvious question was whether I could skip the accounting software entirely on the data side. Pull the transactions in automatically, run them through what I already have, and only pay for something at the very end to actually file with HMRC. Or ideally not pay at all.

Short answer: no. Not for free, not in the UK, not without a commercial contract.

The first thing to understand: you cannot talk to your bank directly

I assumed open banking meant I could write a script that talks to Lloyds' API and pulls my own statements. It does not work like that.

Lloyds does run a proper developer portal with an open banking API, and there is a sandbox you can play in for free. But sandbox is fake data. To get production credentials you have to be FCA authorised or registered as an Account Information Service Provider, hold an eIDAS certificate, and be enrolled in the Open Banking Directory before the bank will issue you anything. That is a licence application with fees, ongoing compliance obligations and months of lead time. It exists for firms offering a service to the public, not for one bloke wanting his own statements.

So you have to go through an aggregator. Somebody who already holds the licence and rents it out.

What I found, and what happened to it

GoCardless Bank Account Data (formerly Nordigen). This was the famous free one. Genuinely free tier, fifty bank connections a month, 720 days of history, refresh four times a day, properly licensed. Half the tutorials and self-hosted budgeting tools on the internet still point at it.

It is closed. New signups were disabled around mid-2025 and the product is being wound down. If you find a guide recommending it, that guide is out of date.

Enable Banking. This is what everyone now recommends as the replacement, and on paper it is exactly right. Free "Restricted Production" mode where you link your own accounts and access only those. Self-serve signup with no sales call. Proper documentation, REST API, native SDKs, sessions valid up to 180 days. Regulated AISP.

Better still, the Firefly III project has a whole tutorial for it. Their words: Enable Banking offers a free restricted mode that allows you to access your own bank accounts without a paid subscription, perfect for personal finance management. There are active GitHub issues and recent release notes showing it being maintained.

I was fairly convinced. Then I looked at their own market documentation, which lists every country they cover.

Austria, Belgium, Bulgaria, Croatia, Cyprus, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden.

Thirty countries. No United Kingdom. Ireland is there, we are not.

So that is another hour gone. It looked genuinely promising right up until the last click.

Everything else. Teller offers 100 free live connections but is US only. SimpleFIN is about $15 a year and also US only. TrueLayer, Tink, Yapily and Plaid all cover UK banks properly, but production access is entirely sales-led, with contracts, minimum monthly commitments and typically four to twelve weeks of onboarding. Their public pricing pages have quietly vanished, which tells you everything about who they are aimed at.

The only survivor I found is open-banking.io, which claims UK and EEA coverage across 2,600 plus banks for about €3 a month, no certificates required, credential-based rather than the usual OAuth dance. It has recently been added as a provider to the Firefly III data importer. That is the least established of the lot and I have not tested it, so treat it as a lead rather than a recommendation.

The self-hosted container route, for completeness

If you are in one of the countries that is covered, the setup is genuinely nice and worth knowing about.

Firefly III is a self-hosted personal finance ledger, and the Firefly III Data Importer is a separate container that feeds it. Both are official Docker images and actively maintained, with the importer currently on v2.3.x and recent releases full of provider-specific fixes. The importer is deliberately kept separate from the main application for security reasons.

You never hand your bank login to any of it. Open banking redirects you to your bank's own login page, you authorise there, and the aggregator receives a token. Nobody except the bank sees your credentials, and with a self-hosted setup your transaction data stays on your own machine.

One thing worth being clear about, because I misread it myself at first: the CSV import option in Firefly III is an input, not an output. It does not download statements for you and hand you a CSV. You download the statement from your bank by hand and upload it. That is the fallback for when you have no bank connection, not automation. If you already have scripts that parse statements, CSV import is a sideways move at best.

Where that leaves it

The free indie open banking tier for UK accounts does not really exist. The genuinely free options are all either wound down or scoped to countries that are not this one, and every provider that does cover Lloyds properly wants a commercial agreement and a sales conversation.

And even if I had solved the bank end, the other half was never settled: whether an individual can realistically get production credentials for HMRC's MTD API, or whether that approval process is built purely for commercial software vendors. I never got an answer to that, and it would decide the whole thing.

So the DIY route dies on both ends. Which, annoyingly, makes paying somebody about a hundred quid a year to deal with Lloyds and HMRC on my behalf look considerably more reasonable than it did when I started.

Writing it up anyway. Someone else is going to go down this exact path, find the same enthusiastic tutorials pointing at a service that shut two years ago, and end up where I did. At least now there is a signpost.

Update on the Making Tax Digital software saga. I actually signed up to Xero and tested it properly, and I have come to the conclusion that it would make my life harder rather than easier. Writing up why, because the reasons are not what I expected going in.

First, credit where it is due

Everything I was told would work, worked. I need to say that clearly because the rest of this post is critical.

The Lloyds bank feed connected first time, from a personal account, and pulled in all three of my accounts including the credit card. That was the one thing nobody could answer for me and the thing that had killed my previous attempt with different software. It just worked.

Tracking categories, which is the mechanism for running two sole trades in one subscription, are present on the cheapest £7 a month plan. Nobody could tell me that either. Not the pricing page, not the support chatbot, not their community boards. It is there.

Bank rules can set a tracking category, and they can be scoped to a single bank account. I set up a rule matching POST OFFICE and it fired correctly on the very first transaction I looked at.

So on paper, every single requirement I set out was met, for about 84p a month including VAT during the introductory discount.

And then I actually looked at the interface

Here is the reconciliation screen. This is where you spend your time.


Look at what that is showing. Three transactions. Three. Taking up the full width and height of the screen.

Each transaction gets multiple lines of interface. The bank line on the left, then a whole panel on the right with Match, Create, Transfer and Discuss tabs, a Who field, a What field, a Why field, a Business dropdown, a Tax Rate dropdown, an Add details link, a View details link, an Edit rule link, a Don't apply rule link. And an OK button.

All of that could sit on one line. Date, description, amount, category, done. Instead it is spread over half a dozen rows per transaction, so you scroll and scroll for very little actual information.

The killer: every single line needs approving

This is the bit that finished it.

Even when a bank rule has matched a transaction perfectly, and has already filled in the account code and the business it belongs to, you still have to click OK on it. Individually. One transaction at a time.

I looked for a select-all or a bulk confirm. There does not appear to be one. Compact view just makes everything slightly smaller, it does not change the number of clicks.

My postage account alone had 55 transactions waiting on connection, and that was only what the feed had pulled so far. I do two posting runs a week, so that is well over a hundred lines a year on the simplest account I have, every one needing an individual click on something that is already fully automated and needs no human judgement whatsoever.

And that is the easy account. The current account and the credit card are considerably messier.

I understand why it works this way. For a proper accountant, reconciliation is the step where you check the bank against your invoices, and going line by line is the entire point. Fine. But I have no invoices to check against. For my purposes it is a rubber stamp exercise on data that is already correct.

There is an Auto-reconcile add-on at £3.50 a month, which is AI matching. I do not know whether it removes the clicking or just improves the suggestions, and at ten times the cost of the base plan during the discount period I was not inclined to find out.

Why this matters more to me than it might to you

I have RSI in my hands. Hundreds of clicks and a lot of scrolling is not a minor annoyance, it is the deciding factor. A layout that wastes vertical space and demands a click per row is genuinely the wrong tool for me, however good it is underneath.

The comparison that actually decided it

What I do now: log into the bank, download the statements, roughly fourteen file clicks across the postage and electronics accounts, upload to my server, run my scripts. Everything comes out totalled up in a page I built, laid out how I want it, one line per transaction.

What Xero offers: the download step disappears, which is genuinely nice, and then I click OK several hundred times through an interface that scrolls forever.

That is not a saving. That is moving the work somewhere less convenient and paying for the privilege.

And the 90 day thing seals it. Open banking connections have to be reauthorised every three months. So I still have to log into my bank quarterly regardless. The single thing the direct connection was supposed to save me, I still have to do anyway. At that point I may as well download the statements while I am in there.

On direct bank access, which I went round in circles on

I spent a good while investigating whether I could just pull my own bank statements automatically with an API call and skip the accounting software entirely. Short version: no.

You cannot talk to your bank directly. Lloyds runs a full open banking API, they have a developer portal, the whole thing works. But to get production access you need to be FCA authorised or registered as an Account Information Service Provider, hold an eIDAS certificate, and be enrolled in the Open Banking Directory. That is a licence application with fees and months of lead time, aimed at companies offering a service. There is a sandbox you can play in, but it returns fake data.

So you have to go through an aggregator who already holds the licence. And there:

GoCardless Bank Account Data, formerly Nordigen, was the free one everybody recommended. It is closed to new signups and being wound down. Half the guides on the internet still point at it.

Enable Banking is the replacement everyone points to now, and it does have a genuinely free tier for accessing your own accounts. Their marketing shows a Union Jack among the supported countries and their bank list does include the UK. It covers exactly six UK banks. Barclays, HSBC, NatWest, RBS, Coutts and ABN AMRO. Lloyds is not one of them, and Lloyds is the largest retail bank in the country.

Teller and SimpleFIN are US only. TrueLayer, Tink, Yapily and Plaid all cover Lloyds properly but production access is entirely sales-led, contracts and minimum commitments, and their public pricing pages have quietly disappeared. The only remaining option I found claiming UK coverage was open-banking.io at about €3 a month, which I have not tested.

So even the best case is a monthly fee to read my own bank statements. Which strikes me as slightly mad in 2026.

That is the thing that really gets me. All this regulation exists specifically to make bank data portable and accessible. It has been enormously successful, 17 million users, billions of API calls a month. And there is no provision whatsoever for an individual to simply read their own account programmatically. You are either a licensed business or you are downloading PDFs like it is 1998. There is no middle option and nobody appears to be working on one.

Where I have ended up

Going round in circles looking for a simple, efficient solution that as far as I can tell does not exist.

So the plan now is to keep my own scripts, which already work and are laid out the way I want, and find a cheap or free bridging tool purely to push the final numbers at HMRC. There are several: My Tax Digital is free, MTDSorted Lite is free, AffordableMTD is free until April 2027 then £19.99 a year, MTD 4 Income Tax is about £20 a year. The question for all of them is whether they handle two separate sole trades, which is the thing that eliminated nearly every full accounting package I looked at.


The moral, if there is one

Test the thing before you buy it. Everything about Xero checked out on paper and it still turned out to be wrong for me, for reasons that no feature list, review or support answer would ever have told me. The deciding factor was the shape of one screen.

Cost of finding that out: about a fiver and an evening. Cost of not finding out: a year of subscription and a lot of unnecessary clicking.


The genuinely annoying bit

There goes the best part of a day, and where I have landed is this: I have to pay a third party company for the privilege of telling HMRC what I owe them.

Read that back. Paying for software to pay my tax bill. The world has gone quietly mad.

And the thing that really grates is that I could write this myself. I have the scripts, I have the server, the bank has a perfectly good API sitting right there. What stops me is not capability, it is layers of rules and regulations that exist for perfectly sensible reasons and which, taken together, mean an individual is simply not allowed to read his own bank account programmatically. Every route out is blocked by a licence I will never hold or a contract aimed at companies a thousand times my size.

Lloyds are not helping either. You can export a whole year of CSV data for a current account, but the credit card only goes back six months. No idea why. Same bank, same login, same export screen, different rules. Still, given I am now going to be doing this every quarter anyway, six months of CSV is more than enough for a quarterly run. So I may switch to CSV for the card as well and drop the PDF parsing entirely, which would be one less fragile thing in the chain. The PDF format changed under me once already this year and silently stopped extracting half the transactions, so getting rid of it is no bad thing.

Small consolation for what this has all cost in time.

The wider picture is what it is. There has been pushback on Making Tax Digital for years and it has changed nothing, because the people it lands hardest on are the ones with the least voice. Whatever the stated aims, the practical outcome is that a lot of very small traders now have to buy software they never needed, four times a year instead of once, and a handful of software companies do rather well out of it.

For me that is a few days a year of my life spent on paperwork that produces nothing, sells nothing, and changes my tax bill by exactly zero pence. Days that could go on designing boards, fixing the server, or frankly anything at all with a point to it.

The small operator absorbs it, as usual, because there is no alternative on offer.

Right, final instalment on this. I cancelled the Xero subscription.

What I was actually trying to buy

Worth restating, because it was never complicated. I wanted something that would pull my bank and credit card transactions in automatically, let me tick which of my two businesses each one belonged to, and file the quarterly updates. My own scripts already total everything up. The only thing I was paying for was an end to downloading statements by hand every quarter.

The software genuinely does the first part. The Lloyds feed connected first time, personal accounts and credit card, and the tracking categories and bank rules did exactly what I needed for two sole trades on one subscription. Credit where it is due.

Then it hands all the manual work straight back at the other end. Every transaction needs an individual click to confirm, even when a rule has already matched it and filled everything in correctly. No bulk confirm that I could find. So the one job I wanted done gets done, and a few hundred clicks a year appear in its place. That is the opposite of what I was buying.

I have RSI in my hands, so that is not a minor irritation, it is the whole decision.

The cancellation, which was its own adventure

I went to cancel. It would not let me log in without two factor authentication.

Fine, except my tablet had nothing on it that could handle it. I tried the QR reader, which cheerfully announced it was opening a browser and then failed, because a 2FA code is not a web link and a QR reader has no idea what to do with it. So that meant installing an authenticator app I did not previously want, on a device I did not previously need, purely to close an account I had opened that morning.

Let that sink in. Two factor authentication, an app install and a scanning rigmarole, in order to stop paying somebody 70p a month.

And then Lloyds moved the goalposts as well

Somewhere in the middle of all this I discovered Lloyds have quietly changed the layout of the credit card statement PDFs. They have added a Card Ending column, so every transaction line now starts with the card number instead of the date.

My parser has been silently failing on it. Not erroring, not warning, just extracting fewer transactions and looking perfectly happy about it. April and May came out fine and then it just stopped, and I would never have noticed if I had not been staring at the output for other reasons.

Another couple of hours with AI to work out what had changed and patch it. All of that just to be in a position to start last year's tax return, which I still have not actually begun.

Two days

That is where I am. Two days gone on this. Not doing anything productive, not designing anything, not fixing anything that needed fixing. Two days of jumping through hoops trying to find something that fits how I already work, and ending up back where I started.

And it is all of it, not just the tax. Every single thing now comes with an account, a password, a verification code, a text message, an app you have to install, a QR code to scan, a session that times out, a portal, a chatbot standing in front of a human. Every one is small on its own. Together they eat days.

Trying to run what is basically a hobby business in 2026 is a colossal joke. Not because of the work, the work is fine. Because of everything bolted onto the side of the work.

Where I am leaving it

I am doing no more on this. It has taken enough.

My scripts work, they are actually in better shape than they were two days ago, and I know which free tool I will use when the time comes. That is enough to be going on with.

From here I wait until HMRC write and tell me I have to change. On my figures that is most likely April 2028, which is a good while away. No point signing up voluntarily and starting the clock early, and every point in leaving it alone.

Two years is a long time. Something better may turn up. The software might improve, someone might build the simple thing that does not exist yet, or the rules might shift again as they have several times already.

Or, at the rate the store is going, my turnover will have dropped far enough that I do not meet the threshold at all and the whole problem solves itself. Not exactly the outcome I would choose, but it would at least save me the clicking.

What I actually wanted was simple: something that would pull my bank and credit card transactions in automatically, let me tick which of my two businesses each one belonged to, and file the quarterly updates. I already have my own scripts that total everything up, so the only thing I was trying to buy was an end to downloading statements by hand every quarter.

The software itself works well and the bank feed connected to my personal Lloyds accounts first time, including the credit card, which I was not expecting. Tracking categories and bank rules on the Simple plan do exactly what I needed for running two sole trades from one subscription.

The problem is the reconciliation workflow. Every transaction needs an individual OK click, even when a bank rule has already matched it and filled in the account code and tracking category correctly. There appears to be no way to bulk confirm rule-matched transactions. One of my accounts alone generates well over a hundred lines a year that need no human judgement whatsoever, and I would still be clicking OK on every one. I have RSI in my hands, so hundreds of clicks and a lot of scrolling is a genuine barrier rather than a minor irritation. The reconcile screen also uses a great deal of vertical space per transaction, which makes it worse.

So the one job I wanted done, getting the data in without manual work, is done, and then a large amount of manual work is added straight back at the other end. It ends up being more effort than downloading the statements myself, which is the opposite of what I was paying for.

Just as off-putting was that I could not ask a single question before registering. There is no phone number, no contact email, and the contact page only lists an address for job applicants. Every route says register first, then raise a case. So the only way to find out what I was buying was to buy it. This entire signup and cancellation could have been avoided by a five minute conversation up front.

When I did get in, I landed on an AI chatbot. It told me the Simple plan was formerly the Starter plan and had a 20 bank transaction monthly reconciliation limit. Both were wrong. Starter was retired in 2024, and that limit was removed from Starter back in September 2020. Had I believed it I would either have walked away from the only plan that suited me or been upsold to a plan three times the price. I had to check its answers against your own pricing page and community boards to work out which were true.

For a decision involving my tax compliance, having no way to reach an actual person to discuss things is a serious problem, and it is the main reason I would not recommend Xero to anyone in my position.

So I am staying with my own scripts and a bridging tool for the HMRC submissions.

Honestly, I think there should be a law that every web designer, software developer and app builder gets a smart tap on the hand with a hammer for every click and every scroll they inflict on the rest of us. They would learn very quickly how to build a decent interface. Most of them seem to be teenagers who can scroll and tap all day long and think nothing of it, whereas I just want the job done in the least time possible so I can get on with something constructive.

We are supposed to be living in an age where things get easier and simpler, and instead every single thing gets more complicated and eats more of the day, especially anything to do with running a business.

Case in point: paying my tax bill last year. The website was so peculiar that after wading through page after page I got dumped out at the end with no confirmation, no receipt, no indication whatsoever whether the payment had actually gone through. So I did it again, to be safe. Of course it had worked the first time, and the money went out twice. I have not had the duplicate refunded, and I am hoping it gets knocked off this year's bill, assuming there even is one, which on current trading looks optimistic.

Why on earth can it not be one single page? Reference number, amount, pay. That is it. That is the whole transaction. Instead it is a dozen screens of hoop-jumping that takes an hour and leaves you unsure whether it worked. Total madness, every single time, and I cannot begin to describe how sick to death of it I am.

And the real kicker is that just as you finally learn one of these systems, they redesign the whole thing and you start again from scratch. People have actual lives to be getting on with. This is all absurd.

A question about the 45p a mile mileage rate, because I think I am missing something

Separate topic from the software saga, but it came out of the same evening's digging and I would be interested if anyone can tell me where my reasoning falls down.

HMRC's simplified mileage rate is 45p a mile for the first 10,000 business miles, then 25p after that. The important bit, which I had not appreciated until now, is what that 45p is supposed to cover. It is not a fuel allowance. It is meant to cover the entire cost of running the vehicle for business: fuel, insurance, road tax, servicing, repairs, tyres and depreciation. All of it, in one flat figure per mile.

The alternative is claiming actual costs, where you work out what the vehicle really costs and claim the business proportion. You cannot mix the two, and once you start using the flat rate on a vehicle you cannot switch back to actual costs for that vehicle later. So it is worth getting right first time.

Here is what bothers me.

I tested this myself some years back, out of curiosity rather than for tax purposes. Out of my way on a dual carriageway at a steady 50, I get somewhere around 50mpg. Across town, stop-start, first and second gear most of the way, I have measured it as low as about 15mpg. Not 30, not 25. Fifteen.

Diesel is currently 186.4p a litre on the government's own weekly average, so a gallon is about £8.47. Run 50 miles:

At 50mpg that is one gallon, £8.47 of fuel. The flat rate would pay £22.50, leaving £14 for everything else.
At 15mpg that is 3.3 gallons, £28.25 of fuel. The flat rate would pay £22.50. That is nearly £6 short before a single tyre has been paid for.

So on a steady run the rate pays well over twice the fuel cost, with plenty left for the servicing and depreciation it is also meant to cover. On town work it does not even cover the diesel, never mind anything else.

And town work is exactly the sort of driving I do. Van with a machine and a load of water in the back, and around here it is uphill on the way there and somehow uphill on the way back as well. Foot to the floor in second or third grinding up hills, revs up, going nowhere fast, for miles at a time. Genuinely, I have gone ten miles out of my way on a main road and used less fuel than going the direct route through the middle of town.

Worth adding that the 45p and 25p rates have not changed since 2011. Diesel has gone from around £1.35 to £1.86 in that time, and it is still climbing: up nearly 19p a litre since mid-July, with fuel duty due to rise 3p in January and another 2p in March, then rising with inflation from April.

For what it is worth, this is not a new complaint. I remember a conversation years back with one of the site engineers where I used to work, blokes who spent their lives driving all over the country. He reckoned he had worked out the true cost of running a vehicle at somewhere around £2 a mile, and that was twenty years ago. That was his own calculation rather than any official figure, and I cannot verify it or remember how he arrived at it, so take it as a point of reference rather than a number to rely on. But if someone doing that mileage for a living thought the real cost was several times the approved rate even back then, it does rather suggest the flat rate has been on the mean side for a very long while.

So my question is this. A flat national rate per mile appears to reward exactly the kind of driving that costs least and penalise the kind that costs most. Somebody doing motorway miles in an economical car does very nicely out of it. Somebody in a loaded van crawling round a town centre is effectively subsidising them, and increasingly so as fuel gets dearer against a rate that has been frozen for fifteen years.

Am I missing something? Is there a reason it works this way that I have not grasped, or is it simply that a single national figure was easier to administer than anything more sensible and everyone just lives with it?

Interested to hear from anyone else running a van about which route they went and why.

The Xero cancellation is still pending likely will not fully cancel until the next billing period next month.

Keep getting tons of junk email from them. Clicked on the unsubscribe link in one of the emails.. And of no great surprise...


I mean Evilbay haven't fixed links in their own emails for over a decade now. Dissolve pointless them even sending out emails when they don't work anyway. The links just go nowhere other than page not found.

So then I was hunting round Xero interface and jumping through the login hoops again going around pages which did not help or simply did not load at all.. About 10 minutes later found a preferences language seems to be well hidden away where I can actually unsubscribe from all the junk !

Like I keep saying, everything is just a x10 more efffort than it needs to be these days. These companies cannot even maintain their own infrastructure properly any more.