The productivity story everyone is talking about
Nobel Prize winning economist Robert Solow famously said in 1987: “You can see the computer age everywhere but in the productivity statistics.” More recently, investment in B2B SaaS in the UK has continued apace – in Q1 2025, SaaS companies secured £1.67b, a 41% increase from Q4 2024, and across 2024 they raised £5.04b, up 20% year on year.
Yet since 2019, despite the emergence of AI and much else besides, UK productivity improvements have been tepid at best. The Office for National Statistics’ latest flash estimate shows output per hour only slightly above its 2019 average, and the Office for Budget Responsibility’s March 2025 outlook bakes in a subdued trend for the years ahead. What can possibly be holding productivity back?
We can’t claim to know the answer, but in our neck of the woods, we see clear evidence of productivity drains that counter-intuitively increase the most software applications businesses invest in.
Where the hours really go
Let’s zoom in. In banks, fintechs and in the many non-financial brands that handle payments, lending or payouts, teams still spend chunks of the week hopping between systems – re-keying data from a procurement tool into a bank portal, copying identifiers from a card-scheme dashboard into a CRM, reconciling a PSP settlement file against a general ledger export. The swivel-chairing exists precisely because finance sits next to the tools people use, not inside them. When money movement, account data and rules live outside the workflow, every approval, exception and audit step spawns another tab – and more time.
Consider three familiar scenes. A chargebacks analyst assembles case evidence with a timer ticking, but the documents live across a ledger, a scheme terminal and email. An onboarding specialist reviews a sanctions hit, yet the context that clears the match is marooned in another tool, so the investigation restarts from scratch. A treasury operator triages statement breaks on Friday afternoon because two systems round timestamps differently. None of these is a dramatic blocker. All of them are a slow leak of attention and hours.
A conservative model of the swivel-chair tax
Two anchor facts set the scale. OECD-wide employment stood at 662 million in May 2024, and average labour productivity across the OECD was USD 67.5 per hour in 2022. These figures are published by the OECD in its employment release and its productivity analysis.
A knowledge-worker lens keeps the scale clear. OECD employment stands at 662 million; if knowledge roles account for roughly 70% of jobs in the target firms, that is about 463 million people. Valuing time at the OECD’s average GDP per hour of USD 67.5 and using a realistic 43-week working year, a saving of 43 minutes per knowledge worker per week returns about USD 964 billion in output annually. This sits well below published upper bounds for the digital context-switching penalty, with field research summarised by Harvard Business Review indicating close to four hours a week lost to app switching.
Even under a stricter definition, the conclusion holds. On Eurostat’s measure, high-skilled occupations account for around 44% of EU employment; applying that filter, the same USD 964 billion follows from saving roughly 68 minutes per high-skill worker per week — still only a fraction of the observed four-hour toggling ceiling and entirely proportionate to finance-adjacent work. See Eurostat for the occupational split and the OECD’s productivity reference in its Compendium.
Why embedded finance automation changes the plot
If the problem is fragmentation, the fix is not another dashboard – it is making the money layer programmable inside the tools where the work begins. That is what embedded finance does when it is wired for automation rather than just connectivity. When payments, accounts and compliance checks are exposed as first-class, rule-driven actions inside the workflow, entire hops disappear. A marketplace payout can be initiated, approved and posted from the procurement surface without a CSV ever touching a desktop – with bank-confirmed status written back automatically. Statement reconciliation stops being a hunt through filenames because references and counterparty data are born together with the payment, so matching is by ID, not by guess. KYC and sanctions context follows the entity wherever it appears, so a reviewer is not rebuilding the case file in a new screen. The gains show up as fewer errors, cleaner audit and faster cycle times – but what they are really buying back is attention.
Bridging the firm-level fix to the national debate
The macro backdrop is stubborn. UK output per hour is only slightly above its pre-pandemic level and the fiscal watchdog has pencilled in a cautious path for the years ahead. Against that context, a practical way to lift firm-level productivity is to remove thousands of these micro-frictions in financial operations. This will not, by itself, move the national numbers overnight – but it is a credible, measurable contribution that leaders can deliver this quarter while bigger reforms grind through the system.
Bottom line
SaaS investment by business continues apace. AI promises more results with less human effort. But as long as these tools are locked out of the financial domain, the gaps will continue to sap productivity. Embedded finance offers the means to join things up completely, so the full potential of automation can be unleashed. And thereby, at least for knowledge workers, deliver that elusive growth in productivity.
Method note: This article uses a single lens – the output lens – valuing time at the OECD’s GDP per hour worked. The model scales a saving per worker per week over roughly 43 working weeks. With the full OECD workforce, 30 minutes saved per worker per week lands near USD 0.96tn. Restricting to knowledge workers at 70% of employment requires about 43 minutes per worker per week to reach USD 964bn. Using the EU’s measured high-skill share of 44% implies roughly 68 minutes per worker per week for the same total. Inputs are from the sources hyperlinked above.
