The Sunday Shortlist
UK HR software: Five months, the same shortlist.
Measured June to October 2026; snapshot 28 September to 25 October 2026. Monitoring sponsored by seedli.ai.
On this page
- What we measured
- The judging sheet
- The verdict: three suites, then everyone else
- Five months, the same shortlist
- Compared, not chosen, then kept
- One company, many names
- When the model changes, the long tail moves
- Nobody owns a criterion
- What claiming it looks like
- Getting the page cited beyond your own site
- The lesson
- Sources
- The Sunday Shortlist
The Sunday Shortlist decodes how AI makes decisions about a market category and audience.
This week we analyse HR software for UK mid-market companies
We unpack the questions UK HR leaders ask AI when they choose an HR platform for a growing company, the shortlist that comes back, and the criteria that decide it, across all four stages of the decision journey: from first comparison to who they are told to stay with and recommend.
This market has been measured every day since April. The surprise is how little has moved. Since June, the same three names have held the top of the final recommendations, month after month, in a market full of challengers.
What we measured
From 28 September to 25 October we ran this market's buying questions through the models. Gemini, Claude and ChatGPT answered the comparison and the final choice: 1,125 answers comparing providers and 1,056 making the final choice. The loyalty stages (retention, advocacy) rest on 174 and 139 answers. For the trend, we compare only months in which the models and questions stayed the same: June to October for the final choice.
The buyers' questions are about replacing what they have, without legal risk:
- "What modern HR platforms can replace our legacy HRIS system?"
- "Are there HR platforms designed specifically for UK compliance requirements?"
- "What security and compliance controls are built in for UK employment law and GDPR?"
- "How long does it typically take to implement an HR system?"
The judging sheet
When a buyer asks a model to choose, the model behaves like a judge: it applies criteria, checks each provider's evidence against them, and cuts the ones that fail. All three models use the same ten criteria here. The heaviest are cost, expected outcomes, product fit, regulatory safety, and trust and reputation. The eliminators are specific to the UK:
- Expertise, the hardest cut: "Vendor lacks UK employment law knowledge or HR transformation implementation experience."
- Capability: "Platform cannot handle UK payroll, pension auto-enrolment, or statutory leave management."
- Compliance: "UK GDPR and employment law violations expose company to ICO fines and legal liability."
The test underneath all three: can the vendor prove it knows how employment works in the UK, not just how HR software works?
The verdict: three suites, then everyone else
| Name | Evaluation | Decision | Retention | Advocacy |
|---|---|---|---|---|
| Workday | 84.4% | 99.2% | 100.0% | 100.0% |
| SAP SuccessFactors | 78.7% | 87.9% | 85.6% | 98.6% |
| Oracle HCM* | 60.2% | 47.0% | 19.5% | 35.3% |
| Personio | 52.2% | 12.8% | 83.9% | 69.8% |
| BambooHR | 35.8% | 10.9% | 81.0% | 65.5% |
| UKG* | 24.5% | 1.0% | 1.7% | 2.2% |
| ADP* | 8.4% | 24.1% | 56.9% | 74.8% |
| Ciphr | 19.8% | 2.2% | 23.0% | 6.5% |
| Sage* | 16.4% | 9.4% | 39.1% | 12.2% |
Share of answers that name the company at each stage. * = the models use several names for this company (see below); an answer naming two of them counts once. Retention and advocacy rest on 174 and 139 answers, so those columns move in coarse steps.
Workday is the default: named in 99.2% of the answers that make the final choice. SAP SuccessFactors follows at 87.9% and Oracle's HR suite at 47.0%. The buyers describe themselves as growing, mid-sized UK companies; the shortlist they get back is led by the global suites.
Five months, the same shortlist
Watch the final choice month by month, and almost nothing happens:
| Decision share | Jun | Jul | Aug | Sep | Oct |
|---|---|---|---|---|---|
| Workday | 99.2% | 99.4% | 99.1% | 99.3% | 99.2% |
| SAP SuccessFactors | 88.9% | 90.6% | 90.1% | 86.9% | 87.4% |
| Oracle HCM Cloud (main name only) | 37.1% | 39.6% | 40.5% | 36.8% | 34.9% |
| Personio | 12.5% | 11.7% | 10.5% | 14.6% | 12.7% |
| BambooHR | 11.9% | 10.9% | 10.7% | 13.3% | 10.7% |
Five months of daily answers, and no name in the table has moved by more than about four percentage points. The one name trending upward is ADP under its own name, from 11.9% in June to 16.9% in October, with a dip in September. For every challenger, the lesson is the same as the table: this answer will not drift your way on its own.
99.2% → 99.2%
Workday's share of final recommendations in June, and in October. In between: never below 99.1%.
Compared, not chosen, then kept
The challengers have a pattern of their own. Personio is named in 52.2% of the answers comparing providers, falls to 12.8% at the final choice, and then reappears in 83.9% of the answers about which platform to stay with. BambooHR does the same: 35.8%, then 10.9%, then 81.0%. ADP runs the other way, from 8.4% when buyers compare to 74.8% when they are asked who they would recommend.
One plausible reading: the models know the challengers well enough to compare them and to call them easy to live with, but not well enough to stake the final recommendation on them. At the moment of choice, the models fall back on the safest-looking name.
One company, many names
The models also split companies into several names, and in this market it is the large vendors that pay for it. Oracle's HR suite appears as Oracle HCM Cloud, Oracle Cloud HCM, Oracle Fusion Cloud HCM and Oracle Fusion HCM: the main name alone holds 50.7% of comparison answers, all four together 60.2%. Sage appears under seven names. UKG appears under four, including Kronos, a name that Kronos and Ultimate Software retired when they merged and became UKG in 2020.
Every variant is a separate entry in the models' memory, and a share split across entries is weaker than the same share under one name. Declaring which names belong to you is the job of Organization schema with alternateName and sameAs.
When the model changes, the long tail moves
The same five months also show what happens when the model behind the answers changes. In April and May, ChatGPT's answers came from gpt-4o-mini, and two names appeared regularly: Bullhorn, a recruitment platform, and "Cezar", which looks like a garbled form of the UK vendor Cezanne HR. When gpt-4o-mini was replaced by gpt-5-mini in late May, both disappeared from every stage and have not returned. Nothing happened in the market. The model changed.
For a vendor in the long tail, that is the real risk: visibility built on how one model happens to answer can vanish with the next model version. Visibility that holds across models is worth more than visibility in one.
Nobody owns a criterion
Ten criteria decide this market. Not one has an owner: no vendor is the name the models reach for on UK employment law, UK payroll and pensions, implementation experience or data protection, the questions that eliminate hardest.
These are not questions a challenger needs a global brand to win. They are questions of documented UK expertise: which legislation the platform handles, how auto-enrolment and statutory leave work in it, how many UK implementations of what size have been completed.
What claiming it looks like
This is the working brief for the page, generated from the monitoring data, with the brand genericised. The frame is "publish the method": the hardest cut in this market is missing UK expertise, and the answer is a page that shows how the vendor's implementation handles UK law, step by step, in a form the models can quote. For how to build this kind of page, see the How it works playbook.
Title: "How [our platform] delivers UK HR transformation: the method"
The position the page takes: UK employment law knowledge belongs inside the implementation method, not alongside it as an optional service. When it sits outside the delivery team, statutory gaps build up through every phase and only surface at an audit or a tribunal claim. Defensible because every claim on the page points to a documented phase, a named role and a checkable record.
The intent family the same page must also answer (one prompt spawns two to three searches behind the scenes):
- "Does [our platform] understand UK employment law?"
- "How long does a [our platform] implementation take in the UK?"
- "Who supports us during and after go-live?"
The structure: five H2s, each a claim the models can lift as a standalone answer:
- "Why UK employment law knowledge must be part of the implementation method, not a separate service"
- "Inside our UK implementation: what happens at each phase"
- "How we validate UK statutory compliance before and after go-live"
- "What our implementation team brings to complex UK projects"
- "How to assess any HR vendor's real UK expertise before you sign"
Key Takeaways for the top of the page (each a self-contained claim a model can cite):
- "UK law expertise is part of every implementation: [which statutory areas the team covers, e.g. Working Time Regulations, TUPE, right to work]."
- "Each phase ends with a compliance checkpoint: [your actual phases and what is checked at each]."
- "UK clients work with [the role that owns the project] from [when], through go-live."
- "Across [number of projects] UK implementations over [period], [on-time rate] went live on schedule."
Key Snippet, placed early: "[Our platform]'s UK implementations follow [number] phases with a compliance checkpoint at each; across [number of projects] UK projects since [year], [on-time rate] went live on schedule."
Slug: /uk-hr-implementation-method, the buyer's question shape, not a content-type label.
Kept alive: the project numbers updated each quarter, the statutory list reviewed whenever UK employment law changes, the review date shown on the page.
Two notes. First, every bracket is a fact from your own delivery records, including the method itself: describe the phases, roles and checkpoints you actually run, not a generic sequence, because the page is only worth quoting if a buyer can hold you to it. Second, the last H2 is the counterintuitive move that works: by giving buyers four questions to ask any vendor, including you, the page becomes the checklist the models quote when the expertise question comes up. Ask: where UK law knowledge sits in the team and whether it is billed separately; which compliance checkpoints the method contains; who owns the project from day one; and the track record of on-time, compliant go-lives for companies of the buyer's size.
Getting the page cited beyond your own site
In this market the models cite institutions and analysts, not vendors: the CIPD, the Information Commissioner's Office, The Pensions Regulator, Acas, HMRC, XpertHR, HR Magazine, Gartner, Forrester, G2 and the Recruitment and Employment Confederation. No vendor site appears among them.
The practical consequence: a vendor's UK expertise counts when it is visible where those sources look. That means CIPD and XpertHR content that names the vendor's approach, analyst and review profiles that describe UK payroll and compliance specifically, and a page on the vendor's own domain that documents its UK legislation coverage with dates. ChatGPT and Claude lean on text: structured pages, registers and review platforms. Gemini rewards fresh, dated material tied to the questions buyers actually ask.
The lesson
This is a remarkably stable market: five months of final answers, the same three names on top, and challengers that buyers compare, but the models do not choose. Stability is not safety, though. The long tail showed how quickly a name can disappear when the model changes.
Your category has its own version of the UK-expertise question. The vendor that documents it first gives the models something better to fall back on than a famous name.
Sources
The measurement behind this edition: window 28 September to 25 October 2026, evaluation and decision resting on 1,125 and 1,056 answers from Gemini, Claude and ChatGPT, retention and advocacy on 174 and 139; month-by-month figures from June to October 2026, months with model or question changes excluded. Criteria, eliminators and buyer language extracted per model per run; the full method is documented at how Seedli measures AI visibility. The UKG merger and rebrand verified against the companies' own announcement, linked above.
The Sunday Shortlist
Get the decode, and the toolkit
One step left. A confirmation email is in your inbox. Click it, and your toolkit downloads.
That didn't send. Check the address and try once more.
The Sunday Shortlist decodes one category at a time: the questions buyers ask AI, the shortlist that comes back, and the criteria that decide it. Subscribers get every decode first, and the free toolkit gets you started tonight: a worksheet and three prompts to run the first check on your own brand.
One email a week · unsubscribe anytime · Terms