In September 2026, Trafalgar House released findings from its 2026 Trust & Confidence Survey showing that 24.4% of respondents agreed their pension would enable them to live comfortably in retirement, down from 26.3% in 2025. Another 17.1% selected “don’t know,” up from 15.8%. The research covered more than 2,000 UK adults and was completed at the beginning of 2026, according to the provider. Publication in September should not be mistaken for the fieldwork date.

These answers measure expectations, not projected retirement income or the actuarial adequacy of any scheme. Respondents will have different ages, savings histories, household resources and definitions of “comfortable.” The survey can identify a communication challenge, but it cannot establish that a particular employer plan is insufficient or that confidence fell because benefits were cut.

Balances are not an income plan

A pension statement may accurately show a current pot while leaving a member uncertain about future income. Contributions, investment returns, inflation, tax, retirement date and the chosen method of drawing benefits all affect the translation. Employees may also hold several pensions and other assets. Providing a single balance without assumptions can lead to misplaced certainty as readily as anxiety.

Communication should therefore show ranges, assumptions and trade-offs in plain language. Members ought to see what changes when they contribute more, defer retirement, choose a different risk profile or consolidate information across schemes. Such illustrations need caveats: they are scenarios rather than guarantees. Administrators should test whether people understand those caveats rather than simply tracking email opens.

Governance implications for employers

Trustees and employers have different legal roles, yet both have an interest in usable member information. A governance pack can pair operational accuracy with comprehension: error rates in records, time to resolve queries, member understanding, engagement with retirement tools and actions taken after contact. A high portal login count does not show whether workers can make sound decisions.

For global benefits managers, local systems differ sharply. A UK defined contribution projection cannot be transplanted into countries dominated by statutory or defined benefit arrangements. The common principle is to explain likely income, uncertainty and choices in terms suited to each jurisdiction. Where advice is regulated, signposting should respect that boundary.

A better member test

The survey’s uncertainty figure should invite a question about actions, not just sentiment. Compare the information a member receives at joining, after a job change and in the years before retirement. Check whether projections incorporate realistic contribution histories and whether members can find the assumptions without specialist knowledge. Some will need help reconciling old pots; others need a clearer distinction between nominal and inflation-adjusted income. A communication experiment can compare two statements using the same underlying numbers and test what people actually understand. Record whether the format changes intended action, not just confidence. The sponsor and trustees should also consider how vulnerable members and people with limited digital access receive equivalent support.

Evidence still needed

The published release does not provide a scheme-level breakdown that would connect confidence with benefit generosity. Employers should therefore resist attributing the finding to any one contribution policy. Administrative records, member interviews and a transparent income projection would offer a stronger basis for decisions than one national sentiment measure.

What to test next

This editor recommends running a comprehension test with different age and income groups before redesigning a member portal. Ask participants to interpret a range of possible incomes and identify one action available to them. Compare results with the administrative data; a reassuring presentation that obscures risk is not a success. Trafalgar House’s survey supplies a prompt for this work, not a benchmark for every scheme.