Every reporting technology provider promises automation.
“Automate your entire reporting process.”
“Zero manual effort.”
“Reports generated in minutes.”
The reality is more nuanced – and understanding where automation actually works is more useful than the marketing version.
Some parts of client reporting can be fully automated. Others can be significantly accelerated with technology support. And others will always require some human judgment and cannot be automated without compromising quality.
Knowing which is which is essential to evaluating any reporting platform – and to ensuring you’re not being oversold a solution.
Parts that can be fully automated
Data extraction and assembly. Ingesting performance data, holdings, allocation breakdowns, risk metrics and benchmark returns from third party and internal systems is a fully automatable step. A modern and flexible reporting platform does this without human intervention, every production cycle, with consistent formatting and calculation logic.
Performance and attribution calculations. These should be undertaken within specialist upstream performance and attribution systems, and NOT within your reporting system. Do not be lured into putting calculations in the wrong place, and then using your client reports to check data.
Template population. Once data is assembled and any aggregations, and data shaping are undertaken, populating all the sections of a report template – performance tables and charts, asset allocation doughnuts, risk statistics, top holdings – all these are an automated step. The template is configured once; the system fills it in for every client, every cycle.
However, more modern and flexible systems like Opus Nebula are dynamic in how the template flexes: in these cases the template flexes based on data. A single template may be used for many many reports, dynamically flexing between specific client requirements, different funds and strategies, between funds and shareclasses and between various languages. If you ever want to change something, it is clearly quicker, easier and cheaper to amend one template, rather than a plethora, and this approach also guarantees consistency between all the outputs and on the timing of any changes.
Regulatory disclosures. Specific regulatory disclosures, standard risk warnings, and other required statements can be incorporated into templates and auto-populated from the relevant data sources. Regulatory updates to the required content can be implemented centrally and pushed to all templates, without any duplication. The system should automatically assign the appropriate disclosures, statements and footnotes, based on data such as, client, mandate, fund, registration and distribution, share class, period since launch etc.
Distribution. Routing completed, approved reports to the correct clients and third parties through the correct channels – portal, fund centre, email, print – is fully automatable once the distribution logic is configured per client. Flexibility is required whether the reports are sent individually, or in a batch, whether meta data is required in the PDFs, if passwords are required on the files, what the filename should be. These requirements should all be configuration, not needing to be built each time.
Parts that can be significantly accelerated
Commentary. Investment and market commentary cannot be automated in the sense that a machine can replace the judgment of a fund manager about what their intentions were and why their portfolio performed as it did.
But the process of collating, structuring and incorporating commentary can be made much faster: pre-populated data context, structured input fields, automated reminders, and workflow tracking.
Quality assurance. Automated QA checks – flagging inputs outside expected ranges, checking for missing data fields, comparing outputs against prior period figures – can catch a significant proportion of errors before human review of the report. They do not replace human review, but they make it much faster and more focused – especially so if the checks are undertaken on the data prior to production of the report.
Configuration for individual clients. Where clients have specific ‘additional’ requirements – additional benchmarks or comparators, different classifications and breakdowns, variations on how ‘cash’ is treated for analysis and comparison purposes, perhaps an additional data section, or personalised commentary – these can be configured within the platform and applied automatically rather than being managed manually for each client relationship.
Parts that genuinely require human judgement
Investment narrative. The fund manager’s explanation of what happened in the markets and the portoflio, why, and what it means for the future, cannot be replicated by automation. AI tools can assist with drafts, but the judgment and accountability for the content must sit with a human.
Final review and sign-off. An experienced person needs to review the completed pack before it reaches clients. Automated pre-production checks, and post production automated validation of items such as page limits being exceeded etc. reduce the burden of this review, but they cannot substitute for it.
Understanding how the system works and how pages are generated helps people look in the right place, rather than looking at every report, and every where on that report. The signatory is accountable for what goes to clients.
Exception handling. When something unexpected happens – a data feed delivers incorrect figures or fails to deliver all the data, commentary is late – a human needs to diagnose and resolve it. Systems can flag the exception; they cannot decide what to do about it.
The realistic expectation
A modern, flexible reporting system can automate the majority of the labour required for a typical reporting production cycle. What remains – commentary, review, exception handling / follow up – cannot be entirely removed, but it can be focused on work that adds genuine value rather than production administration. When the process is largely automated, this frees up the reporting team from “doing the work” to “managing the work”, and typically can allow the team up to 10 x the productivity that was previously possible, and frees the team up to take on more value-added activities.
Frequently asked questions
Can AI write fund manager commentary?
AI tools can generate structured first drafts of commentary based on information provided such as performance data, attribution, transaction information and market context. These drafts can be reviewed and approved by fund managers rather than written from scratch, which saves time. The key requirements are that the fund manager retains responsibility for the content, the AI-generated text is reviewed and enhanced rather than published directly.
What is the biggest barrier to reporting automation in practice?
Data quality and the flexibility of the reporting system. Automation is only as reliable as the data it works with. Firms with fragmented, inconsistent data environments – multiple portfolio systems, unreliable custodian feeds, manual data adjustments – find that automation downstream is limited by the quality of data upstream. Addressing data foundations is often a prerequisite for effective reporting automation. Older, less flexible reporting systems may require individual customisations, and multiple templates to achieve the desired report content. In this case, automation will be reduced and manual checking of reports will be increased.
Does automating reporting reduce the need for a reporting team?
It changes the team’s composition and what they spend their time on. Production tasks that previously consumed most of the team’s time are automated and handled by the platform. What remains is management of the process, oversight, quality assurance, exception handling, client relationship support, and continuous improvement of the reporting process. Some firms reduce headcount; others reinvest the capacity in value-adding activities.