Crafting Recommendations that Stick

On writing persuasively in the service of positive healthcare change

Author

Satyan Chari
November 10, 2025

A lightly altered version of the original Photo by frame harirak on Unsplash

What makes for a good recommendation?

My colleague and respected human factors expert, Thom Loveday, recently shared some observations relevant to this question. Within a broader set of points, he highlighted that many recommendations (aimed at guiding positive change or improvement in healthcare), even those put out in public-facing formal reviews, often suffer from problematic levels of ambiguity. Thom focused on the importance of clarity as a fundamental starting point. I tend to agree. Yet, a little more can be said about crafting good recommendations - because, whether something is ‘good’ is a much more contingent question. Good in what way, for what purpose, in which circumstances, and for which audience?

As someone who has been around the healthcare block a few times, I have a good deal of experience in writing recommendations over the years: good ones, middling ones and truly abysmal ones. I have come to learn that there is a lot at play, very little of which gets discussed properly.

In this piece, I take a slightly provocative position: suggesting that while clarity is paramount, a recommendation’s real value lies elsewhere - in its capacity to mobilise action in the real world. By extension, the capacity of a recommendation to drive real change has everything to do with the context from which it came and into which it is being delivered. Ultimately, recommendations don’t change anything, people do, so recommendations must be persuasive to the people involved in ways that lead to positive change otherwise they are pretty useless. Below are some observations on what it takes to craft such recommendations.

Finding the generative middle ground

There is always some degree of separation between the explicit requirements set for a piece of work, what is implicitly desired, and what is actually needed. We all navigate these spaces in our various roles but not always with intent. A late evolution in my own practice has been to devote much more time to develop an understanding of hidden tensions at the start of any review that I am brought on to. Simple questions around the parameters and constraints involved can reveal much. Other times, more thoughtful probing is required. Asking about organisational realities, the current concerns of leadership and inviting early perspectives on the problem (or even those adjacent to it), can help you connect the dots more clearly.

Typically, the pragmatic options for real improvement must be negotiated out of this mileu of cross-purposes, hidden considerations and pragmatic realities. A set of recommendations that slant too heavily towards meeting the explicit and implicit requirements can be clear and actionable but ultimately damaging to the long-term interests of the organisation. We should not, in good conscience, go down this path. Equally, recommendations that slant too heavily towards the actual need (when all indications are that it might be a bridge to far) can alienate the very stakeholders we need to influence. Within all of this complexity reside real opportunities to find a middle path, one that energises necessary movement.

Consider the following examples:

Scenario 1: The ED Medication Error

A patient in a busy Emergency Department (ED) receives a duplicate IV antibiotic dose, resulting in a moderate adverse reaction. The error was identified by a family member. The nurse involved is highly competent but was working in an under-staffed, high-pressure shift and reported significant cognitive load from working with the EMR.

Explicit Requirement: The hospital's Quality and Safety Unit commissions a Root Cause Analysis (RCA). The official Terms of Reference state: "To review the incident, identify any deviance with the hospital's Medication Administration Policy, and recommend corrective actions to prevent recurrence."

Implicit Requirement: The ED management is facing significant pressure regarding performance metrics (e.g., wait times, length of stay). The unstated expectation for the review is to deliver a timely and focused resolution. The preference is for recommendations that address immediate compliance, knowledge or behavioural gaps (retraining, reminders etc) rather than recommendations that would trigger a broad, complex, and resource-intensive review of the EMR platform or departmental staffing models.

The Actual Need: The EMR's alert system does not visually differentiate between low-priority advisories (e.g., "pharmacy review pending") and high-priority critical warnings (e.g., "duplicate dose"). Clinicians are exposed to hundreds of low-value alerts per shift, creating a high signal-to-noise ratio that predictably leads to cognitive strain and inattentiveness to alerts. The actual need might be to look at the design of the EMR's alert hierarchy and user interface, not retraining staff to navigate a flawed system.

Scenario 2: The Faltering Digital Transformation Journey

  • The Scenario: Two years after a major Electronic Medical Record (EMR) implementation, clinician satisfaction is at an all-time low. Staff surveys cite chronic system slowness, poor integration between the EMR and lab/radiology systems, and clunky workflows. As a result, staff have developed numerous workarounds such as using insecure personal mobile devices to share patient data.

  • Explicit Requirement: The Hospital Board, concerned about the low return on a significant capital investment and with an awareness of the emerging risks related to workarounds, commissions an external "Post-Implementation Review." The directive is: "To evaluate user adoption, identify key areas for workflow optimisation, and develop a targeted training strategy to improve compliance and system utilisation."

  • Implicit Requirement: There is significant reputational investment attached to success of the EMR project. The unstated expectation is that the review will validate the integrity of the core platform. The desired outcome is a set of recommendations focused on behavioural solutions. There is a clear organisational bias against recommendations that would suggest a fundamental flaw in the vendor choice, initial configuration, or that would require significant new capital expenditure.

  • The Actual Need: The source of the workarounds is not poor user training or compliance but a deep misalignment between the platform's architecture and the hospital's operational needs. The actual need is a strategic (and costly) path to address these core technical and integration deficits, as no amount of training can fix a system that is currently not fit-for-purpose.

Both of these scenarios offer no easy solutions. Its up to us, as change agents, to creatively and collaborative construct what is possible within these problem spaces.

In the medication error case, the deeper need for a timely (and contained) resolution could perhaps be softened through well-constructed conversations with leadership on the potential negative downstream consequences of taking a hard knowledge-based approach to managing the problem. If not, a slightly tangential path might be more palatable: for instance by providing a small amount of protected resources to enable clinically-led quality improvement initiatives around IV antibiotics. While the end point might still be a knowledge-based intervention, allowing some latitude for a such a project to be devised and implemented locally can enhance morale by building autonomy and create positive dividends in many unforeseen ways.

For the Digital Transformation example, the paradigm of 'training will fix it' is clearly unsuitable. A middle path might involve reframing the 'Actual Need' not as a single, cataclysmic capital request, but as a phased multi-year journey to acknowledge and clear the ‘technical debt’ that has been accumulating. This approach respects the Board's fiscal constraints but purposefully moves the conversation beyond surface-level fixes.

It might start by publicly validating some of the clinicians' frustrations (a crucial relational step) and setting up a small taskforce of passionate clinicians and capable IT leads. This team wouldn't be asked to "fix the EMR" (which is too big) but possibly to investigate and inform a suitable solution to alleviate the single most painful workflow within 90 days. This builds momentum, it demonstrates responsiveness, and could yield a tangible proof-of-concept that enables future conversations around the deeper, more fundamental improvements that are actually needed.

Recommendations as a Catalyst, Not an Artefact

We often think of recommendations as the final product of a review; the "answer," carefully bound in a report that lands on a desk. This is a mistake. A report is a static artefact. A recommendation that sticks must be a catalyst - flexible, useful and usable. No surprise then that its design must suit its intended users. This means both the leaders who have to get behind the recommended work and teams who must enact the change.

In essence, a sticky recommendation must be:

  • Specific enough to be actionable without being prescriptive. A recommendation to "Improve communication" is an artefact. Its so obtuse as to be devoid of any value. In contrast, a recommendation to “trial a daily interdisciplinary team huddle to discuss high-risk patients” is better. It’s specific and practical, but it doesn’t stifle or bind teams to a new path. While it doesn’t provide a solution, it offers a mechanism to get to one.

  • Fine-tuned for adoption. A recommendation that requires a stressed organisation to carve out a $5 million budget and 18 months of work is both overwhelming and impractical. Instead, a recommendation that calls for trials of small test changes invites engagement, its generative, it reduces risk, and if it validates the early path, could gently catalyse rethinking of deeply embedded organisational logics that might have derailed a larger, more decisive proposal at the very start.

  • Owned. A recommendation addressed to "the organisation" will be owned by no one. A good recommendation gently but clearly identifies who (a role, a committee, a team) is best placed to carry it forward.

The goal always is to provide a handle for organisations and stakeholders to grasp. If it's too abstract, too big, or has no obvious owner, it will simply sit on the shelf: clear but perfectly useless.

The Real Work: Persuasion & Mobilisation

❝

"If you wish to persuade me, you must think my thoughts, feel my feelings, and speak my words,"

- Cicero

Photo by Antenna on Unsplash

Cicero’s advice to students of rhetoric has a lot to offer us two thousand years later. Clarity is the entry fee, but persuasion is what we are going for. Persuasion, in this context, comes from deeply understanding the world of the people you are trying to influence - their pressures, their constraints, their hidden fears, and their unexpressed hopes and then working within that space rather than one of our own choosing. A recommendation that "sticks" emerges from that work because it resonates deeply on a human level. No amount of technical analysis will ever deliver that. Do this right, and your recommendations will naturally:

  • Acknowledge their reality by showing that you have genuinely listened and see the problem from their perspective. You are validating the real-world constraints and daily pressures they face, not just working off the technical details.

  • Paint a compelling, positive vision by framing the change not as a critique of the present, but as an achievable and desirable future. The recommendation should tap into their quiet aspiration for things to be better.

  • Provide a safe first step in the context of what is possible, the expressed fears and constraints. The suggested path should reduce the barrier to entry, making the first move feel logical, accessible and low-risk.

Ultimately, crafting recommendations that stick is less an act of technical writing and more an act of social and political navigation within the complex, messy, and deeply human realities of healthcare.

Subscribe and never miss an issue, unlock additional resources at the end of each newsletter and get full access to previous content!

 Post script

For a whole host of professional and personal reasons (followed by many weeks of ‘blank page syndrome’), The Human Stream newsletter had fallen into unintentional abeyance for about six months. I do apologise for the extended break - thank you for sticking around. I’m looking to forward to getting back into a rhythm of writing. See you in a month!

The Human Stream is a monthly newsletter for clinician improvers, safety and quality professionals, governance teams and healthcare leaders. The Human Stream compiles insights, topic overviews and practical tips from contemporary safety and systems sciences, all in an easy-to-read, information-rich package, conveniently delivered to your mailbox!

The Human Stream is free because of sponsorship from Nuansys Healthcare. Find out more:

AI Use Declaration: AI was used to generate the first version of case studies used in this post. All other material is written by a human. AI was used to support background research, in enhancing and proofing of content. All quotes are verified and directly taken from source documents.

  1. My Take on SMART for goal setting: https://thehumanstream.com/p/beyond-smart