A field guide for Medical Affairs

The Engagement Playbook

Experts aren’t leads. The relationships that change clinical practice are built before you need them, and your own network is the best data you will ever have. This guide shows how to organise it, prioritise it and grow it, with AI as a guide, not a base. These are lessons from my own path, shared to compare notes. Take what’s useful, adapt it to your market, and tell me what you’d do differently.

Derieck Phillip Reid Sing
Derieck Phillip Reid Sing, MD
Medical Affairs · KOL engagement · digital health
CLINICAL QUESTION Societies Guideline groups Digital voices Hospital committees Specialists Patient groups
If you only have a minute
01Experts aren’t leads.Build the relationship before you need it.Principles → 02The most influential expert is often not the one who decides.Map both.Starting from zero → 03Know where you stand, not just who matters.Four questions, one honest answer.Health check → 04The colder the contact, the more specific the reason.Ask for a perspective, never a favour.How to open → 05Your network is the dataset.AI categorises, then creates, then discovers.Your data + AI →

Everything below goes deeper. Jump to what’s useful to you and skip the rest.

Why now

Most teams don't lack relationships. They lack structure around them.

Who do we engage, and why? Who owns each relationship? Does any of it change clinical practice? Three shifts make these questions harder, and more important, every year.

Influence has spread out

It sits across societies, committees, guideline groups and digital channels, not only with a few senior names.

Organisations are matrixed

The same expert hears from global, regional and local teams, often in the same month.

AI changes the workload

Organising what you know, preparing for meetings and learning from them no longer needs a large team, if your own data feeds it.

Six principles

What held up when building engagement from zero

Built across eight markets with no brand presence, no network and no history. They apply just as well to a mature team rethinking its approach.

1

Start with the clinical question, not the contact list

Define what needs to change in practice and which evidence gaps stand in the way. The right experts follow from that, not from a database of famous names.

TryWrite one sentence per priority: "Clinicians need to know ___ to change ___."
2

Build the relationship before you need it

Experts are not leads to be captured. Trust is earned over time by giving before asking: relevant data, a genuine question, their perspective taken seriously, a follow-up that shows you listened. The colder the contact, the more specific your reason to reach out has to be.

TryBefore your next request, list what you have given this person in the last year.
3

Map influence beyond titles

The person who shapes practice is often a society working-group member, a committee chair or the colleague everyone calls for advice, not always the most senior professor.

TryAsk three field colleagues: "Who do clinicians call when they're unsure?"
4

One owner per relationship

When several teams contact the same expert without coordination, credibility suffers. Agree one owner, and let others contribute through that person.

TryPull last quarter's contacts across teams and count the duplicates.
5

Design compliance in, not on

Clear rules on fair-market value, scientific balance and separation from promotion make engagement faster, because teams stop second-guessing every interaction.

TryTurn your top five compliance questions into a one-page decision guide.
6

Measure quality, not volume

Meeting counts tell you little. Insights gathered, evidence gaps closed and changes in practice tell you whether engagement is working.

TryReplace one activity KPI with an outcome KPI next quarter.
AI in engagement

Four myths worth retiring

Tap a card to see what actually holds up.

The framework

Connect → Organise → Engage → Learn

A cycle, not a project. It starts with people, not software. Select a step to see what to do, and where AI helps.

Connect Organise Engage Learn PEOPLE FIRST AI SECOND
AI is a guide, not a base. It organises, drafts and suggests. Relationships, judgement and trust stay with people.
Starting from zero

No network, no brand, no history. Now what?

A new market, a new therapy area, a launch. Databases give you names; they don't give you a way in. This is the sequence that worked across eight markets, and became a model others adopted.

1

Listen before you present

Visit, ask what is hard in their daily practice, and say little. Your first credibility comes from understanding their problem, not from your slides.

2

Map who decides, not only who speaks

In many systems the KOL isn't the decision-maker. Technical committees, procurement boards, pharmacy and nursing leads decide what reaches patients. Know both maps.

3

Earn credibility through technical validation

Small, hands-on sessions where specialists test the science themselves persuade more than any presentation. Let them find the value.

4

Go where they already gather

National congresses and society meetings already have the audience. Workshops held alongside them cost less and reach more than standalone events.

5

Let your first advocates open doors

One respected clinician who introduces you is worth fifty cold emails. Discovery tools help you choose whom to meet; introductions do the rest.

6

Turn what worked into a model

Write down what worked in market one: the sequence, the materials, the people. It becomes the template for the next market, and the next team.

Your network + AI

Your network is the dataset

Generic AI knows what's published. It doesn't know who answers your calls, who changed their practice after your workshop, or who the whole ward asks for advice. That knowledge is yours. Feed it in, and AI becomes useful.

Where AI earns its place, in order

1 · CategoriseMake sense of the network you already have 2 · CreateBriefs, summaries, educational drafts 3 · DiscoverWho's missing? Last.

The more of your own knowledge you put in, the higher on this ladder AI is useful. With no data of yours, all it can do is step 3, and that's the weakest.

Starting from zero? Discovery moves first, but only to decide whom to meet. Introductions, congresses and societies do the rest. Once you have a network, the order flips back.

Five steps to feed your network into AI

  1. Gather what you have. CRM exports, your meeting notes, congress contacts, your LinkedIn connections export, society lists you are entitled to use.
  2. Give it structure. One row per person, the same columns for everyone. The template below is a starting point.
  3. Add what only you know. How you met, what they care about, what they said last time, who trusts them. This is the column no vendor can sell you.
  4. Protect it. Personal data needs a lawful basis and an approved tool. Use your company's approved AI, or replace names with IDs before anything leaves your hands.
  5. Give context, then ask. Tell the AI your clinical question and your criteria, then ask it to categorise, find gaps and draft. Check everything it returns.
Template

Network file columns (paste into a spreadsheet)

ID, Institution, Specialty, Society / committee roles, Main clinical interests, How we met, Last interaction (date + topic), What they care about, Influence on practice (1–5), Openness to exchange (1–5), Relationship owner, Consent / data basis, Notes
Tool · Prioritise

The engagement priority scorer

Rate an expert, society or committee on five criteria. The scorer suggests a depth of engagement. Use it to make prioritisation explicit and discussable, not to replace judgement.

Examples:
Partner

Weights: scientific relevance 30%, influence on practice 25%, evidence-gap fit 20%, reach 10%, openness to scientific exchange 15%. Adjust them to your strategy; what matters is that the team agrees on them.

Tool · Relationship health

Influence tells you who matters. This tells you where you stand.

Every mapping tool measures how influential an expert is. None of them measures the relationship. Think of one expert you work with and answer four questions honestly.

Answer the questions

Your relationship status and the next move will appear here.

The colder the contact, the more specific the reason

How to open, by temperature. The rule underneath: ask for a perspective, never a favour, and let them offer more.

Cold

Open with something specific to them: their paper, their talk, a question their work raised for you.

Ask for their view on one clinical question. Nothing else.

Never a generic introduction, a product, or a meeting request in the first message.

Warm

Open with what you did with their last input. Closing the loop is the strongest signal you listened.

Ask for a reaction to something concrete: data, a draft, a plan.

Never let the only contact be when you need something.

Strong

Open with the real problem, early, before decisions are made.

Ask for honest disagreement. If they never push back, it isn't as strong as you think.

Never overuse it, or route every team's request through them.

Tool · Coordinate

Who owns what

A simple ownership model removes most duplication. Adapt the levels to your organisation; the point is that everyone can see it.

RelationshipGlobalRegionalLocal / affiliate
International society leadershipOwnsContributesInformed
Global guideline authorsOwnsContributesContributes
Regional experts and congressesInformedOwnsContributes
National societies and committeesInformedContributesOwns
Practising specialists and digital voicesInformedInformedOwns
Tool · AI prompts

Five prompts, in the order to use them

Categorise first, create second, discover last. Each prompt keeps AI grounded in your own data and sources. Use anonymised information, and only tools your company has approved.

Measure

Swap activity metrics for outcome metrics

You don't need to drop activity counts. Pair each with a measure that shows whether the activity mattered.

Before you quote an AI vendor’s numbers

Many headline figures in this space (“9x faster shortlisting”, “up to 200% more experts”) come from single-client case studies. They may well be true. Ask five questions before they go into a business case:

  1. What was the baseline, and who measured it?
  2. One client, or many?
  3. Over what period?
  4. Does it measure quality of engagement, or only speed and volume?
  5. Can we test it on one therapy area first?
Digital opinion leaders · Responsible AI

Two checklists worth printing in your head

From building a network of more than 100 health creators across the Americas, 15 of whom became collaborators, and from designing an AI-guided health platform. Tick them off as you go.

Before working with a digital voice

    Before using AI in engagement

      Self-check

      Where is your team today?

      Six statements. Answer honestly; it takes a minute.

      Get started

      A 90-day plan

      Start with one therapy area or one market. Prove the approach, then scale it.

      DAY 1 DAY 30 DAY 60 DAY 90 Agree Clinical priorities and evidence gapsGather your network into one fileSet AI and compliance guardrails Build AI-assisted categorisation of your networkValidate with the people who know themAgree segments and ownership Run and learn One engagement cycleCapture insights in one formatReview what changed, then refine
      Derieck Phillip Reid Sing
      About the author

      Derieck Phillip Reid Sing, MD

      MD · MSc Medical Scientific Liaison · ICH-GCP

      Physician in Medical Affairs, across pharma and medical devices. As Regional Medical Director, I built KOL networks and medical society relationships from zero across eight markets, and built and led a multi-country medical team. Before that, Medical Affairs at Novartis and pharmacovigilance at Pfizer. As founder of a digital health start-up, I designed an AI-guided health platform and built a digital opinion leader network across the Americas.

      I'm still learning, and this is not the last word on engagement. It's what I've seen work, written down in case it helps someone. If you see it differently, or something here doesn't hold in your market or therapy area, I'd genuinely like to hear it.

      linkedin.com/in/derieckreidderieckreid.comderieckrs@icloud.com