10 TACTICS ✣

Ambassador and Champion Programs.

Structured recognition programs for the company's most engaged external community members. When designed well, they are among the highest-leverage tactics in DevRel — multiplying the company's reach through people who genuinely love the…

ENTRY · 4 MIN · 880 WORDS

Structured recognition programs for the company’s most engaged external community members. When designed well, they are among the highest-leverage tactics in DevRel — multiplying the company’s reach through people who genuinely love the product. When designed badly, they corrode trust and turn community goodwill into transactional exchange.

What these programs are

Variously called:

MVPs — Microsoft, since 1999.
Heroes — AWS (since 2014), GitLab.
Champions — MongoDB, IBM, Salesforce.
Experts — Google (GDE), Cloudflare.
Ambassadors — HashiCorp, Snyk, Auth0, Datadog, MuleSoft, NVIDIA Inception (startups).
ACEs — Oracle (four-tier).
Stars — GitHub.
Innovators / Wavemakers / Squad / SupaSquad — Snowflake, DigitalOcean, Supabase.

The common pattern: a vendor identifies external community members making outstanding contributions, recognises them publicly, and provides benefits that empower them to continue.

Selection is meaningful. Either nomination-based or rigorous application; not “anyone who fills the form.”
Recognition is visible. A public directory, a badge, a title.
Access is real. Early product access, direct lines to engineering and product, advisory-board-style participation.
Career value is real. Speaking opportunities, content amplification, visibility that matters when changing jobs.
Reciprocity is balanced. Members give time; the company gives access, visibility, swag, and (where appropriate) money.
Terms are bounded. Annual renewal cycles prevent calcification.

What kills programs

Treating members as a marketing distribution channel. “Could you tweet about our new feature?” gets old fast.
Inconsistent staffing. A program manager who departs and is not replaced.
Recognition gap. Members feel less recognised over time as the company scales.
Selection drift. Standards fall; original members feel devalued.
Failure to deliver promised benefits. Promised early access that never materialises.
Acquisition or restructure. The program manager is gone; the program staggers on.

Design choices

Open vs. selective?

Selective programs (Microsoft MVP, AWS Heroes, GDE, MongoDB Champions) produce status value.
Open programs (AWS Community Builders, DigitalOcean Wavemakers) build pipelines into selective programs and reach more emerging contributors.

Most large companies run both — emerging-talent program and elite recognition program.

Multi-track?

Some programs have multiple tracks for different contribution types:

Supabase SupaSquad — Contributor, Content Creator, Trusted Host, Event Speaker.
Oracle ACE — Apprentice, Associate, Pro, Director (tiers).
IBM Champions — Multi-domain badges + Champion status.

Multi-track allows the program to recognise diverse forms of contribution.

Term length?

Annual renewal (Microsoft MVP, AWS Heroes, Oracle ACE) is standard.
Lifetime status is unusual and risky — calcifies the membership.
Multi-year terms (Cloudflare Developer Experts) are intermediate options.

Geographic distribution?

For global programs:

Regional ambassadors representing distinct markets.
Local user-group support alongside the headline program.

HashiCorp Ambassadors span 31 countries; AWS Heroes span 57. Geographic diversity is both signal and outcome.

Benefits that work

In rough order of impact:

Direct line to product and engineering. The single most-valued benefit by most program members.
Early access. Beta features, previews, design feedback opportunity.
Visibility. Featured in company blog, conference speaker placement, social amplification.
Access to other members. A private Slack/Discord, an annual summit.
Conference travel sponsorship. When members give talks about your product.
Swag with actual taste. Generic t-shirts are insulting; well-designed members-only items work.
Cash / vouchers / credits. Often controversial; works for some programs (AWS credits to Community Builders); risks transactional dynamics elsewhere.
Career visibility. Reference letters, intros to recruiters at member-employer firms.

Benefits that don’t work:

Vague “exclusive content” (members can usually find the info elsewhere).
Webinars-only. Members already have access to public webinars.
Discounts on products members don’t buy (developer-individual members usually aren’t the buyer).

Common program-management failures

No dedicated program manager. Programs need an owner. Without one they decay.
Slow nomination cycles. Inbound nominees go cold during long evaluations.
Inconsistent recognition cadence. Some quarters celebrated; others silent.
No clear path from emerging-talent to elite program. Community Builders should know how to become a Hero.
Overlapping internal advocate work. Confusing whether company employees or ambassadors take a speaker slot.

Measurement

Track per program:

Member count and renewal rate.
Member-generated content volume (talks, posts, videos).
Reach of member-generated content.
Product feedback received through the program.
Member-influenced revenue (where measurable).
Member NPS. This is critical — your members can leave.

Member NPS is the leading indicator

If your members’ satisfaction is declining, every other metric will follow within 12 months. Survey members at least twice a year. Act on the feedback visibly. The decay-detection point is the place to intervene.

Sizing

How big should an ambassador program be? Some reference points:

Microsoft MVP — ~4,000 globally across hundreds of technology areas.
AWS Heroes — ~270 (very selective).
Google Developer Experts — ~1,000 in 85 countries.
MongoDB Champions — Several hundred.
HashiCorp Ambassadors — ~100 in 31 countries.
Cloudflare Developer Experts — Smaller, dozens.
Oracle ACE — Several hundred across tiers.

Smaller-and-selective tends to produce more visible value per member; larger-and-inclusive tends to produce more aggregate reach. There is no single correct size.