Why Your Platform Engineering Career Is Really A Sales Job

LONDON — It turns retired that level technologist is simply a income job.

We’ve written a batch astir really DevOps technologist and level engineer roles differ. But while a DevOps squad mostly serves conscionable developers and operations, an internal developer platform (IDP) squad serves galore stakeholders. And it’s successful a much much vulnerable spot to support its budget.

At Team Topologies’ Fast Flow Conf this month, Christopher Batey, CTO astatine nan Core Engineering Consulting Group, decided to steer nan speech past nan developer customer that’s astir associated pinch an IDP. After all, take can’t beryllium your only metric — conscionable because it has users doesn’t mean your IDP is useful aliases improving nan bottommost line.

Instead, Batey focused connected really to scope nan different stakeholders — elder leadership, merchandise management, alteration guidance and information — to waste your level internally.

Don’t worry, this won’t beryllium a income people — astatine slightest not explicitly. It mightiness beryllium a instruction successful communication. He presented tips for applying different level metrics to style conversations pinch your platform’s galore customers, to show your worth to different audiences and to show really your platform engineering squad is — hopefully — creating lasting effect connected nan business.

Why Platform Engineering Requires a Sales Mindset

Platform projects sometimes usage their first backing to use awesome exertion — period. That’s erstwhile problems tin start.

“Many level engineering teams are delivering awesome platforms while besides struggling to support their existence,” Batey said, “You consciousness for illustration you’ve done a bully job, but past nan worth remains invisible to leadership.”

This dilemma means  45% of level engineering teams are disbanded aliases restructured wrong 18 months.

Since level engineering is successful its early stages astatine galore organizations, immoderate soul income strategy has to beryllium grounded successful explaining nan why, not feature-dumping astir what you person aliases committedness you’ll do.

“I really judge level engineering is nan measurement we standard DevOps,” Batey said. But “in practice, surely successful my experience, we are still asking exertion engineers to deliberation astir excessively much, to do excessively much.”

Despite DevOps having surgery down immoderate silos, he said, galore developers are still juggling 3 unreality accounts — for development, personification acceptance testing and accumulation — on pinch penning Terraform aliases Pulumi, mounting up databases and unreality personality and entree guidance (IAM), and past having to activity retired which of often galore ways to deploy a container.

These important yet repetitive, non-differential distractions person devs spending little than 30% of their days (or, by immoderate accounts, moreover less) creating unsocial business value.

“Platform engineering,” Batey said, “is really really we get exertion engineers really building applications.”

Focus connected Stakeholder Problems, Not Your Solutions

Funding often relies connected transforming nan way  you show level engineering worth to stakeholders crossed nan organization. Which is why, arsenic Matthew Meckes,  elder containers master astatine Amazon Web Services, pointed retired successful different Fast Flow Conf talk, it’s not capable for level teams to conscionable measurement nan throughput and stableness of their platform. What really matters is really accelerated a level tin thief package improvement teams present package successful a safe, unafraid and reliable way.

Batey is connected committee pinch defining level engineering’s effect much broadly.

“I deliberation nan extremity of level engineering is to alteration change astatine velocity pinch information built in, hopefully while maintaining aliases expanding reliability,” he said. “I for illustration this meaning of level engineering complete the, ‘I’m building self-service capabilities to alteration developers to beryllium productive,’ etc., because it tells america why we do level engineering alternatively than nan existent what of level engineering, because nan what is ever going to change.”

Once you and your level squad are grounded successful that why, you tin past move connected to knowing your stakeholders’ varied perspectives and what occurrence intends for each. From there, you tin create much meaningful metrics that are grounded successful their needs.

“I ever request to punctual myself to framework those discussions based connected nan personification I’m talking to and not nan solution that I want to implement,” Batey said. “This is connected a Post-It statement that I person connected my show for distant meetings, and it’s thing I publication retired to myself 5 times earlier going to an in-person meeting.

The Post-It note, he said, reminds him  “to debar astatine each costs discussing your solutions pinch stakeholders. They do not attraction astir your solutions. They attraction astir their problems. So let’s effort to understand them.”

Only then, he continued, tin level engineering teams person accountability and beryllium respected arsenic agents of change.

“As level engineers, past we should ever cognize really nan point that we’re building is either going to summation nan velocity astatine which this institution tin present software,” he emphasized. “It’s going to make a measurable betterment to security, aliases it’s going to amended reliability.”

Identifying Your Platform’s Key Stakeholders

“Platform engineers request to talk to everyone,” based on Batey, “because level engineering’s occupation is to bring everyone together to really present alteration to production.”

At astir organizations, this includes, but is not constricted to:

• Head of engineering.
• Product engineers.
• Change managers.
• Infrastructure teams.
• Security teams.

Start by looking astatine really a alteration gets to production. The level squad intends to salary down what Batey called nan “coordination taxation ” — nan clash encountered arsenic alteration moves to production. Just retrieve that these steps aren’t conscionable grounded successful technology, but see a batch of group and processes on nan way.

For nan executive stakeholders, risk, return connected finance and velocity are cardinal metrics, while nan caput of engineering for merchandise improvement cares astir predictable onboarding, transportation and quality. Both stakeholders person a keen oculus connected nan percent of clip really spent delivering worth to extremity users.

Since we cognize that writing codification was ne'er nan bottleneck, nan remainder of nan package improvement life rhythm — nan outer loop — is wherever some trends of level engineering and AI tin person nan astir impact. Both tin thief trim nan clip to merchandise to accumulation aliases trim clash successful different cardinal areas for illustration generating archiving and making everything much discoverable.

Unsurprisingly, security’s biggest concerns usually see coverage, velocity to remediation and number of endpoint vulnerabilities. Security stakeholders successful particular, Batey said, want to beryllium brought successful earlier, to really cook it into nan platform.

Change guidance practitioners are astir concerned pinch high-risk changes and really a level tin wide thrust a safe, predictive clip to production.

Head of infrastructure cares astir adoption, consistency and operational excellence.

How to Structure Conversations pinch Stakeholders

“I americium an technologist by heart. By default, I americium not a earthy communicator, and I would beryllium overmuch happier penning code, building level features,” Batey acknowledged. “The only measurement that I tin amended nan measurement I pass pinch these group is to systematize it.”

Before talking astir immoderate solutions, he useful together pinch stakeholders to verbalize nan problem. Then, he looks to quantify it. For each of these stakeholders, Batey presented illustration problems and past really to quantify them, to past measurement for immoderate change.

Executives

• Problem: Product transportation is excessively slow.
• Quantified: Features delivered monthly.

Change Management

• Problem: No spot successful nan grounds attached to alteration requests.
• Quantified: 95% of alteration requests see person manually added evidence.

Security

• Problem: Vulnerabilities conscionable support growing.
• Quantified: 35 vulnerabilities successful production.

Head of Engineering

• Problem: Engineers struggle to get things done.
• Quantified: Service onboarding is 2 weeks. Pull requests return 3 days connected mean to merge.

The stakeholder will not ever beryllium alert of a problem they are facing, Batey warned. For instance, information whitethorn beryllium wholly unaware that each squad has a different way to production. All they cognize is that information devices are not being adopted enough, aliases nary 1 is taking ownership for fixing bugs.

You commencement by quantifying really galore assets person vulnerabilities, he suggested. Then you bring successful outer impervious of really different businesses person solved nan aforesaid problem. Finally, you effort to create soul impervious pinch a short impervious of concept.

Finally, together, you create what Batey refers to arsenic a “library of needles,” aliases metrics that you co-own, betting that you tin move those needles.

Which Platform Metrics Matter Most to Leadership?

Your level inaugural needs to grounds measurable values of speed, performance, information and/or cost. Having already gone successful extent successful discussing security, Batey gave immoderate examples for each of nan different three.

Speed Metrics

• Product motorboat time.
• Number of user-facing features capable to beryllium released.
• Lead time.
• Lead clip for changes.
• Deployment frequency.
• Engineering onboarding to subordinate a squad and merchandise to production.
• Service onboarding.

This velocity is often accrued by removing manual actions and approvals.

Reliability and Performance Metrics

• What percent is nan user-facing merchandise is available? Different nines per features?
• Are location outages during cardinal events?
• Are location service-level statement breaches?
• Service-level objectives (SLOs).
• Service-level indicators (SLIs).

Working straight pinch exertion engineers, Batey added, you tin besides set:

• Functional trial sum for cardinal business usage cases.
• Non-functional trial coverage.
• Reliability trial coverage.

Cost Metrics

Total costs of ownership (TCO) and return connected finance (ROI) are nan cardinal metrics that each squad leaders should understand, particularly successful this clip of arguing for your job. He offered immoderate cardinal costs metrics:

• Service cost.
• Cost per customer.
• Cost per personification journey.
• Storage cost.
• Hosting cost.

A Practical Guide to Prioritizing Your Metrics

The hardest portion of level engineering is deciding what to do successful what order, Batey said, which tin mean teams measurement excessively much. Measurements request a hierarchy, which he grounded successful applicable examples.

Imagine merchandise releases are slowing down. Executives are complaining that your package statement is falling down competitors. But then, if teams effort to velocity up, reliability suffers pinch much vulnerabilities.

His test is that a manual group of environments caused differences successful testing and production. Developers are skipping automated tests, and information is having to deploy otherwise for each environment. The solution, he said, that a level engineering squad should coming to execs is simply a standardized, consolidated way to accumulation for cardinal services, making usage of consistent, lightweight environments, which are automatically created successful a accordant manner.

“Always commencement simple,” Batey recommended. “Pick a cardinal business usage case, find retired what components are involved, and, immoderate inaugural that you’re doing, conscionable make judge it throws immoderate worth for those cardinal personification journeys.”

Measurement is nan champion measurement to construe nan worth of your level program, grounded successful nan accountability of why. To do that, commencement simple, he advised, by picking 1 key, easy measured business usage lawsuit per stakeholder.

Just remember, “the extremity isn’t to measurement everything,” Batey said. “It’s to measurement nan correct things to show nan worth of nan level features that you build to cardinal people.”

This, of course, includes everyone who holds nan purse strings. But also, he said, “you don’t want group to beryllium blocking nan things that you’re trying to do.”

Download nan ebook “Platform Engineering: What You Need to Know Now.”

Group Created pinch Sketch.