SQL. You possibly can construct fashions in your sleep. You’ve run dozens of A/B assessments. So why aren’t you getting promoted?
The reality is, most knowledge scientists plateau not as a result of they lack technical abilities, however as a result of they don’t perceive what really adjustments between ranges. Firms hand you a ladder with imprecise rungs labeled “affect,” “scope,” and “strategic considering” after which count on you to determine what these phrases imply in follow.
Many good and expert knowledge scientists get caught at L4 for years, grinding on tougher technical issues, whereas their friends leap to L5 by shifting how they give thought to their work. The profession ladder in knowledge science isn’t a straight line of accumulating extra instruments and strategies. It’s a collection of basic shifts in the way you outline issues, create worth, and affect selections. Every promotion requires you to play a special recreation and it isn’t at all times clear precisely when the foundations have modified.
On this publish, I’ll break down what adjustments at L3, L4, L5, and L6; not by way of summary competencies, however in concrete behaviors and mindset shifts. These are the patterns I’ve noticed throughout dozens of promotions (and seemingly high-performing however stagnant careers) at main tech firms. Let’s decode the hidden ladder collectively.
L3 → L4: Turning into Dependable
The bounce from L3 to L4 is deceptively easy to explain however surprisingly laborious to execute: you need to shift from being an executor to being an proprietor.
At L3, you’re given well-defined duties. A PM or senior knowledge scientist scopes the work, breaks it into steps, and checks in steadily. You write the SQL question. You construct the dashboard. You run the experiment. Another person worries about whether or not you’re fixing the fitting downside, whether or not the metric is sensible, or what occurs after you ship.
At L4, you personal the end result. The distinction reveals up in dozens of small behaviors that compound into a very totally different working fashion.
Ending cleanly turns into non-negotiable. L3s can get away with “I constructed the mannequin, right here’s the pocket book.” L4s ship: documentation that others can use, code that passes overview on the primary attempt, outcomes introduced in a manner that results in clear selections. If you hand off work, nothing comes again to you with “wait, what does this column imply?” or “are you able to rerun this with the corrected knowledge?”
Constructing belief means your supervisor stops checking your work. They know that if you say the evaluation is completed, it’s really achieved: edge instances dealt with, knowledge high quality verified, outcomes sanity-checked in opposition to instinct. Early-career knowledge scientists usually underestimate how a lot of L4 is just proving you gained’t create surprises. Reliability isn’t glamorous, but it surely’s the muse of every thing that comes after.
Asking higher questions separates L4s from individuals caught at L3. When a PM asks for “conversion charge by section,” an L3 builds the question and returns the numbers. An L4 asks: “Are we making an attempt to determine which section to focus on, or validate an current speculation? As a result of that adjustments whether or not we must always have a look at conversion charge or incremental carry.” You begin seeing the why behind requests, which suggests you may usually resolve the precise downside fairly than simply the acknowledged query.
Seeing the following step earlier than being advised could be a very powerful L4 conduct. You end analyzing experiment outcomes, and as an alternative of ready for somebody to ask, you’ve already drafted three follow-up experiment concepts with tough scopes. You notice an information high quality challenge and file the ticket to repair it earlier than anybody notices the bug of their dashboard. You ship the quarterly metric overview and proactively flag the one metric that’s trending in a regarding path.
Right here’s what this appears to be like like in follow: You ship your first undertaking end-to-end with out PM handholding. Perhaps it’s redesigning the person onboarding movement. You don’t simply run the experiment, you write the one-pager proposing it, outline the metrics with the PM, implement the logging with the engineers, analyze the outcomes, current to management, and coordinate the complete rollout. Six months earlier, 5 totally different individuals would have pushed these steps. Now it’s you. That’s the L4 transition.
The L3 → L4 bounce is about proving you might be trusted with greater issues. As soon as your supervisor is aware of you’ll end what you begin, see round corners, and ship high quality work with out supervision, they may give you ambiguous initiatives. Which brings us to L5.
L4 → L5: Turning into Strategic
If L3 → L4 is about dependable execution, L4 → L5 is about turning into the one that defines what issues are price fixing within the first place.
That is the place most knowledge scientists get caught. They preserve perfecting their execution—working cleaner experiments, constructing extra refined fashions, mastering new instruments—whereas lacking the basic shift their firm expects. L5 isn’t about doing the work higher. It’s about deciding what work ought to exist.
Scoping ambiguous issues turns into your core talent. At L4, a PM palms you a query: “Why did engagement drop final month?” You examine and return a solution. At L5, management says “development is slowing” and also you flip that into 5 concrete hypotheses, a prioritized investigation plan, and a timeline for decision-making. You’re snug with ambiguity as a result of your job is to resolve it for others.
This reveals up in the way you method new initiatives. A product crew says “we have to enhance retention.” An L4 asks: “What evaluation would you like?” An L5 pushes again with construction: “Let’s outline what retention success appears to be like like first. Are we optimizing day-7, day-30, or long-term engagement? What’s the enterprise case? Are we making an attempt to hit an org-level KPI or validate a product guess? That adjustments every thing about how we must always method this.” At L5, you’re doing the strategic considering that no person else has time for.
Designing metrics separates L5s from L4s greater than some other talent. At L4, you measure what you’re advised to measure. At L5, you perceive that metric alternative is technique. When your organization debates whether or not to optimize for each day lively customers or time spent per session, you’re the one that can articulate the tradeoffs: DAU optimization would possibly juice short-term engagement by means of notifications whereas degrading long-term person expertise; time-per-session may reward addictive options over useful ones. L5s don’t simply calculate metrics, as an alternative they think about whether or not they’ll drive the fitting conduct. It’s understanding you could optimize the mistaken metric completely and nonetheless hurt the enterprise.
Influencing PMs turns into a core a part of your job. At L4, you’re responsive: PMs set priorities and also you execute. At L5, you’re proactive: you notice alternatives within the knowledge and persuade PMs to guess on them. This could be analyzing person conduct knowledge, noticing {that a} small section has 10x increased lifetime worth, constructing a enterprise case for why the product crew ought to focus their subsequent quarter on increasing that section, and driving the dialog in planning conferences till it’s on the roadmap.
This affect requires a very totally different communication fashion. You cease answering questions and begin shaping which questions matter. You write technique memos, not evaluation experiences. You current suggestions, not findings. The evaluation continues to be rigorous, but it surely’s in service of driving selections, not documenting what you probably did.
Considering in tradeoffs could be the deepest L5 mindset shift. L4s search the “proper reply.” L5s perceive that the majority product selections contain competing values with no clear winner. And your job is to make these tradeoffs specific so leaders can determine. Ought to we launch the characteristic with recognized bugs to hit a deadline, or delay for high quality? There’s no purely data-driven reply, however you may quantify the tradeoffs: “Launching now reaches 2M customers throughout peak season however traditionally our buggy launches see 40% increased assist prices and 15% increased churn. Right here’s what that appears like in greenback phrases.”
Right here’s what L5 appears to be like like in follow: You lead a brand new experiment technique that adjustments the product roadmap. Perhaps development has stalled and the crew is working disconnected assessments. You intend a scientific testing framework: outline a coherent person journey to optimize, map out the highest-leverage factors to check, create a shared metric tree so groups aren’t optimizing conflicting targets, and set up a six-month roadmap of experiments sequenced by dependency and studying worth. This isn’t evaluation, it’s technique. You’ve formed how your whole product org thinks about development for the following two quarters. That’s the L5 transition.
The L4 → L5 bounce is about increasing from executing options to defining issues. When you’ve confirmed you may take ambiguous conditions and switch them into clear paths ahead, you’re able to scale your affect past your individual work.
L5 → L6: Turning into a Multiplier
The L5 → L6 transition is the toughest to make, and the simplest to misconceive. It’s not about being a extra strategic particular person contributor. It’s about turning into a drive multiplier: somebody whose affect scales by means of others. This may be as each an IC or as a supervisor.
At L6, your worth isn’t measured by the standard of your individual analyses. It’s measured by how a lot better you make everybody else’s work. This can be a brutal psychological shift for high-performing particular person contributors who constructed their careers on private excellence.
Setting frameworks others use turns into your main output. An L5 would possibly run one of the best experiment in firm historical past. An L6 creates the experimentation framework that makes each crew’s experiments higher. This might be a call tree for statistical check choice, a template for experiment one-pagers that forces groups to assume by means of success metrics upfront, or a standardized method to measuring incremental carry that turns into the corporate default.
Usually an L6’s output seems to be nothing groundbreaking individually however optimizes work throughout the org. For instance, a reasonably primary “metrics overview guidelines” that forestalls dozens of groups from making metric-selection errors that will waste quarters of labor.
Mentoring issues greater than you’d count on. At L6, you’re accountable for growing L4s and L5s throughout the group, not simply your quick crew. This goes past code evaluations. You’re educating individuals assume strategically. When an L4 brings you a thorny evaluation downside, you don’t resolve it for them—you ask the questions that assist them resolve it themselves: “What choice does this evaluation must assist? Who’s the viewers and what do they already consider? What would make you assured sufficient within the consequence to guess your credibility on it?”
The multiplier impact reveals up right here: one hour of your time educating an L4 scope issues correctly would possibly save them dozens of hours over the following 12 months, and enhance each undertaking they contact. Your impact-per-hour by means of mentoring usually exceeds your affect from doing the work your self.
Driving alignment throughout groups turns into vital at L6 since you’re engaged on issues too large for any single crew to personal. Perhaps knowledge high quality points are hurting three totally different product areas, however no person owns the underlying instrumentation. An L5 would possibly doc the issue and escalate. An L6 convenes the stakeholders, builds consensus on severity, proposes an answer that works throughout all groups’ constraints, and drives it to completion regardless of crossing organizational boundaries.
This requires a very totally different influencing toolkit than L5. You’re not convincing one PM to prioritize your undertaking, you’re aligning a number of groups round a shared downside when every crew has competing priorities. You get snug with statements like: “I do know this creates further work in your crew this quarter, however right here’s why it unblocks three different groups and saves us all six months of ache later.” You make invisible issues seen, and also you make coordination issues tractable.
Recognizing systemic knowledge points earlier than they change into crises is basic L6 work. You discover that experiment outcomes have been inconsistent currently, dig in, and uncover {that a} latest instrumentation change broke randomization for five% of customers in a refined manner that no person caught. An L5 would file a bug. An L6 sees the sample: that is the third instrumentation challenge this 12 months, which suggests the issue isn’t particular person bugs, it’s that we lack a testing and overview course of for instrumentation adjustments. You intend the method, get buy-in from engineering management, and implement it. Six months later, instrumentation high quality has improved throughout the corporate, and most of the people by no means know you’re the explanation why.
Right here’s what L6 appears to be like like in follow: You create an experimentation overview course of that improves high quality org-wide. It’s not simply higher documentation, it’s the implementation of a light-weight peer overview system the place any experiment over a sure dimension will get reviewed by an information scientist from one other crew earlier than launch. You write the rubric, practice the reviewers, run the primary 20 evaluations your self to calibrate requirements, and set up a suggestions loop to enhance the method. Inside two quarters, experiment high quality has measurably improved (fewer invalid assessments, higher metric choice, clearer documentation), and groups throughout the corporate are making higher product selections as a result of their experiments are extra reliable. You personally reviewed 5% of the experiments, however your framework improved 100% of them. That’s the L6 transition.
The L5 → L6 bounce is about scaling your self by means of techniques, individuals, and processes. Your work turns into extra summary: you’re optimizing how the group works, not fixing particular person issues. It might really feel much less satisfying in some methods (no person celebrates an excellent framework launch), however the leverage is extraordinary.
The Sample Throughout Ranges
Trying throughout these transitions, a transparent sample emerges: development isn’t about studying extra strategies; it’s about increasing the way you see and form the work.
L3 → L4: You increase from process completion to downside possession. The query adjustments from “Did I do what I used to be requested?” to “Did I resolve the issue fully?”
L4 → L5: You increase from downside fixing to downside definition. The query adjustments from “How do I resolve this?” to “What downside ought to we be fixing?”
L5 → L6: You increase from defining issues to constructing the techniques that assist others outline and resolve issues. The query adjustments from “What ought to our crew work on?” to “How can we make all the group simpler?”
Every stage requires you to zoom out. Every stage requires you to let go of the satisfaction of hands-on work and embrace extra summary, extra leveraged contributions. Every stage requires totally different abilities—however extra importantly, alternative ways of interested by what your job really is.
The toughest half? No one tells you this explicitly. You must decode it from imprecise suggestions about “affect” and “strategic considering” and “scope.” Firms count on you to determine that promotion isn’t about getting higher at your present stage’s recreation, it’s about noticing the brand new recreation and beginning to play it earlier than anybody asks you to.
So right here’s my problem to you: Replicate on which mindset you’re presently embodying, and which one you need to develop into.
Are you executing reliably however nonetheless ready for others to scope your work? You’re prepared to start out appearing like an L5: suggest the issue definition as an alternative of ready for it. Are you already scoping issues however solely in your personal initiatives? You’re prepared to start out appearing like an L6: search for the frameworks and techniques that will make everybody’s work higher.
Don’t anticipate a promotion to vary how you’re employed. Begin taking part in the following stage’s recreation now. Your promotion is the corporate recognizing that you simply’ve already made the shift—not allowing you to start out.
The concepts on this publish come from my new ebook, The Strategic Data Scientist: Level Up and Thrive in the Age of AI (Amazon affiliate hyperlink).
The ebook gives actionable frameworks, detailed plans, and real-world examples for making use of these concepts instantly in your day-to-day work—together with particular workouts for making every stage transition, templates for the paperwork try to be writing at every stage, and methods for working with AI instruments whereas constructing the strategic abilities that preserve you forward of automation.
If this publish resonated with you, buy the book now to be taught way more.
