A data team roadmap is one of the most reliable diagnostic artefacts you can read. Whether you are an executive sponsoring the team, a peer leader trying to integrate with them, or an investor evaluating a company's data maturity, the roadmap tells you a lot about whether the team will deliver on what they have promised.
Here are the patterns I look for.
Properties of roadmaps that deliver
The roadmap is denominated in user-visible outcomes, not in internal initiatives. A line item that reads "ship the forecasting dashboard for finance" is denominated in outcomes. A line item that reads "modernise the data platform" is denominated in scope. Outcome-denominated roadmaps are testable. Scope-denominated roadmaps are not.
The roadmap names the customer for each item. Every line has a named team or operator who has agreed that the item is useful to them. Items without a named customer are usually internal priorities that have not been negotiated with the rest of the business.
The roadmap distinguishes between new work and maintenance, and protects time for both. A roadmap that is one hundred per cent new work is fictional. A roadmap that is one hundred per cent maintenance has stopped delivering. Healthy roadmaps run sixty to seventy per cent new work and reserve the rest for keeping existing systems alive.
The roadmap has fewer items than the team has people. Most struggling roadmaps fail by overloading. A team of eight with fifteen committed initiatives in a quarter is not going to ship fifteen things. They are going to ship four and hold sixteen partial deliveries. A roadmap with five committed initiatives, ranked, leads to five things shipped.
Properties of roadmaps that will not deliver
The roadmap reads like a vendor pitch. Items are named after technologies or products. "Snowflake migration." "Dbt adoption." "Implement Looker." This is a roadmap that is proud of its tooling and unable to articulate what users will be able to do once the tooling is in place.
The roadmap has too many strategic initiatives. Strategic initiatives are valuable, but they tend to expand. A team running four strategic initiatives in parallel is a team that has not chosen.
The roadmap has no measurable outcomes. The criteria for success are described in adjectives, not in numbers. "A modernised platform" rather than "a platform on which the finance team can model their close in under four hours." Adjectival success criteria cannot be falsified, which means the work cannot be evaluated.
The roadmap has a heroic Q4. Q1 and Q2 have a few items. Q3 has more. Q4 has everything that did not fit anywhere else. This is a roadmap whose author has not made the cuts the calendar will eventually force them to make. The cuts are better made now.
What to ask the team lead
A short list of questions that surface a lot.
Which item on this roadmap are you least confident you will ship. The answer is informative either way. If they cannot name one, they have not stress-tested the plan. If they can, the conversation about why is worth having early.
Which item on this roadmap was added because a stakeholder insisted, against your judgement. Every roadmap has at least one. The lead's awareness of which one is a sign of how politically aware the team is.
What did the team ship last quarter, and which item produced the most user-visible value. The answer reveals whether the team has any feedback loop with its users. Teams without a feedback loop produce roadmaps that drift further from business value each quarter.
What is the team's maintenance load, in person-days per week. A team that cannot quantify this is a team whose new-work plan is fictional, because they do not know how much capacity is actually free.
The synthesis
A good data team roadmap is short, customer-named, outcome-denominated, and honest about maintenance. A good data team lead can tell you which items are at risk and why. The combination of these two artefacts predicts delivery better than any other diagnostic I have seen.
Bad roadmaps are not always the lead's fault. Sometimes the lead has produced a clear plan and been overruled by a stakeholder coalition. The diagnostic still applies, but the intervention is at a different level. Either way, the roadmap is the artefact you read first.