Execution 3 - Data-Informed Decision Making
Google Project Management Certificate · Course 4: Project Execution - Running the Project
Tracking progress and managing quality both produce numbers. This module is about what to do with them. Data is simply a collection of facts, and data analysis is the process of collecting and organising those facts so you can draw conclusions, solve problems and support your goals. Roughly 2.5 quintillion bytes of data are created every day, so the project manager’s skill is not gathering more of it, it is picking the few numbers that actually change a decision.
The module runs in a straight line: what data is worth to you, which metrics to track, how to decide which of them matter right now, how to handle data ethically, the analysis process itself, then visualising the result and presenting it to stakeholders. The last half is the one the module’s activity exercises, since the deliverable is a stakeholder presentation built from project data.
The framing idea throughout is that data alone convinces nobody. Facts become persuasive when they are shaped into a narrative and paired with a visual, because that is what people remember.
Why data matters to a project manager
Section titled “Why data matters to a project manager”Data is already part of ordinary life: athlete stats, screen-time reports, health tracking, job sites that match you to roles from the experience and location you typed in. A runner training for a 10 km race reads pace in minutes per kilometre off a GPS watch and adjusts her training from it. Netflix watches genre, rating and repeat-viewing data points to guess what you will enjoy next. Same mechanism, different scale.
For a project manager, using data daily buys you five things:
Worked through the running Plant Pals example at Office Green: buying-pattern data showing that every best seller is a tropical plant tells you what to order from the supplier next, and tells you something real about customer preference. Tracker data on tasks completed, escalations raised and issues logged against an internal process tells you where the pain is concentrated, so you know which process to fix first.
Data does not stay useful on its own. It becomes a powerful tool only through critical analysis, application and execution - collecting it is a third of the job at most.
Data, metrics and analytics
Section titled “Data, metrics and analytics”Three words that get used interchangeably and should not be.
| Term | What it is |
|---|---|
| Data | The information itself - the numbers and feedback available about your project |
| Metric | A quantifiable measurement used to track and assess a business objective; how you measure the data. Metrics follow from goals, so they vary per project |
| Analytics | Using data to answer questions, discover relationships and predict unknown outcomes |
Once the metrics are chosen, the questions to keep asking are: what do these metrics actually mean to me, how do I want to use them, can I find patterns that let me predict, can I find something to optimise, and what lessons does this project’s data hold for the next one?
Metrics group loosely into productivity and quality, but the categories are not walls. All project data is interrelated, and the same metric applied to a different part of the project tells you something different.
Productivity metrics
Section titled “Productivity metrics”Productivity metrics measure progress and output over time, and let you track or predict how effective and efficient the team is.
| Metric | Meaning | What you ask of it |
|---|---|---|
| Milestones | An important point in the schedule marking progress, usually the completion of a deliverable or phase | How many milestones hit in a given window? |
| Tasks | An activity to be completed within a set period | What percentage finish on time, and how long do they really take? |
| Projections | A prediction of an outcome from what you know now, e.g. six months to finish with today’s resources | Are my forecasts on trends, duration, cost, performance getting more accurate? |
| Duration | Total time from start to finish, also usable per milestone | Will I meet the deadline? Broken into hours, days, weeks, months, sometimes years |
On-time completion rates are the headline number for stakeholders. High rates mean you are meeting your goals; low rates mean deadlines are slipping and something needs fixing, whether that is a process or the way you estimate scope, complexity and timeline. Tracking real task duration is what makes the next timeline estimate better. Predicting the future is impossible, but refining how you make projections is both achievable and valuable.
Quality metrics
Section titled “Quality metrics”Quality metrics relate to achieving acceptable outcomes.
| Metric | Meaning | Why it signals quality |
|---|---|---|
| Number of changes | Any difference from what was originally planned or required | Monitors risk. Small changes that compound are an early warning of a bigger problem. Keep a change log, a record of all notable changes, to explain to stakeholders why something is slow or expensive |
| Number of issues | A known, real problem that may block a task, e.g. legal approval delayed on an advert, or too few confirmed seats to secure a conference venue | Issues usually cause changes; patterns in them point at processes to refine |
| Cost variance | The difference between actual spend and budgeted spend | Low variance means you estimate budgets well. High variance means re-examine your estimating, or your expense tracking. Example: budget for 250 conference attendees, 275 turn up, the venue bills you for the extra |
Happiness and satisfaction
Section titled “Happiness and satisfaction”Google project managers use a sub-set of quality metrics called happiness metrics, covering the user’s overall satisfaction with a product or service: visual appeal, likelihood to recommend, ease of use. They are captured through a well-designed survey, or indirectly through revenue, customer retention and product returns.
A customer satisfaction score is normally a combined figure, the average of several happiness metrics. If a customer rates appearance 6/10, ease of use 7/10 and likelihood to recommend 8/10, the overall score is 7/10. What counts as an acceptable score is not for you to decide alone - agree it with stakeholders by discussing which aspects of the project matter most.
Adoption and engagement
Section titled “Adoption and engagement”Also quality-related, and easiest to remember as a party.
Every project defines its own adoption metrics. Common ones: conversion rates, time to value (TTV), onboarding completion rates, frequency of purchases, whether users leave a rating, whether they complete a profile. Engagement metrics might be the daily usage rate of a feature, or tracked orders and customer interactions.
Choosing what to track: signals
Section titled “Choosing what to track: signals”There is far more data available than is useful, so the module offers a body-temperature analogy. Normal is around 98.6 degrees Fahrenheit. Cross 100.4 and your body starts sending signals - sweating, aching, dehydration - because the brain’s job is to notice what threatens overall health. Project data works the same way.
Two ways to decide what is important:
The worked example. A manufacturer is releasing a portable home appliance by Q3, and the stakeholder’s worry is the deadline. The available data: it is Q1, you are 2,000 dollars over budget, but the burndown chart (time against work done and work remaining) puts you 30 days ahead of schedule. Look further and tasks have grown 10 percent in three weeks because stakeholders keep adding features, and the team’s productivity is falling as late nights burn them out.
The tempting signal is the budget overrun. The right signals are the ones tied to time and scope, because that is what the stakeholder said they care about. The honest read of the data: the deadline is reachable provided the feature requests stop. Productivity metrics then let you forecast how the team absorbs the extra scope at their current rate, and you communicate that forecast rather than a bare reassurance.
Data ethics
Section titled “Data ethics”You will collect data from focus groups, interviews and questionnaires, and it will often contain PII (personally identifiable information) that could identify, contact or locate someone. You then report on it to stakeholders, customers and the team, so handling it ethically protects your organisation, your project and your position.
Organisations practise data ethics to comply with regulations, demonstrate trustworthiness, ensure fair and reasonable use, minimise bias and build positive public perception. Two principles carry most of the weight.
Data privacy
Section titled “Data privacy”Proper handling of data: why it is collected and processed, privacy preferences, how personal data is managed, and individuals’ rights, all in line with the law across collection, processing, sharing, archiving and deletion. Three practical moves:
| Practice | What it means |
|---|---|
| Raise privacy awareness | Make sure every team member, plus outside vendors, contractors and stakeholders, knows your organisation’s data security and privacy protocols |
| Use security tools | Encrypted storage and password managers reduce breach exposure; in tools like Google Workspace or Microsoft OneDrive, restrict access to named individuals |
| Anonymise data | Blanking, hashing or masking identifying fields so people stay anonymous and the data can then be shared freely. Anonymise names, phone numbers, social security numbers, email addresses, photographs and account numbers |
Data bias
Section titled “Data bias”Bias creeps in through slanted survey questions, a sample group that does not represent the population, a sample that lacks inclusivity (for instance excluding people with disabilities), or even the collection method: give people too little time and rushed answers produce more mistakes and poorer data. Think about bias and fairness from the first collection through to presenting conclusions. Four common types:
| Bias | What goes wrong | Example |
|---|---|---|
| Sampling | The sample does not represent the population as a whole | Nobody over 65 included, or whole socioeconomic groups excluded |
| Observer | Different people observe the same thing differently | Stakeholders in different parts of the world read identical data and reach different conclusions |
| Interpretation | Reading an ambiguous situation as strictly positive or strictly negative when several readings are valid | Inconclusive survey results make some people anxious, and anxiety picks a side, while a calmer reader assesses several angles |
| Confirmation | Searching for or interpreting information so it confirms what you already believe | Asking only the stakeholders you already know agree with you |
Data analysis
Section titled “Data analysis”The everyday version: you are saving for a big purchase, so you build a budget, review it, and find weekly expenses exceed weekly allowances mostly through eating out. The adjustment follows from the finding. Gathering was only half the work; the analysis is what turned numbers into a plan.
The ride-share example. Analysts spot heavy midweek rush-hour demand for drivers in one city, and riders struggling to get picked up at peak times. As PM you pick the data points with the team: peak traffic times, average daily rider requests, number of available drivers. The analysis points at a solution, incentives for drivers to work the city at peak times. Drivers feel appreciated, driver supply rises, customer satisfaction rises.
Qualitative and quantitative data
Section titled “Qualitative and quantitative data”| Type | What it is | Example from the ride-share case |
|---|---|---|
| Quantitative | Statistical and numerical facts | Rider requests counted, and how they rose at specific points over a period |
| Qualitative | Subjective qualities that numbers cannot capture | User feedback about the service |
You will use both to inform decisions, make improvements and share insights. Project managers most often apply analysis to spot repeated behaviours and build a solution on the prediction that follows.
The six steps of data analysis
Section titled “The six steps of data analysis”Run through with a gym that is losing memberships.
-
Ask. Frame the analysis with key questions, starting with: what is the problem? Compare the current state of the business to the ideal state and name the obstacle, as specifically as you can, targeting the problem rather than its symptoms. “Why do we keep losing members?” is weak; “what factors are negatively affecting the member experience?” tells you what to look for. Also identify your stakeholders and confirm exactly what they are asking of you - analysing all business risk is a different job from analysing weather and seasonal risk.
-
Prepare. Collect and store the data. Here you survey members about their experience in three defined categories, upkeep of the facility, customer service and membership cost, plus a free-text field. Have an organised system for tracking and filing the responses as they come back.
-
Process. Clean the data. Enter it into a spreadsheet or another tool and strip out inconsistencies and inaccuracies: duplicate responses from the same member, biased data, typos and other errors. Cleaning is what lets you claim later decisions rest on facts and are fair and unbiased.
-
Analyse. Look closely to draw conclusions, make predictions and decide next steps. Transform and organise the data so the full scope of the result shows, and build charts to expose trends, patterns or gaps needing more research. In the gym case, 50 percent of free-text responses cite outdated equipment and 75 percent cite expensive fees. Put side by side, the two inform each other: members feel the experience is not worth the price, so investing in equipment is what protects the membership fee.
-
Share. Use data visualisation to present the findings in a form your audience can digest, and offer the insight so stakeholders can make an effective, data-driven decision.
-
Act. The business takes the insight and puts it to work against the original problem.
Telling a story with data
Section titled “Telling a story with data”The instructor’s own example runs alongside the steps: a Google Maps project to label every business in the world (restaurants, hotels, gas stations and the rest) with only a small team of engineers, presented to vice presidents from both Google Maps and Google Search, since users reach businesses through both.
Data visualisation
Section titled “Data visualisation”Three reasons a PM reaches for one: it filters, focusing the audience on the most important data points and insights; it condenses, compressing long ideas into a single image; and it sticks, aiding information processing and enhancing memory. That last one is why visualisation is an active part of storytelling rather than decoration.
Dashboards and KPIs
Section titled “Dashboards and KPIs”A dashboard displays a tight summary of metrics, stats and KPIs. A key performance indicator is a measurable value showing how effective the organisation is at reaching a key objective - it signals whether you are on track for your success criteria.
What earns a place on one:
The point is summarising, not reproducing. If the objective is a 95 percent customer satisfaction score after three months and you hold thousands of survey responses, the dashboard shows the average score and the pace towards the goal, not a spreadsheet of every response. Same with a plan holding hundreds of tasks at varying completion.
Common project charts
Section titled “Common project charts”| Visualisation | What it is | Use it for |
|---|---|---|
| Burndown chart | A line chart of time against work done and work remaining, outstanding work on the vertical axis and time horizontal | Letting the team picture how many tasks are left |
| Column chart | Bars signalling performance and progress | Comparing different activities, or progress over time, e.g. customers and plants delivered year over year |
| Pie chart | Composition, the parts-to-whole relationship | Showing what makes up the whole at a glance |
| Infographic | A concise visual summary of information, at Google a one-pager or one-sheeter, mostly graphics with supporting text | Presenting complex information quickly and professionally, especially when you will not be there to explain it |
Choosing the right chart
Section titled “Choosing the right chart”Decide what you want to show before you pick the shape. The type of data and the message together choose the chart.
| Your message | Chart | Best practice |
|---|---|---|
| Show relationships between two variables | Scatter plot - dots positioned by two variables, often with a trend line showing direction. Example: happiness score on the y-axis against life expectancy on the x-axis, rising together | Start the y-axis at 0 so the data is represented accurately |
| Compare values | Bar graph - size contrast between two or more values, also good for clarifying trends and patterns. Example: motivation by time of day, low in the morning and climbing to the end of the day | Use consistent colours, accent colours to highlight important points, and horizontal labels |
| Demonstrate composition | Pie chart - how much each part makes up the whole. Example: a day split into work, sleep, eating, commuting, television | Avoid too many categories, make the slices total 100 percent, order slices by size |
| Analyse trends and behaviour over time | Line graph - change over time, and can carry more than one series. Example: popularity of dogs against cats, both trending up, dogs always above | No more than four categories to avoid clutter, and put highly variable data at the top of the chart |
Beyond these, use visuals to show changes over time, frequency, correlations between things, and to analyse value and risk, and make them accessible so the story is understood by everyone. For tooling, project management software such as Workfront and Jira tracks activity and returns readable results on project health, while data analysis tools such as Tableau handle the visualisation side.
Presenting to stakeholders
Section titled “Presenting to stakeholders”Data alone will not convince anyone that your decisions were right or that the project mattered - an effective presentation is what conveys the work. You will present often: kickoff, weekly status updates, the final project presentation. Start by asking what do I want my audience to know, think or do as a result of this presentation, then build around the big picture and keep it simple.
Precise, flexible, memorable
Section titled “Precise, flexible, memorable”| Quality | What it takes |
|---|---|
| Precise | Identify the problem you are solving for this audience and cut anything that dilutes the narrative. Use designing for five seconds: the audience should grasp a slide within five seconds, so include only the most relevant data points and no wall of text |
| Flexible | A stakeholder may leave early or arrive late. Know what you would cut to go from an hour to 30 minutes, or to five. Preparation is what buys flexibility, and it also prevents the small stumbles that distract from the narrative |
| Memorable | Tie together the analysis, the visualisation and the narrative. Use stories and repetition. Mind your delivery: upright posture, hands at your side, raise your tone to make a point, intentional pauses, about half your normal speaking speed, short sentences, eye contact, a warm expression. Then have confidence, since the research is done |
Preparing the presentation
Section titled “Preparing the presentation”-
Get clear on the goal, and seek input. Be specific about what you want out of the meeting and frame the discussion around it. “We need two engineers who have worked in this industry” beats “we need more resources”. Check those goals with your manager or stakeholders and take their feedback before you build, and if you were invited to present, confirm exactly what the requester hopes to gain.
-
Create a delivery plan. Give every slide a headline, the single sentence it exists to make, then a couple of supporting points such as anecdotes, charts or data. Build in signposts that cue the audience to where you are going. Limit the slides in the main deck, and keep backup slides for hard questions, trade-offs and alternatives, hidden at the end until needed.
-
Respect the audience’s time. Invite only the people who need to be there, and send the deck ahead where you can.
-
Practise. Do a mock run with the team, coordinate coverage and handoffs if there is a co-presenter, and rehearse the Q&A including what you will say when you do not know an answer. Be ready to run the whole meeting alone if a co-presenter does not appear. Practise in front of a mirror or record yourself to catch awkward phrasing.
-
Prepare for surprises. Be able to summarise the key points quickly if time is cut, and to pivot the content towards whatever your audience cares about most.
-
Open well and pace it. Spend extra preparation on the introduction, where nerves peak and an early win builds confidence. Get straight to the point: state the problem, say why you are in the room and what you will cover, and lay down ground rules such as whether questions come during or after. Guide the audience through with transitions like building on this point or as I mentioned before, and read their cues to adjust pace.
-
Follow up. Send summary notes, action items and time frames. Debrief with your manager or key attendees on what they took away, what went well and what could have been better, and review next steps.
Making it accessible
Section titled “Making it accessible”You cannot communicate effectively if the audience cannot access the content.
| Area | Practice |
|---|---|
| Slide design | Clear and simple. Avoid crowding graphics, text or animation, which is hardest on people with visual or cognitive impairments. Do not let animation remove content people are still reading, and avoid flashing or flickering, which distracts and can trigger seizures. Simple can still be beautiful |
| Always use some visual | Voice-only is hard for people facing a language barrier or a hearing or cognitive impairment. Even one slide of main points helps |
| Alt text | Add alternative text to images, drawings and diagrams so screen readers can describe them. Select the object, right-click, choose alt text in Slides or PowerPoint |
| Charts | Data-heavy charts with small fonts are hard to decipher, so state the takeaway on the slide itself or in the speaker notes |
| Never rely on colour alone | Critical information needs text. Highlighting a new part of a flowchart in a different colour excludes colour-blind viewers, so add a cue such as the word new. If the deck leans on images, add a written summary at the end |
| Captions | Caption all audio and video, check YouTube auto-captions for accuracy and request a captioning service if they are poor, and use real-time captioning where available. It also helps with diverse accents, fast speakers, microphone trouble and chatty rooms |
| Contrast and text size | The difference between text and background is the contrast ratio; aim for 7:1. Bigger text is usually better, so stand at the back of the room and check you can read your slides. Avoid all capitals, which is harder for some readers including people with dyslexia |
| Share in advance | Send slides a few days ahead so people can arrange what they need, for instance following along with screen-reading software. If you cannot, send a bulleted outline, and add a glossary for acronyms and technical terms, which also helps interpreters and captioners |
Revision summary
Section titled “Revision summary”Next: Leadership & Influencing Skills → - leading a team you do not manage.