Data Driven Decision Making: Insights from a Marketing Lead

Data driven decision making is the future. 

If you want to stay competitive, data driven decision making (DDDM) is the only way to run your business. No more gut feelings, no more risky hunches. If you have data, you should be using it to make decisions that actually work. And if you don’t have data, you need to start collecting it now.

Why should you trust my expertise?

I was the first marketing hire at Sapiengraph, and have since risen to a ‘head of marketing and growth’ role, leading a team of six. 

I’ve picked up all my data skills on the job. I learned how to do basic SQL queries, and built many Looker Studio dashboards from scratch to track our growth.

And the company? We’ve had 200% growth every year since 2019, all bootstrapped. No VCs, no investors. Just results. That track record speaks for itself. 

Why you need data driven decision making in your business

If there’s one thing I learned from Steven, the CEO of our company, it’s how to be productively lazy. 

I don’t mess around with processes that make life more difficult, more manual, or more time-consuming. If it can be automated, I automate it. If I can streamline it, I will streamline it. Once you implement data driven decision making, you get the following:

Streamlined processes

Data driven decision making eliminates endless guesswork and ‘gut feelings.’ You’re working off data, not hunches. This means fewer debates, faster decisions, and ultimately, better outcomes. Set up your data pipelines now, and you’ll thank yourself later. 

We keep a running list of every failed marketing attempt, especially campaigns, with the data to back up why they tanked. That list is open to the whole company. 

Anyone can dig in, see what’s been tried, and either build on it or understand why it’s dead in the water. We don’t need to keep telling people why we’re not doing something because this list exists. It saves time and prevents repeats of past mistakes. 

Fact-based decisions

Andrew McAfee put it well: “We’re too quick to override the computers, even when their answer is better.” We think we know better, but we’re wrong more often than we think. 

Data-driven decisions remove ego from the equation. They’re not based on overconfidence or guesswork. They’re rooted in facts. The numbers will tell you what is working and what doesn’t.

For example: We paid $250 a month to sponsor a couple of YouTubers. They brought in 20k in revenue. In contrast, we spent $1800 a month on email marketing. It brought in $3.7M. 

Sure, $250 is a small amount when compared to 20k in revenue, but the data shows that the ROI for email marketing is vastly superior for us. (BTW, these numbers aren’t real. Not trying to get fired.)

Cost savings & problem solving

Data itself is valuable, but doing the proper analyses is even more valuable. You can see the moment that something stops or starts working, so it’s easier to catch issues and cut out processes that waste money, effort, and time. 

We keep detailed logs of each campaign that we’ve tried, and the successes and failures that came with it. It’s tedious, yes, but these systems reduce unnecessary effort, keep growth steady, and solve problems before they blow up.

I don’t want my team to waste time reinventing failures. I want them to focus on what hasn’t been proven to fail yet, which means their energy goes toward things that are more likely to work. 

That’s how you keep productivity high and resources focused where they matter. 

7 steps, and 12 questions you should be asking.

Step 1: Define your objectives

Start with a clear question or objective. Ask yourself this question: What is my problem?

This is something we’re very firm about. I’ve even got a compilation of our CEO Steven emphasizing this.

Here’s my problem: Every time we hire someone new, they come to me saying things like “I think we should do this. I think we should do that” but they don’t tell me what they’re looking to solve. 

Nine times out of ten, when I prod them on the ‘why’, they will raise a goal or a problem that we’ve tried to solve in the past. Great initiative, not so good on the execution. 

Example: “We should be active on Quora”. Why? To generate traffic? It doesn’t work. How do I know? We tried it and didn’t get many conversions so we put it on the backburner. The data shows that being on Quora isn’t going to generate us a lot of traffic. 

Now, if the goal is to make sure no one impersonates us on Quora, then yes, us creating an account and posting there would make us credible and make sure no one could impersonate us. 

Every suggestion needs to address a specific problem, presented as Problem → Suggested Solution. No exceptions. 

Second question you need to ask yourself: When does it need to be solved?

Not everything is urgent. Some objectives, like hitting 1M followers, can take years. Others, like fixing customer satisfaction issues, might need attention now.

Prioritize by urgency and importance. Is this a problem for today, next month, or five years from now? For example, data shows building a credible social media presence takes 1-5 years, with a smaller ROI than email marketing. So, prioritize emails that drive responses over posting on X.

If it’s urgent, focus on getting data fast. 

Step 2. Identify relevant data sources

Once you have your goal or your problem, figure out how to collect your data. You cannot solve what you cannot measure. If it’s not measurable, figure out a way to measure it. If it doesn’t relate back to solving a problem you have, ignore it.  

Third question: Where do I get this data?

You have two sources: internal and external. Internal sources give you data your business captures naturally during daily operations. 

Think sales numbers, customer counts, leads your team is chasing, pageviews, bounce rates, or inventory costs. These are your quantitative data, cold hard numbers. We track these through tools like Redash and PostHog.

But qualitative data is just as critical. It’s where the insights often live. Customer feedback and interviews reveal bottlenecks and UI issues we’d never catch otherwise. For example, exit surveys and email check-ins with subscribers have highlighted key areas for improvement.

For employees, we use Google Forms for regular check-ins to spot internal friction points. And my team knows they can always approach me directly with problems, so we can tackle them together. Numbers matter, but context and clarity come from listening to the people behind them.

External sources refer to data that comes from outside your company. Things like market research reports, social media, industry conferences, trade shows, government and regulatory reports, academic publications, review platforms, financial reports, news media. 

Market research reports can give you insights about your industry. It can show you what your competitors are doing and how the market is evolving. 

If you need extra B2B data about your customers, or new hires, you can get that from 3rd party data providers like Sapiengraph. We provide firmographic data for companies and enrich details about employees such as past employment and education.

Our sales team uses our own product to build leads lists, while others have used Sapiengraph to source for VC deals.

Fourth question: Is this the correct data?

It’s tempting to hoard each and every piece of information you get, but that’s a one-way street to getting overwhelmed and looking at the wrong data. 

Every time you get tempted to do something new, check your goals and your plans. If it’s in the plan, save it for later. Much later. If it’s not in the plan, forget it. 

Step 3: Collect and organize data

People talk about collecting and organizing data but they’ve got it backwards. You need to organize before you collect. 

This fifth question is one that many startups fail to ask themselves: How will someone else find this in the future?

Why? Because data is useless if you can’t even find it. Before you start hoarding files, figure out your system. Where are you storing it? How are you naming it? What categories do you have?

Obviously, as a tech company, we’ve got a couple of private servers, but you don’t need fancy infrastructure on day one. A shared folder in the cloud works fine. 

Messy data organization is like a debt with 50% interest—it compounds fast and cripples progress. You think you’re saving time now, but future-you and your team will pay for it tenfold.

Decentralize data ownership. Folders should be shared, set permissions early on and have a process for handing over access whenever someone joins the team. 

My entire writing and marketing team has access to all the files, folders, and analytics for both Proxycurl and Sapiengraph. It makes it easier for the writers and marketing team to collaborate on content without needing a constant back and forth on “Can you export a list of these users?”. 

We also have a transparent company structure and a list of who’s in charge of what, so every writer and marketing associate knows who to ask for help.

Once you’ve got all that out of the way, ask yourself this: How do I collect the data?

Go back to your initial objective and think about what data can help solve the problem or achieve goals. 

If your goal is to figure out which industries are your best customers, you should have a CRM like HubSpot or Salesforce to keep track of deals, win rates, deal sizes, sales cycle length, and lead sources. 

Don’t have money for either? Then you’ve got to do manual data entry using Google Sheets or Excel then use formulas to calculate the metrics. 

If your goal is to boost your engagement, then you need to be tracking what’s happening on your website and your pages–things like web traffic, time spent on pages, API usage (if any), email open rates, link clicks, and churn rates. 

Companies on a budget can consider just using Google Analytics. It’s free. Personally, we use Posthog because it accounts for adblockers and it shows us what our web visitors are doing on our site, like rageclicking. 

This one in particular is one of our devs testing a new feature. Wonder what got him all riled up?

Step 4: Ensure data quality

This is where many businesses stumble. Throwing everything into a report and expecting results is wishful thinking. The ‘easy’ part of data-driven decision-making—collecting data—is done. Now comes the harder question: Do I have good data?

This is what I learned through years of marketing experience: If your data is flawed, your results will be too. 

Great data needs to be accurate (up-to-date), complete (no gaps), and unique (no duplicates). But let’s face it—perfect data doesn’t exist right away. Getting from bad data to great data takes significant effort, and starting with poor-quality data only makes the process harder.

The solution is simple: Get better sources. That’s why the team prioritizes accurate and fresh data when growing Sapiengraph and our sister product, Proxycurl. That’s why we also made templates that will help you enrich missing data directly in your spreadsheets. It’s all about giving you the cleanest, most actionable data to start with.

Unfortunately for everyone, no data is perfect. Next question: How can I make my data better?

The answer is data preprocessing. You have to pre-process data to ensure a smoother data driven decision making process. 

Here are the three things to look out for:

• Missing values: Either fill them in, or exclude them. If you have 1000 customers but only the location data for 500 of them, your ‘top 10 countries’ world map is going to be very biased.
• Duplicates: Figure out a way to deduplicate them while retaining the latest information. Otherwise, you’ll end up sending the five emails to the same customer. 
• Outliers: Just like Spiders Georg shouldn’t have been counted because he ate 10000 spiders, you shouldn’t include your biggest subscriber in your regular dataset. They’re going to skew your data.

Step 5: Analyze the data for insights

Data analysis is where the raw numbers start to tell a story. 

Look, I’m not much of a coder, and we have two kickass devs at the company whose job is specifically to write queries on Redash and create metrics for us to study. They’re the ones who come up with the formulas needed to track complicated business metrics like

• lifetime value
• customer acquisition cost
• customer segmentation and so on.

But if you’re a founding marketer with no coding chops like me, start with the basics like Excel or Google Sheets. There’s no shortage of YouTube tutorials teaching you how to use formulas and create charts. And hey, if that fails, throw it to ChatGPT, Claude, or hire someone off Fiverr.

This is where the next question comes in: How to analyze my data?

Here are where things go wrong pretty quickly if you don’t know what you’re doing. Some of the data is easy to analyze. If you have a data table that looks like this:

Then you can easily turn that into a chart that looks like this:

And you can simply say “Let’s focus on locking down the American market”. 

But not all data can be easily analyzed or visualized like this. There are a lot of different analytical techniques out there. 

For example, using Python or R to turn a bunch of customer IDs, lifetime values and customer satisfaction scores into a PCA analysis showing you that customers from the tech industry buy more and report higher satisfaction. 

Once you’ve got your data cleaned up and analyzed, you should ask: What is my data telling me?

Taking the PCA plot from the above as an example. We know that our customers from tech-related industries are pretty happy with us. The same is less true for those in retail. That’s a problem that we now need to solve because we want ALL of our customers to be happy regardless of industry. 

Step 6: Make data-driven decisions

Hoarding metrics does nothing. You must act on those metrics. Knowledge is nothing without action. Ask yourself, what is my next move?

Now that I know the retail subgroup of our clients aren’t too satisfied, I’m in the process of touching base and asking them why. But how do we get people to talk to us? 

Our A/B email test played out with formal emails getting fewer replies than casual emails. The data shows that people prefer casual emails. 

It could be due to the rise of AI, where people feel like formal speech is more likely to be generated by AI, whereas informal emails are more likely to have a person on the other side of it (There is. It’s me. I write the emails).

So now that I have solid data that people want informal language, I went ahead with it. Our sales and marketing teams all use less formal language when engaging with customers and the results prove themselves.

It gets people to tell us their needs and wants so we can focus on fixing problems. 

Step 7. Monitor and refine

Pooja Agnihotri, wrote in their book, 17 Reasons Why Businesses Fail : Unscrew Yourself From Business Failure, “The purpose of data is to learn on time what is working and what is not and take any corrective actions according to that.”

After you’ve made some data-driven decisions, you have to keep an eye on them .They need constant evaluation to measure their success and ensure they stay relevant. 

Which brings us to the final question: Is this working?

Every team should have measurable, achievable metrics to see if what they’re doing is actually working. 

Either you need objectives and key results (OKRs) or key performance indicators (KPIs). 

For the marketing team, this can look like achieving a certain amount of clicks on an article, getting ranked on Google search ASAP, driving down customer acquisition cost through better and more efficient marketing, and decreasing the lead to customer conversion rate.

Whenever we try something new, like posting certain kinds of content, we have to set goals for it. Example: If I post a YouTube video about our applicant tracking template, does it increase our page views? And do people on those pages convert to customers? 

If my video has 1000 viewers, and too few of them head to our webpage to subscribe, then we need to figure out what’s going wrong and if the effort is not worth the investment. 

Always revisit your initial goals to see if your OKRs and KPIs are getting you closer to those goals or solving those problems. 

If your data driven decisions are working, you should be seeing less churn, higher engagement, better employee satisfaction, and higher retention rates in both customers and employees.

Tools to get started in data driven decision making 

Making the switch to data driven decision making is not easy, I won’t lie to you. Even if you follow all the steps above and reflect on the answers to each question, implementing proper data governance and infrastructure is difficult. 

The key is to begin with tools that are easy to implement. My CEO Steven wrote a full guide on the sales tools we use to hit 1.2M ARR in 12 months, but I’ll give you a quick run down, specific to our marketing and sales teams.

Google Forms

Job applications and 1-on-1s are done through Google Forms. No fancy ATS systems. The Google ecosystem is pretty great and sends all applications directly to the Google Sheets, where we can easily keep track of applicants and feedback.

Google Sheets 

Everyone has this nowadays. This is also why we built a Google Sheets extension for Sapiengraph. It’s a familiar environment for a lot of people and there are a lot of tutorials available on how to use it. If you get stuck anywhere, ChatGPT or Claude can help you solve issues. 

It’s easy to use, accessible, and powerful enough for basic storage and analysis. It can do everything Excel does and can help with simple data visualization tasks. And if you need a formula it doesn’t have, you can probably write a script for it. Or get Claude to write a script for it. 

Sapiengraph 

Of course we’re going to use our own product. Since we’re already in Google Sheets most of the time, it makes sense to use our own custom formulas to prospect for leads and potential hires. It’s how we got one of our marketing managers

Get key firmographic data from us. We use our own product for recruitment, marketing and sales too.

All you need are the LinkedIn URLs of your clients/customers and you can do your data enrichment directly in your Google Sheets. 

Posthog 

We use Posthog for a lot of things. Mostly to track customer behavior on our webpages. Recently, one of my team noticed that lots of people were coming to one of our pages but only the average session duration was less than a minute. 

Data showed us that there was a problem. We later discovered that one of our videos had broken and a quick reupload later, people were back to scrolling the entire page.

Redash 

If you’re not new to coding and writing queries, you can integrate Redash with your web analytics or sales software and easily create visualizations to track your metrics. We use it to get an overview of our sales and marketing costs and ROI.

Sendy

The marketing team uses this to send bulk emails at a fraction of the cost of traditional services like Mailchimp. For me, it’s a great affordable outreach automation tool. It’s also great when it comes to collecting data because it allows us to create specific customer lists, track email open rates, click-through rates, and campaign performance.

Mistakes I’ve learned from 

We’ve already covered data quality–inaccuate, incomplete, biased data gets you terrible results. But that’s not the only problem that you’re going to face when you pivot towards data driven decision making. 

Skill gaps

Not everyone in your company is going to be a data expert. Do not expect them to become a whiz overnight, even with training. Be realistic about what your team can do.

If it’s urgent, hire a consultant. Or bring in an external data analyst. Outsource your analyses to another company. Let those pros lay the groundwork, clean your data, and implement systems while your team gets used to the new data infrastructure. 

Meanwhile, invest in upskilling and promoting data literacy. Hold workshops or provide employees with the time and finances to take data literacy courses. 

It takes time to cultivate a data-driven culture in a company, so be patient. Remember, it’s not just about hiring data scientists; it’s about getting everyone in the team to think with data. It’s about creating an environment where everyone can act with confidence because they understand the numbers that drive the business.

Biases

You might think that data is free from bias. It’s not. Nothing is bias free. Not completely. 

Take for example, survivorship bias. You look at a company out there and say “Hey, they’re doing this thing right now! If we do the same, we will also be successful!” 

Unfortunately, most businesses you see are survivors. For every thriving business you see, there are at least 10 more that never made it. You just don’t know about them. Sometimes the data you collect is only available because it’s the data of the survivors. 

Regardless of how perfect your data is and how much you know, it’s likely that some part of you will still be clouded by some kind of bias. 

There’s the availability bias, the tendency of humans to stop searching for more information when they think they have enough. The confirmation bias, the tendency of people to look for things they already know, thus only noticing data that fits their preconceived notions. Anchoring bias, when the first piece of information you find becomes the baseline for all your adjustments. 

In the words of our CEO Steven: “I don’t even trust myself. That’s why I want to quickly put something to market and listen to what customers say or do.”

No matter how good a data driven decision making process is, these biases are going to creep in. Which is why data driven decision making is a continuous and iterative process that undergoes constant refinement.

“One of the first things taught in introductory statistics textbooks is that correlation is not causation. It is also one of the first things forgotten.”

― Thomas Sowell, The Vision of the Anointed: Self-Congratulation as a Basis for Social Policy

Where I see DDDM in the next few years 

So where do we go from here?

Artificial intelligence: If you’re not on it, you’re losing out. We’ve used AI to enhance the lead generation process. The developers at Sapiengraph get free Cursor plans (an AI code editor), completely sponsored by the company. Machine learning and AI both will be a huge part of future data driven decision making strategies. I suggest you start learning how. 

And as the world becomes more connected, more people are going to want to disconnect. Privacy will be sacred so have plans to deal with stricter privacy laws. 

Finally, prescriptive analytics will become a thing. Predictive analytics have been around for a while, but they’re evolving quickly. The future will include prescriptive analytics—the ability to not just predict what will happen, but to prescribe the best course of action in accordance with real-time happenings. 

“With data collection, ‘the sooner the better’ is always the best answer.”

Wise words from Marissay Mayer, former Yahoo! CEO. 

So don’t wait any longer. Take these free templates and start your data driven decision making process now.