<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jola, Thinking Like A PM on Medium]]></title>
        <description><![CDATA[Stories by Jola, Thinking Like A PM on Medium]]></description>
        <link>https://medium.com/@mojolaaaa?source=rss-a991af1b16d1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Y-0DwV8mATpxHqFWG0uVkA@2x.jpeg</url>
            <title>Stories by Jola, Thinking Like A PM on Medium</title>
            <link>https://medium.com/@mojolaaaa?source=rss-a991af1b16d1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 13:16:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mojolaaaa/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Incremental Changes vs Big-bang Releases For Software Products.]]></title>
            <link>https://medium.com/@mojolaaaa/incremental-changes-vs-big-bang-releases-for-software-products-e352c75eadbc?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/e352c75eadbc</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Thu, 17 Sep 2026 08:35:28 GMT</pubDate>
            <atom:updated>2026-09-17T08:35:28.511Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Thinking like a PM again.</em></p><p>I’ve been a loyal Piggyvest user for years. I mean, as far back as my Uni days. Like, the kind of user that refers people without expecting anything in return. If you read my tweets or my <a href="https://medium.com/@mojolaaaa/financial-illiteracy-can-be-expensive-ask-me-how-i-know-73667e2df792">articles</a>, you’ll notice I praise them a lot. I even did in my most recent article. It’s been my go-to for locking money away from myself, stacking interest, and generally feeling like I’m doing adulting correctly. For the first time ever, my favorite app. is properly frustrating me.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/590/1*JjY4XQvPXH_2yBsRh_EBpQ@2x.jpeg" /></figure><p>They just did one of those big-bang releases. I saw a couple of tweets from the marketing team hyping it up and I was pretty excited to see what was released. A couple of things are currently failing and it made me remember advising one of my former bosses against doing a big release all at once. I was very confident. He requested for a total revamp of the app to be released to the users. And one thing to take note of is how you can’t solely rely on “testing” the application. You can test something over and over in a controlled environment. Everything works beautifully. Then you release it to thousands of real people using different phones, different networks, different behaviours and different levels of patience… And the product you’ve spent ages testing, has developed a personality of its own.</p><p>That’s why I’ve started appreciating incremental releases more. Small changes being shipped regularly and watched closely. WhatsApp does this really well. Sometimes I open the app and realise something has changed, but I can’t even remember when it happened. A small feature just appears and smething gets slightly better. I think that’s the wisest way to go about it when it has to do with software products.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/590/1*1oghS-olR3n6ZIzuLa0iFw@2x.jpeg" /></figure><p>Sometimes you genuinely need to make a major change. Maybe the underlying architecture has changed. Maybe there’s a migration that has to happen. Maybe several teams or external partners have to move together. In Piggy’s situation, it was a major UI upgrade and additional features. I’m guess they wanted to catch up with the Dangote IPO FOMO.</p><p>I THINK THE REAL QUESTION SHOULD BE:</p><blockquote>What actually needs to be released all at once?</blockquote><p>You honestly don’t always need to make a big entrance. Sometimes, the best product improvements are the ones users barely notice because everything just keeps getting a little better. Most high-performing software organizations have moved decisively toward the incremental end of the spectrum while retaining the ability to execute coordinated larger changes when needed.​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e352c75eadbc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Financial Illiteracy Can Be Expensive. Ask Me How I Know.]]></title>
            <link>https://medium.com/@mojolaaaa/financial-illiteracy-can-be-expensive-ask-me-how-i-know-73667e2df792?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/73667e2df792</guid>
            <category><![CDATA[money]]></category>
            <category><![CDATA[investment]]></category>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Wed, 09 Sep 2026 08:52:31 GMT</pubDate>
            <atom:updated>2026-09-09T08:59:45.716Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Turns out greed has a syllabus too.</em></p><p>In 2021, I lost my life savings.</p><p>I was earning around ₦300,000 a month, and I knew nothing about the stock market. Not even NGX or the US stock market.</p><p>I knew that I liked saving money. I’ve always liked the feeling of having a bulk sum sitting somewhere. I think that there’s something comforting about knowing that if anything happens, you have something to fall back on. I’ve had that instinct since before I could name it.</p><p>Then a friend put me onto something “agriculture investment blah blah” I honestly can’t remember the full name or nature of the investment.</p><p>So I dropped in ₦25,000.</p><p>And then my common sense left the building.</p><p>The ₦25k started returning money. Fast. Fast enough that I sat there thinking, wait, is this actually working? Once the money doubled, I put in back the ₦50,000 I got. That grew too. Now I had a problem, and the problem was that I became greedy, and thought “if I put more money, i’ll make more money”. I put in my life savings at the time and thought I had arrived. I didn’t arrive anything. I started seeing pepper and the stars.</p><p>*blur, blur, blurship*</p><p>The returns stopped coming. Then came the excuses. There was a problem with withdrawals. They were working on it. It would be resolved.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VI07CSFm8YpHXBkm8tFFAg@2x.jpeg" /></figure><p>Eventually, I accepted that I wasn’t getting my money back. I was distraught, and because I knew I couldn’t suffer in silence, everybody knew. I told my mum. I told coworkers. I kept hoping someone would say they know about the company, and it was just a minor glitch on the app. That’s what we call “bugs” these days.</p><p>My boss at the time, Dr. Kunle, bless his heart, said something I did not appreciate in the moment. He said it was good, actually, that I lost that money this young. That it would teach me something. Lmaooo.</p><p>The way I rolled my eyes. Good keh? How is losing my entire savings good?</p><p>Years later, I think i’ve come to the understanding of what he meant. I mean, it’s cheaper than learning it now in Tinubu’s economy.</p><p>I wasn’t financially literate. I think we underestimate learning about finance early enough. I was just lucky enough to make money twice before I made the big mistake. I didn’t understand risk. I didn’t understand return. I didn’t understand that a high return should make you ask more questions, not simply put in more money. Most importantly, I didn’t understand that your savings and your investments are not the same thing. Money for your needs is not investment capital. Your emergency money is not investment capital.</p><p>I have learnt that your life savings should never become the money you use to test whether you have finally figured out how money works. I mean, I still like having money. I still like watching it grow. But i’m more cautious.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/494/1*Lrs8S5zi5tQXkdd9ZsvF3A@2x.jpeg" /></figure><h3>What You Should actually walk away with After Reading This</h3><ul><li>A result is not a strategy. One good outcome tells you almost nothing about whether the underlying thing is sound. You need repetition, not a single green number, before you call something “working.”</li><li>Speed is mostly a red flag, not a reward. If money is compounding faster than any regulated product you’ve heard of, the honest question is what’s funding that speed. Usually it’s the next person’s deposit, not a real asset.</li><li>Separate your money by job, not by mood. Savings, emergency funds, and risk capital have different jobs. The day you let one do another one’s job is the day a bad quarter turns into a bad year.</li><li>Ask people who have nothing to sell you. Every fintech, every “investment opportunity,” every friend with a link has an incentive. Find the boring, unpaid opinion before you find the exciting, sponsored one.</li><li>Losing money young is a discount, not a disaster. I paid a school fee at ₦300k a month that would have cost ten times as much at a bigger salary with bigger obligations attached to it.</li><li>Diversify across regulation, not just across assets. I now spread money across platforms that answer to actual regulators (SEC Nigeria, US SEC) instead of one platform that answers to no one.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*bdfChVobLdLvkqp-Cm-ZqQ@2x.jpeg" /></figure><h3>The products I actually trust with my money now</h3><p>None of these will promise you 100% in a month by the way.</p><p>PiggyVest is where my discipline lives, and if you know me or read my stuff, you already know this is the one I preach the gospel of. I’ve been using PiggyVest since uni. Autosave takes small amounts out on a schedule so I never even see the money to spend it. Safelock locks funds away and pays interest upfront, so I physically cannot touch it even when I want to. Flex Naira is for my controlled spending. The safe move here is simple: lock it and forget it exists.</p><p>RiseVest is for dollar plans. It puts money into ready-made US stock, real estate, and fixed income portfolios, so I’m not out here picking random stocks I don’t understand. If the goal is just “protect this money from naira wahala,” the fixed income plan is the calm option.</p><p>Ladda mixes savings with real investing, treasury bills, mutual funds, a little real estate, plus financial education from their sister company. Flexible savings is the safe lane. The locked investments need real commitment, so I only put money there that I know I won’t need before the year ends. I hear they’ll have Dangote IPO available on the app.</p><p>Hisa is where I actually buy stock. It’s regulated in the places it operates. I’ve loved my experience so far and I think they have more stock options compared to other apps. My safe approach here is small and often, not one big lump sum trying to time the market.</p><p>Bamboo is my other stock app, same fractional share setup.</p><p>Autopilot is new for me and I’m careful with it. It lets you copy the trades of certain investors, public ones included, straight into your own account.</p><p>Then there’s Fomo and Bybit. Still learning these, honestly. Crypto hasn’t always appealed to me because I think it’s stressful. I’m open to learning but it’s not intuitive like a savings app, and I’d rather say I’m new at something than fake confidence I don’t have. Ask me again next year.</p><p>I’m not a financial advisor and this isn’t advice. I’m just a girl who has made financial mistakes, and putting it out there so that you don’t.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=73667e2df792" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Whoever Holds Your Data Holds Your Trust]]></title>
            <link>https://medium.com/@mojolaaaa/whoever-holds-your-data-holds-your-trust-04a36deb634c?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/04a36deb634c</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Mon, 07 Sep 2026 19:06:50 GMT</pubDate>
            <atom:updated>2026-09-07T19:06:50.618Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM — Week 4, Day 3</p><p>Today’s focus. Compare three real Customer Data Platforms, Segment, RudderStack, and mParticle, and actually land on one instead of writing the safe, noncommittal version of this article where every option sounds equally reasonable and I never say what I’d pick.</p><p>I want to be honest about something before I get into it. I haven’t deployed any of these three in production. What I have done is sit with the actual mechanism each one uses, not the marketing page, the mechanism, and one question kept surfacing no matter which platform I looked at. Not “how fast can I set this up.” Not “how many integrations does it have.” Just: whoever is holding my customers’ data right now, do I actually trust them with it, and can I prove why.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/669/1*72JH4Me1ovJBoz4iU-n6Dg@2x.jpeg" /></figure><p>The concept, applied. All three CDPs do the same basic job, take one event and route it everywhere it needs to go. Where they genuinely differ is who holds the data on the way there, and that single fact ends up deciding almost everything else about how much I’d trust each one with something sensitive. Segment stores your data on its own servers first, then forwards it downstream, fast to set up, easy for a non-engineer to configure, but the trust is borrowed. You’re trusting Twilio’s infrastructure and Twilio’s judgment before your own customer’s data ever reaches something you control. RudderStack is warehouse-native, your own warehouse becomes the system of record, open-source, meaning the trust stays entirely in-house, at the cost of needing real engineering time to earn it. mParticle sits somewhere else again, no self-serve signup, built for regulated, high-volume shipping, and the trust there is institutional, you’re trusting a governance layer and an audit trail built for enterprises, whether or not you’re actually enterprise-sized yet.</p><p>I kept coming back to the identity resolution piece, because it’s where “who holds the data” stops being abstract and turns into something I can actually picture. All three are deterministic, none of them are guessing who a user is with some probabilistic model, they’re all following an explicit priority list, user_id beats email beats device ID. But the rulebook for that priority list lives in three completely different places. In Segment, it lives inside a dashboard I’d click through and trust was applied correctly. In RudderStack, it lives as SQL, inside my own warehouse, something I could actually open and read line by line, no trust required, just verification. In mParticle, it lives inside an enterprise console, trust delegated to a governance system built by someone else.</p><p>That’s the detail that actually made my decision for me. If I’m building for a fintech, and identity resolution decisions are the exact thing standing between a user we recognize and a user we accidentally count twice, holding the data isn’t a nice-to-have, it’s the whole question. I don’t want to trust that a vendor’s dashboard applied the rule correctly. I want to hold the rule myself. RudderStack is my winner, for now, for that specific reason. Sensitive data never has to leave infrastructure the company already controls and has already gotten audited, which means the trust never has to be borrowed from anyone in the first place.</p><p>Where this connects back. This is Day 4 of Week 3 showing up again, except now it’s not just a concept, it’s a real decision with a real cost attached. Aliasing an anonymous session to a real user_id only means something if I trust the mechanism doing it, and trusting a mechanism I can’t see is exactly the kind of borrowed trust that got the old creator-marketplace product into trouble in the first place. We used whatever the analytics tool gave us by default and never once asked who was actually holding our users’ data or why we trusted them with it. I don’t think I would have made that mistake with RudderStack specifically, not because it’s magic, but because holding the data yourself removes the question of whether to trust someone else with it at all.</p><p>I want to be honest about the tradeoff too, because picking a winner isn’t the same as pretending there’s no cost. Holding your own data means holding your own responsibility for it. RudderStack’s model sits unused, or worse, mismanaged, on a team with nobody available to actually earn that trust through real engineering discipline. A three-person startup with no data engineer yet would probably be making the wrong call picking RudderStack over Segment, and I’d tell them that directly if they asked me. Borrowed trust, managed well by a competent vendor, can genuinely beat unearned self-reliance. My winner isn’t universal. It’s specific to the kind of company I actually want to be building for, one where holding the data is worth the cost of holding the responsibility that comes with it.</p><blockquote>The takeaway. Every CDP is really just answering one question differently: who holds your customers’ data, and does that party deserve the trust that comes with holding it. For fintech, I don’t want to hand that trust to someone else by default. I want to have to earn it myself, on purpose, where I can see exactly what it cost.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=04a36deb634c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The PM’s Verification Test]]></title>
            <link>https://medium.com/@mojolaaaa/the-pms-verification-test-a9882f93b7af?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/a9882f93b7af</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Thu, 03 Sep 2026 12:59:32 GMT</pubDate>
            <atom:updated>2026-09-03T13:28:27.475Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM — Week 4, Day 2</p><p>I used to think “tracking is done” meant one single thing. Like an engineer builds something, the event gets logged, everyone moves on. Turns out that’s actually where the interesting part starts.</p><p>Here’s the simplest example I could think of. Someone opens a fintech app and makes their first deposit. Behind that one tap sits an invisible chain: the app notices what happened, an analytics tool records it, and later the product team opens a dashboard to understand how people are actually using the app. Sounds straightforward, until you realize one question decides whether any of that data is worth trusting at all.</p><p>How do you actually know the event worked? The part nobody sees. Say we’re tracking one event: first_deposit. At first glance, that feels like enough information. It isn’t.</p><p>Imagine two people make their first deposit. One deposits ₦2,000. The other deposits ₦200,000. One uses a bank transfer, the other a debit card. The dashboard still just says “2 first deposits.”</p><p>Same number, completely different behavior sitting underneath it, invisible. That’s why an event needs context, not just a name. Think of the event as the headline and the parameters as the rest of the article.</p><p>Instead of seeing this:</p><p>first_deposit</p><p>You’d rather see:</p><p>Amount: ₦15,000</p><p>Payment method: Bank transfer</p><p>Plan type: Flexible Savings</p><p>Platform: Android</p><p>Now you’re answering real product questions instead of just counting clicks. The sentence that changed how I think. Imagine someone on the engineering team says, “KYC Completed has been implemented.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*fWkVobmv_eFxzVQjxUXbKA@2x.jpeg" /></figure><p>A few weeks ago, I would’ve heard “implemented” and quietly translated it to “done.” Now I hear something else entirely, and a handful of questions show up automatically.</p><p>Did it fire at the right moment?</p><p>Did it happen exactly once?</p><p>Did it carry the right information with it?</p><p>Can we actually see it, right now, in the analytics tool?</p><p>That shift feels subtle, but I think it’s one of the habits that quietly separates a PM who consumes a dashboard from a PM who can actually challenge one.</p><p>The scariest tracking bug</p><p>This is the one that stuck with me. A user completes KYC once. Analytics records it three times. Nobody notices right away. A week later, someone presents a funnel: “Users are completing KYC multiple times.” Now the team starts solving a problem that never actually existed.</p><p>The opposite version is worse.</p><p>The user completes KYC, and nothing gets recorded at all. The dashboard quietly reports fewer completions than actually happened, and the roadmap shifts to fix a drop-off that isn’t real.</p><p>The bug was never loud.</p><p>It was just believable, and that’s exactly what makes it dangerous.</p><p>If I were reviewing this as a PM</p><p>If an engineer told me tracking was ready, here’s what I’d actually want to know before I called it done:</p><p>Does the event name match the tracking plan?</p><p>Does it fire at the correct moment?</p><p>Does it fire exactly once?</p><p>Are the right details attached to it?</p><p>Is it linked to the right user?</p><p>Can we actually see it in the analytics tool ourselves?</p><p>None of those questions require me to write a single line of code. They just require me to already know what good tracking is supposed to look like.</p><p>The thing I’m taking away</p><p>I started this lesson thinking Firebase was just another analytics tool to learn. I finished it with a different mental model entirely. An event existing isn’t enough. A PM should be able to trust that the event happened, happened exactly once, carried the right context with it, and actually made it into the data people are about to make decisions from.</p><p>Because the easiest dashboard to believe is always the one that’s quietly wrong.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a9882f93b7af" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Client-Side vs Server-Side]]></title>
            <link>https://medium.com/@mojolaaaa/client-side-vs-server-side-66aa6dcbc080?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/66aa6dcbc080</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Wed, 02 Sep 2026 09:15:53 GMT</pubDate>
            <atom:updated>2026-09-02T09:15:53.930Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM · Week 4, Day 1</p><p>I kept staring at one event today. Withdrawal Completed. It was the same name, same event, completely different meaning depending on who says it.</p><p>What even annoyed me about it was that i built a tracking plan for a fintech app. I spent time choosing event names that made sense because I thought the hard part was deciding what deserved to be tracked.</p><p>Turns out naming events has actually been one of the easiest part of this process. I’m begging to ask myself “Who gets to say what happened?”</p><p>The Client OR The Server?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gYG23qJh-hfNGMDIaLCOZw@2x.jpeg" /></figure><p>Imagine i open my bank app, type ₦50,000, and press Withdraw. My phone obviously reacts to it instantly. If my phone could talk, it’d confidently say withdraw completed easily. Meanwhile, somewhere on the backend, it’s checking balances or still updating records. It hasn’t agreed with whatever action i just performed on my phone yet. Imagine the transaction fails on the backend. My balance won’t move, my dashboard has already miscommunicated to me.</p><p>Client-side tracking isn’t trying to tell me the final outcome. It’s standing beside the user, watching what they attempted to do. Server-side tracking has a harder job. It waits. It speaks after the receipt exists.</p><p>I think for those working on fintech apps made this lesson stick so quickly. A shopping app can survive a few messy clicks and get off the hook easily.</p><p><strong><em>A banking app doesn’t get that luxury.</em></strong></p><p>Nobody wants a dashboard celebrating revenue that never arrived. In fact, i’ve heard people saying they don’t like using dark modes for their fintech apps. They want to see their money clearly. 😅</p><p>The moment I tap Withdraw, that’s a client-side event. The app witnessed my intention. The moment ₦50,000 actually leaves my account, that’s a server-side event. The system finally has proof.</p><p>Same journey. Different responsibilities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*2NhKbwJ98rJmIDpjKZP2_w@2x.jpeg" /></figure><p>A week ago, I would’ve looked at an event list and asked, “Did we capture everything?”</p><p>Now I catch myself asking a different question.</p><p>“Who’s telling this part of the story?”</p><p>I didn’t expect client-side vs. server-side tracking to change how I think. I thought I was learning where events come from. I was actually learning that good product decisions depend on knowing which voice deserves to be believed. And I have a feeling that’s one of those habits I’ll carry long after I forget today’s documentation.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=66aa6dcbc080" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Wait…. Who Is This User?]]></title>
            <link>https://medium.com/@mojolaaaa/wait-who-is-this-user-fec48607c7f3?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/fec48607c7f3</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Sat, 29 Aug 2026 09:04:07 GMT</pubDate>
            <atom:updated>2026-08-29T09:04:07.043Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM — Week 3, Day 4</p><p>Days ago, I was talking about how important it is to name your events properly. The whole idea of doing that was to make sure if I search for withdrawal, I should be able to find everything related to withdrawal without becoming a detective.</p><p>While I learnt about event tracking, I found out finding the event is only half of the problem. What if i found the event, but I don’t actually know who did it?</p><p>Imagine Chidinma opens the saving app on her phone. She browses the savings plans. She starts onboarding. She signs up.</p><p>She closes the app.</p><p>Later that evening, she opens the app again on her laptop and completes her KYC. Then she uses her phone to fund her savings plan.</p><p>From a normal human perspective, that’s one person. Which is Chidinma. But analytics doesn’t automatically think like a human. It sees events. And those events can arrive with different identifiers attached to them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/1*txw7NNqXhozbLzJkdeSaXA@2x.jpeg" /></figure><p>Before she signed in, her activity could be associated with something like:</p><p>anonymous_4821</p><p>After signing in, it could become:</p><p>user_88213</p><p>Then she logs in on another device and suddenly there is another identity attached to the activity. Same person. Different records.</p><p>What happens when the data knows what someone did, but doesn’t know that all those actions came from the same person?</p><p>That is where identity resolution comes in.</p><p>The basic idea is to connect those separate pieces of activity back to one user.</p><p><strong>So instead of analytics seeing:</strong></p><p>User A browsed savings</p><p>User B started onboarding</p><p>User C signed up</p><p>User D logged in on laptop</p><p>User E funded a savings plan</p><p>it can eventually understand:</p><p><strong><em>Chidinma browsed savings → started onboarding → signed up → logged in on another device → funded her savings plan.</em></strong></p><p>And this is where yesterday’s article connects to today’s lesson. Yesterday, I was concerned about whether I could find the right event. Today, I’m concerned about whether I can connect the right events to the right person. Because imagine having the perfect event taxonomy.</p><p>Everything is beautifully named.</p><p>Savings Plan Viewed</p><p>KYC Completed</p><p>Bank Linked</p><p>Savings Plan Funded</p><p>You search for them and everything is sitting exactly where it should. Lovely, but if those events aren’t tied to the right user consistently, you’re still building your analysis on shaky ground. And this matters because of the questions we’re eventually trying to answer.</p><p>What did users do before they activated?</p><p>Did the people who completed KYC eventually fund a savings plan?</p><p>How many users who started onboarding actually became retained users?</p><p>Those questions aren’t really about individual events anymore. They’re about a journey. And you can’t understand a user’s journey if you’re accidentally treating one person like five different people.</p><p>I think this is another one of those things that sounds very technical when you first hear it, but the PM question underneath it is actually quite simple:</p><p>Do I trust that the data is describing one user’s behaviour as one user’s behaviour?</p><p>Because if I don’t, then my funnel can be wrong.</p><p>My retention can be wrong.</p><p>My activation analysis can be wrong.</p><p>And I could end up fixing a problem that doesn’t actually exist.</p><p>So I’m starting to see analytics instrumentation differently.</p><p>First, I need to be able to find the behaviour. Then I need to be able to understand the details of that behaviour. And then I need to know whose behaviour I’m actually looking at.</p><p>Yesterday was the Search Test.</p><p>Today, I think the test is simpler:</p><p>Can I follow one user from their first interaction to their latest one and still know it’s the same person?</p><p>If I can’t, I probably don’t understand my data as well as I thought I did.</p><p>And honestly, that’s a slightly scary thought for a PM.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fec48607c7f3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Search Test]]></title>
            <link>https://medium.com/@mojolaaaa/the-search-test-4735d002124e?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/4735d002124e</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Mon, 24 Aug 2026 14:34:27 GMT</pubDate>
            <atom:updated>2026-08-24T14:34:27.939Z</atom:updated>
            <content:encoded><![CDATA[<h4>Jola Thinks Like A PM — Week 3, Day 3</h4><p>Today is all about the “Search Test”. Imagine at exactly 1:47 PM, on a hot afternoon, your oga sends a message on your Whatsapp group or Slack channel: “Can you pull every event tied to withdrawals? I need it before the 2 PM standup.”</p><p>It’s obviously a completely normal request. I wrote about Mixpanel and Amplitude the other day. It’s very easy to know about these tools until it’s time to start tracking events and you can’t find results because nothing is where it’s supposed to be.</p><h4>Transfer Completed OR Completed Transfer</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*8vXjmCEh5Ly35FbGd0Oz8w@2x.jpeg" /></figure><p>These are two words that have the same meaning but will give you an entirely different outcome. Most PMs won’t take the object-first rule at face value. So let’s break it down.</p><p><strong>Your first guess will probably be:</strong></p><ul><li>Withdrawal Started</li><li>Withdrawal Completed</li><li>Withdrawal Failed</li></ul><p>If you searched for “withdrawal”, you’ll definitely get these three results in just one search.</p><p><strong>If you had written the exactly like this:</strong></p><ul><li>Started Withdrawal</li><li>Completed Withdrawal</li><li>Failed Withdrawal</li></ul><p>Everything will scatter.</p><p>Remember that Started is sitting under <strong>S.</strong> Completed disappears beside Completed Transfer, Completed KYC, and every other “Completed” event in the product. Failed has gone to live under <strong>F.</strong> Same product. Same user behaviour. Three searches instead of one. This entire thing is just about making sure information is about to be retrieved, not about making sure it sounds neat. The funny part is that most of us will still fail our own tests.</p><p>If I was asked to check an event on the Piggyvest app, trust me, the first thing i’ll type is the action first. “Fund Savings Plan” I’ll do it with no hesitation until I check myself.</p><p>Bad taxonomy is extremely dangerous. It’ll never feel like you’re doing the wrong thing at first. But the headache that’ll come with it in the future when you’re asked to find the event you named on a random Tuesday afternoon is nothing compared to what you’ve ever experienced at a job.</p><p>I’ve lived this before. I don’t think i’ll ever get tired of writing about it. The creator marketplace I worked on had events everywhere, but nobody had agreed on how they should be named. Everyone followed whatever felt was right. Infact, the only thing tracked was “number of paying users”.</p><p>Now that i’m looking back, I honestly couldn’t tell if we were object-first or action-first. We just kept on making up different things sprint by sprint. Which means we had a very bad taxonomy or we didn’t even have a taxonomy at all. We only tracked habits, which is able to drift at any point in time.</p><p>Every tiny decision you make can lead to either an analytical workspace your team trusts or otherwise. So, i’m learning and promising myself to always do the object naming first.</p><p>When someone searches for withdrawal, they should be able to find the whole history on the first try. No one wants to do the work of a detective. One search is enough. That’s the Search Test.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4735d002124e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mixpanel vs Amplitude: Understanding Product Analytics Tools]]></title>
            <link>https://medium.com/@mojolaaaa/mixpanel-vs-amplitude-understanding-product-analytics-tools-a5bdb7137990?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/a5bdb7137990</guid>
            <category><![CDATA[data-as-a-product]]></category>
            <category><![CDATA[data-product-manager]]></category>
            <category><![CDATA[product-data]]></category>
            <category><![CDATA[product-data-management]]></category>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Wed, 19 Aug 2026 09:35:46 GMT</pubDate>
            <atom:updated>2026-08-19T09:35:46.060Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Jola Thinks Like A PM • Week 3, Day 2</em></p><p>I started writing late in the day today, and i’m not sure I enjoy writing when I am tired.. I’ll be exploring that out today. Writing is easily one of the best ways for me to think through problems. I remember going to school early just to write out everything i’d learnt the previous night on the board because I wanted to make sure I understood everything i’d read. If i could explain it on a board without looking at my notes, It meant it stuck. Fun times… hahaha.</p><p>I like to write these intros because I think they’re like little ice breakers to ease you guys into the serious part of whatever I get to write. So, I hope you guys enjoy it. 🤷🏽‍♀️</p><p>Anyway, today i’m geeking out about Mixpanel and Amplitude. One thing you should remember while reading this is: Product analytics turn user behaviour into something you can investigate.</p><p>It’s a simple sentence but there’s quite a lot of things happening in it. A product doesn’t just hand you an insight. A user does something, a product manager captures the behavior, attaches a useful context to it, analyzes it and then, hopefully learns something that helps that PM make better decisions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*TmlBHApo1GYcc1v94rGVHw@2x.jpeg" /></figure><p>I’m going to be thinking about it like this:</p><p><strong>(QBEPAID)</strong></p><p><strong>Question → Behaviour → Event → Property → Analysis → Insight → Decision</strong></p><p>It is easily one of my favorite mental models because it’s there to remind me that my dashboard isn’t where thinking starts for me. The question comes first up until the decision making part.</p><p>If you have ever worked in a product role, you’ve probably heard something like:</p><p>“We should check Mixpanel”</p><p>“Amplitude can give us that”</p><p>Both platforms are widely used for products analytics. They mostly help you understand how users make use of your products. so, you might be wondering, which Mixpanel versus Amplitude, what sets them apart?</p><p>They both have a lot in common. You can check for your cohort analysis, funnel, retention analysis, user segmentation, event analysis, behavioral insight. The difference is in the needs of your team and how you use them.</p><h4>Mixpanel</h4><ul><li>Mixpanel’s strength is mostly used understanding specific user behaviors. You can easily ask: “how many users want to use feature X?” or “where are users dropping off in this particular journey?”</li></ul><h4>Amplitude</h4><ul><li>Amplitude also does this but, it has positioned its strength around product intelligence. It helps the product team connect behavioral data around things like: feature adoption, retention, user journeys, experimentation, and product decisions.</li></ul><p>You can have either of them and still report terrible insights. If your product doesn’t properly track important user actions, your analytics tool has very little to work with. It’s called BAD EVENTS.</p><h3>What Should a Product Manager Look For?</h3><p>As a PM, what matters most is that the tool can answer your questions, that your team trusts the data, that non-technical stakeholders can use it easily, that it fits smoothly into your product flows, and that it helps scale the business.</p><blockquote><strong>Today’s Takeaway</strong></blockquote><blockquote>It should never be a battle between dashboards. There’s nothing like Mixpanel vs Amplitude. And whether you choose Mixpanel, Amplitude, or another analytics platform, the principle remains the same: A sophisticated analytics tool cannot compensate for poor questions or poor data. The real value from product analytics is when you press further.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a5bdb7137990" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Metric Is Only as Good as the Events Behind It.]]></title>
            <link>https://medium.com/@mojolaaaa/a-metric-is-only-as-good-as-the-events-behind-it-a8b9e921d46f?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/a8b9e921d46f</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Mon, 17 Aug 2026 16:03:51 GMT</pubDate>
            <atom:updated>2026-08-17T16:03:51.728Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM | Week 3, Day 1</p><p>So before we go into today’s article, let me tell you guys what frustrated me the most about writing this article. It took me forever to come up with the most ideal infographic for this article. It was so annoying, you guys have no idea.</p><p>But let’s jump right into it. Last week, I spent quite a bit of time learning about metrics. DAU. MAU. Retention. Activation. LTV. etc.</p><p>There were a lot of definitions to remember and honestly, I could explain most of them very well, along the line, at least for what i’ve planned to learn, I knew I had to go deeper into what I needed to know. I could know what a metric meant, understand how it was calculated, and still not fully understand what was happening behind the scenes.</p><p>So today, I went one layer deeper into where these numbers actually come from because if I tell you that a product has a 62% activation rate, that number didn’t just appear on a dashboard, something happened before I ever saw 62%.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*RuaiIGRE5NE14afy6WDYWA@2x.jpeg" /></figure><p>Users did things. They signed up, they completed KYC, linked their bank account, they made a deposit, they created a savings plan.</p><p>Those actions can be captured as events, and that led me to one of the things I learned today: An event tells you what happened, a property tells you more about what happened.</p><p>When a transfer is completed, that’s an event.</p><p>But I want to know how much was transferred, which bank was involved, what currency was used, was the transfer successful, was this the user’s first transfer? Those details are the properties.</p><p>Imagine a dashboard tells you: 50,000 transfers were completed today. That’s great, right? What if 20,000 of those came from existing users, what if transfers above 100,000 Naira were failing more often, what if android users were experiencing more failures, What if one particular bank had an unusual failure rate? 50,000 is still accurate, but it doesn’t tell you very much by itself, the interesting questions are hiding in the context.</p><p>I thought of my role of a product manager in all of this, I don’t necessarily need to be the person the tracking code, but I do need to be clear about what I’m trying to understand about the product. if I want to understand why users are activating or not, I need to understand what they are doing before they activate.</p><p>For the usual example I always use(Piggyvest) the journey looks something like:</p><p>Sign up → KYC → Bank linked → First deposit → Savings plan created. I can now start asking; where are people dropping off and WHY?</p><p>An event can tell me someone abandoned KYC. But it’s not going to tell me why. It won’t tell me the process was confusing, it’s not going to tell me if the app crashed, maybe the user changed their mind, maybe the user didn’t have the required document or maybe do not have an idea on how to use the app. You can always show me the behavior, but it will always tell me the entire story. This was what I found most interesting about today’s learning.</p><p>I spent the last couple of weeks learning how to interpret numbers. Now I’m beginning to understand that a lot of things that happen behind the scenes that determines how useful it can be. If we captured the wrong behavior, we can build a beautiful dashboard, calculate the metric correctly, and still end up asking the wrong question.</p><p>So going forward, I think I want to approach Product analytics a little differently. Instead of asking questions like ”what should we track?”</p><p>I will be asking: “what are we trying to understand?”. Then from here, I’m going to be working backwards.</p><p>Maybe moving forward, I’ll ask better questions that will help me understand what behavior led to an action, what event captures that behavior, what properties would give enough context to understand what happened? And finally, what can we actually do with what we have learnt from this behavior?</p><p>So, beecause I’m beginning to realize that product analytics isn’t about collecting more data, it’s about collecting the right evidence, I’m going to be approaching product analysis differently.</p><blockquote><strong>Today’s Takeaway.</strong></blockquote><blockquote><em>Metrics tell you what is happening. Events help you understand what users actually did. And if you do not capture the right behavior, the number on the dashboard can be perfectly accurate and it will lead you no where or to the wrong conclusion.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a8b9e921d46f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I Built a Metrics Dashboard from Scratch]]></title>
            <link>https://medium.com/@mojolaaaa/i-built-a-metrics-dashboard-from-scratch-1f434482a71c?source=rss-a991af1b16d1------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f434482a71c</guid>
            <dc:creator><![CDATA[Jola, Thinking Like A PM]]></dc:creator>
            <pubDate>Tue, 11 Aug 2026 11:30:36 GMT</pubDate>
            <atom:updated>2026-08-11T11:30:36.174Z</atom:updated>
            <content:encoded><![CDATA[<p>Jola Thinks Like A PM — Week 2, Day 4</p><p>I spent yesterday working on a framework, in a full spreadsheet with formulas that calculate off each other instead of using numbers that I typed in manually. What I had to do was stop learning metrics for a bit because it’s very easy for me to learn on paper, and not actually learn anything. I didn’t want a situation where I will be unable to apply all I have learnt in the real world.</p><p>So what I did was, I built out a Fintech savings framework for a make-believe piggyvest app, organized it across the customer journey I learned from AARRR. Acquisition, activation, engagement, retention, revenue, referral. What I did in total was to come up with 15 metrics in total. Each one has a proper definition, not just a name and the number. It has the numerator, denominator, time period, user population, data source, and the owner of the actual metric.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*12Hfia2bgrkZYE0XV0R5cA@2x.jpeg" /></figure><p>While I was working on this, I found myself asking a question. I’ve never really thought about deeply before: when a business has a data stored overtime, how is the data documented so business users can go back and compare results. I was thinking about how businesses store the data overtime, I reached out to my mentor who has an experience with working with a couple of notable companies or businesses. So I asked him a very simple question. How do you think businesses keep information so teams can compare performance over time. For example, if I want to know what retention look like three months ago, compared to what it looks like now, how does a random business owner go about it?</p><p>I know that businesses are not just about asking questions like what is our retention today? It’s more about comparing retention overtime.</p><p>What changed?</p><p>What metrics improved?</p><p>Which one got worse?</p><p>What was happening in the business when they changed?</p><p>I was curious about what the historical view actually looked like in practice. And then he explained how things have really changed, unlike before that you had to report in excel sheets you have things that you could easily just write SQL for to give you that data. But since I was working with just a framework on a spreadsheet, I created something with four of the metrics live instead of type in singular metrics independently.</p><p>Stickness ratio pulls straight from DAU and MAU rows. So this doesn’t exist on his own on my spreadsheet. It’s simply DAU divided by MAU, recalculating the moment either of those changes. LTV pulls from ARPU. LTV:CAC pulls from LTV and CAC. Payback Period pulls from CAC and ARPU. None of this are random numbers I just get to decide, they are consequences of the other numbers on the sheet.</p><p>LTV:CAC came out to 5.03x.</p><p>On its own, that looked healthy. It was comfortably above the 3x benchmark I was using as a reference for customer economics. If I’d stopped there, I probably would have closed the laptop thinking everything related to revenue economics was in good shape. But Payback Period, built from the same underlying numbers/formula, came out behind target. It will take almost three months to recover the cost of acquiring a user, against a target of two. It was all making sense to me now.</p><p>The number isn’t the insight. The relationship between the numbers is.</p><p>LTV:CAC and Payback Period are built from overlapping input, but they are not answering the same questions.</p><p><strong>LTV:CAC will ask you:</strong> is this customer eventually worth more than it cost us to acquire them?</p><p><strong>Payback Period will ask you:</strong> how long do we have to wait before we recover what we have spent acquiring them.</p><p>Your business can look good on the first question and still have a problem with the second question. And if I only built one formula instead of a framework, I would’ve never seen that. I also noticed that the framework caught something individual formulas would never catch on their own. As I’ve learnt over the week, the framework doesn’t just hold one metric it holds the whole shape of the business, and one metric affects the other.</p><blockquote><strong>Biggest Takeaway Today: </strong>A metric gives you a number. A framework gives you context. And the relationship between the numbers is often where the real insight is.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f434482a71c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>