<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Dorcas Akinnibosun · Blog</title>
    <link>https://dorcas-akinnibosun.vercel.app/blog</link>
    <atom:link href="https://dorcas-akinnibosun.vercel.app/blog/rss.xml" rel="self" type="application/rss+xml" />
    <description>Notes on Power BI, Excel and data analysis from Dorcas Akinnibosun.</description>
    <language>en</language>
    <item>
      <title>Modelling $55.8M of sales in Power BI with a star schema</title>
      <link>https://dorcas-akinnibosun.vercel.app/blog/modelling-55-8m-of-sales-in-power-bi</link>
      <guid isPermaLink="true">https://dorcas-akinnibosun.vercel.app/blog/modelling-55-8m-of-sales-in-power-bi</guid>
      <pubDate>Fri, 23 Oct 2026 09:00:00 GMT</pubDate>
      <category>Power BI</category>
      <description>How five raw tables became a four-page Power BI report for a global electronics retailer: the star schema, the DAX comparisons and what the model revealed.</description>
      <content:encoded><![CDATA[<article><p>Maven Electronics is a multinational electronics retailer with several years of transactions across products, customers, regions, sales channels and stores. The goal was to find patterns in revenue, spot the strongest products and regions, understand who buys, and compare store performance.</p><p>The finished report is <a href="https://dorcas-akinnibosun.vercel.app/work/maven-electronics-sales#live">live on this site</a>. This post is about what happens before the first chart: the model.</p><h2>Five raw tables</h2><p>The data arrived as five separate tables: sales, customers, stores, products and exchange rates. Together they cover 26.3K orders, 15.3K customers and 67 stores.</p><p>Raw files are never ready to chart. I cleaned and shaped each table in Power Query first, so problems are fixed once, at the source, rather than patched inside individual visuals.</p><h2>Why a star schema</h2><p>I modelled the tables as a star schema: the sales table in the middle, holding the transactions, with customers, stores and products around it as lookup tables.</p><p>That shape does three things for a report like this:</p><ul><li><strong>Filters behave predictably.</strong> Picking a store, a product category or a customer group filters the sales through one clear path.</li><li><strong>Measures stay simple.</strong> Totals, averages and comparisons are written once and work on every page and every filter.</li><li><strong>It scales.</strong> Adding a page or a slicer doesn't mean rebuilding relationships.</li></ul><h2>Time intelligence</h2><p>Most business questions are comparisons: this year against last year, year to date, the top performers this period. I used DAX time-intelligence measures (year-on-year, year-to-date, prior year and Top N) so every KPI card shows its comparison automatically, whatever the reader filters to.</p><p>That is how the overview can say revenue reached $55.8M, up 12.1% year on year, and still give the right comparison when you filter to a single region.</p><h2>What the model made visible</h2><p>With the model in place, the patterns were quick to find:</p><ul><li><strong>Revenue peaked in 2019 and declined from 2020</strong>, while the average unit price held steady at $356.8.</li><li><strong>Thursdays and Saturdays drive orders</strong>, with Saturday the single busiest day.</li><li><strong>Loyal customers are 84.6% of the base.</strong> That is strong retention, but also a sign the business may be saturating.</li><li><strong>The United States earns $23.8M; France brings in $1.2M.</strong></li><li><strong>Physical stores generate 79.55% of revenue</strong> and online only 20.45%, which leaves room to grow the online channel.</li></ul><p>Each of those became a recommendation in the <a href="https://dorcas-akinnibosun.vercel.app/work/maven-electronics-sales">case study</a>: promotions on peak days, a better online checkout, localised marketing for weaker regions, tiered loyalty rewards and stock planning for the winning categories.</p><h2>The takeaway</h2><p>The visuals get the attention, but the model decides whether they can be trusted. A clean star schema is why this report can answer a new question in a click instead of a rebuild.</p></article>]]></content:encoded>
    </item>
    <item>
      <title>What 2,300 finance job postings say about the market</title>
      <link>https://dorcas-akinnibosun.vercel.app/blog/what-2300-finance-job-postings-say</link>
      <guid isPermaLink="true">https://dorcas-akinnibosun.vercel.app/blog/what-2300-finance-job-postings-say</guid>
      <pubDate>Fri, 23 Oct 2026 09:00:00 GMT</pubDate>
      <category>Market insight</category>
      <description>Excel still tops the skills list, 81% of roles want a certification and only about 2.7% of applicants get hired. Findings from my live Power BI report.</description>
      <content:encoded><![CDATA[<article><p>People choosing a career in finance, and the recruiters hiring them, make big decisions with very little market data. So I built a four-page Power BI report from 2.3K finance job postings covering 2024 to 2026: 2.2K hiring companies and 467.6K applicants.</p><p>The report is <a href="https://dorcas-akinnibosun.vercel.app/work/finance-job-market-intelligence#live">live on this site</a>, so you can filter every number below by country, experience level or industry yourself. Here is what stood out.</p><h2>Excel is still the most requested skill</h2><p>Excel appears in 626 postings. That is ahead of budgeting (613), Python (604), forecasting (583) and financial modelling (573). Power BI (570) and SQL (551) are close behind.</p><p>The interesting part is how close those numbers are. Finance teams don't seem to be choosing between classic finance tools and data tools any more. They want both in the same person.</p><h2>Competition is intense</h2><p>467.6K applicants and 239.9K interviews produced 12.6K hires. That is a hiring success rate of about 2.7%.</p><p>If you are applying, the maths says a lot of good applications will still be turned down. Volume alone won't fix that. The next two findings point at what might.</p><h2>Certifications matter more than years</h2><p>81% of postings ask for a certification. Meanwhile, average salary barely moves with experience: about $76.7K at entry level against $77.5K at senior level, a difference of roughly $0.7K.</p><p>In this data, the role you hold and the skills you can prove explain pay much better than time served. Average salaries across the dataset range from $59.3K (minimum) to $94.7K (maximum).</p><h2>Hiring takes about a month</h2><p>The average time to hire is 33 days. Retail is the slowest at 33.9 days and investment the quickest at 30.9. Manufacturing has the most hiring activity.</p><p>For recruiters, that spread is small enough that process, not industry, is probably where time is won or lost. The recruitment page of the report breaks it down by company and industry if you want to dig in.</p><h2>Data analyst is the top finance role</h2><p>Data analyst leads the list with 212 postings, ahead of credit analyst, internal auditor, risk analyst and tax analyst. Work modes (onsite, hybrid, remote) and employment types are split almost evenly.</p><h2>How the report is built</h2><p>A fourth page documents the data and the model, because a dashboard is only as trustworthy as what sits underneath it:</p><ul><li><strong>Type validation</strong> in Power Query before anything else.</li><li><strong>A date table</strong> for time intelligence, so month-over-month KPIs compare like with like.</li><li><strong>Skills normalised into their own table.</strong> One posting can ask for many skills, so skills sit in a bridge table instead of a comma-separated column. That is what makes &quot;most requested skill&quot; a real count rather than a guess.</li><li><strong>Light and dark themes</strong> with a toggle on every page, so it works on a projector or a late-night laptop.</li></ul><h2>Try it yourself</h2><p>Open the <a href="https://dorcas-akinnibosun.vercel.app/work/finance-job-market-intelligence#live">live report</a>, pick your country in the slicer and see whether the national picture matches the global one. If you are hiring, start on the recruitment page.</p></article>]]></content:encoded>
    </item>
    <item>
      <title>Why every dashboard I build is interactive</title>
      <link>https://dorcas-akinnibosun.vercel.app/blog/why-every-dashboard-i-build-is-interactive</link>
      <guid isPermaLink="true">https://dorcas-akinnibosun.vercel.app/blog/why-every-dashboard-i-build-is-interactive</guid>
      <pubDate>Fri, 23 Oct 2026 09:00:00 GMT</pubDate>
      <category>How I work</category>
      <description>A static chart answers one question. An interactive dashboard lets each person ask their own. How I design reports people can click through.</description>
      <content:encoded><![CDATA[<article><p>Most data portfolios are a gallery of screenshots. Mine isn't, on purpose: every dashboard I build is interactive, and three of them are <a href="https://dorcas-akinnibosun.vercel.app/#live">published live in Power BI</a> on this site. Here is why that matters, and how I design for it.</p><h2>A static chart answers one question</h2><p>A screenshot shows the answer to the question its author had in mind. But a branch manager wants their branch, a director wants to compare regions, and a recruiter wants one industry. With a static report, each of those is a new request back to the analyst.</p><p>An interactive dashboard lets each person ask their own question. The slicers, KPI cards and charts update together, so filtering to one store or one year changes the whole page, not just one visual.</p><h2>Order pages the way people ask questions</h2><p>I lay out report pages in the order people tend to ask: what is happening, then why, then what to do about it. My <a href="https://dorcas-akinnibosun.vercel.app/work/maven-electronics-sales">Maven Electronics sales report</a> runs from an executive overview, to sales trends, to products and customers, to stores and regions. Each page answers the question the previous one raises.</p><p>Page navigation, bookmarks and drill-through do the rest. You can start at the overview and drill into a single store without hunting for the right tab.</p><h2>Excel can be interactive too</h2><p>Not every team runs on Power BI. For <a href="https://dorcas-akinnibosun.vercel.app/work/metrobank-customer-insights">MetroBank</a> and <a href="https://dorcas-akinnibosun.vercel.app/work/nexa-workforce-analytics">Nexa</a> I built the dashboards in Excel with Power Query, pivot tables and slicers. They ship as working workbooks, so people can filter them in the tool they already use every day.</p><h2>What &quot;interactive&quot; doesn't mean</h2><p>It doesn't mean more charts. Every slicer is another thing a reader has to understand, so I only add filters that match a real question someone will ask, and I keep the default view useful on its own. If nobody touches a single slicer, the page should still tell the main story.</p><p>It also doesn't replace the write-up. Every case study ends with findings and recommendations in plain language, so technical and non-technical readers leave with the same message.</p><h2>Try one</h2><p>Load a <a href="https://dorcas-akinnibosun.vercel.app/#live">live report</a>, pick a slicer and watch the page respond. That is the experience I'm designing for: you asking your own question, and the data answering it.</p></article>]]></content:encoded>
    </item>
  </channel>
</rss>