Trend Analysis in EVM: Forecasting Project Performance

<div class="ge-article-wrapper"><nav class="ge-toc" aria-label="Table of contents"><p class="ge-toc-label">In this article</p><ul class="ge-toc-list"><li><a href="#why-trend-matters-more-than-any-single-value">Why Trend Matters More Than Any Single Value</a></li><li><a href="#the-diagram">The Diagram</a></li><li><a href="#how-to-build-a-trend-analysis">How to Build a Trend Analysis</a></li><li><a href="#worked-example-35m-highway-improvement">Worked Example: £35M Highway Improvement</a></li><li><a href="#trend-analysis-for-spi">Trend Analysis for SPI</a></li><li><a href="#the-confidence-test-for-eac-forecasting">The Confidence Test for EAC Forecasting</a></li><li><a href="#common-mistakes">Common Mistakes</a></li><li><a href="#frequently-asked-questions">Frequently Asked Questions</a></li></ul></nav><article class="ge-article-body"><p>Trend analysis in earned value management means plotting <a href="/en/earned-value/definitions/cost-performance-index">CPI</a> and <a href="/en/earned-value/definitions/schedule-performance-index">SPI</a> over time to identify patterns, trajectories, and inflection points. A single month's CPI of 0.92 tells you almost nothing. Is it recovering from 0.85? Declining from 0.98? Stable for six months? The trend is the story. The single number is just a frame from the film.</p><p>Trend analysis is part of the <a href="/en/earned-value/definitions">earned value definitions glossary</a>. For the full CPI and SPI framework, see the <a href="/en/earned-value/cpi-spi">CPI and SPI formula page</a>.</p><h2 id="why-trend-matters-more-than-any-single-value">Why Trend Matters More Than Any Single Value</h2><p>Here's something I've learned from sitting in hundreds of progress meetings: the absolute value of CPI gets all the attention, but the direction of travel matters more.</p><p>A CPI of 0.88 with a six-month upward trend from 0.78 is a recovering project. A CPI of 0.95 with a three-month downward trend from 1.02 is a project heading for trouble. Same ballpark number. Completely different commercial reality.</p><p>Project boards fixate on the current month. The commercial manager's job is to make them look at the trajectory. That requires trend data, not snapshots.</p><h2 id="the-diagram">The Diagram</h2><pre class="ge-ascii-diagram ge-anim">12-Month CPI Trend: £35M Highway Improvement ============================================== CPI 1.05 ┤ │ ○ 1.00 ┤─ ─ ─ ─○─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ Budget Line │ ○ ○ 0.95 ┤ ○ ○ │ ○ ○ 0.90 ┤ ○ │ ○ 0.85 ┤ ○ │ ○ 0.80 ┤ ← Current CPI │ 0.75 ┤ └┬────┬────┬────┬────┬────┬────┬────┬────┬────┬────┬──── M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 │← Mobilisation →│← Stable →│← Decline begins →│← Decline │ noise phase (trend break) continues │ Annotations: M3-M5: CPI stabilises around 0.98 (healthy) M6: Peak at 1.02 (substructure phase complete) M7: Drop to 0.94 (steel erection starts, new subcontractor) M8-12: Steady decline to 0.80 (subcontractor underperformance) 3-Month Rolling Average: M6: (0.95+1.02+0.98)/3 = 0.983 M9: (0.90+0.87+0.85)/3 = 0.873 ← 11.2% decline in 3 months M12: (0.82+0.81+0.80)/3 = 0.810 ← Still declining Verdict: Deteriorating trend. Not stabilising. Intervention required.</pre><p>That chart tells a story no single number can. The project was healthy through month 6. Something changed at month 7 (new subcontractor for steel erection). The decline has been steady for six months with no sign of stabilisation. This isn't a blip. It's a systemic problem.</p><h2 id="how-to-build-a-trend-analysis">How to Build a Trend Analysis</h2><h3>Step 1: Collect Monthly Data Points</h3><p>At each reporting period, record:</p><ul><li>CPI (cumulative)</li><li>SPI (cumulative)</li><li>CPI (current period only)</li><li>SPI (current period only)</li></ul><p>Cumulative values smooth out noise. Current period values show what happened this month specifically. You need both.</p><h3>Step 2: Calculate Rolling Averages</h3><p>Single-month values are noisy. A concrete pour that slips by a week can swing monthly CPI dramatically. Rolling averages smooth the noise and reveal the underlying trend.</p><div class="ge-table-wrap ge-anim"><table class="ge-table"><thead><tr><th>Average Type</th><th>Window</th><th>Best For</th></tr></thead><tbody><tr><td>3-month rolling</td><td>Short</td><td>Detecting recent trend changes</td></tr><tr><td>6-month rolling</td><td>Medium</td><td>Establishing the underlying trajectory</td></tr><tr><td>Cumulative-to-date</td><td>Full project</td><td>Overall project health</td></tr></tbody></table></div><p>The 3-month average is your early warning system. If the 3-month CPI average is declining for three consecutive periods while the 6-month average is stable, you're seeing the start of a new trend that hasn't yet pulled the longer average down.</p><h3>Step 3: Identify the Pattern</h3><pre class="ge-ascii-diagram ge-anim">Common CPI/SPI Trend Patterns =============================== STABLE: ─────○──○──○──○──○──○──○───── Values within +/- 0.03 for 4+ months. Reliable basis for EAC forecasting. DECLINING: ○──○──○──○ ○──○──○──○ Steady downward movement. Investigate cause. Don't wait for recovery – it rarely happens naturally. RECOVERING: ○──○──○──○ ○──○──○──○ Upward movement after intervention. Confirm with 3+ months of data before declaring recovery. VOLATILE: ○ ○ ○ ○ ○ ○ Swinging by 0.05+ month to month. Unreliable for forecasting. Investigate data quality. STEP CHANGE: ○──○──○──○ ○──○──○──○ Abrupt shift (new phase, scope change, re-baseline). Split the analysis at the break point.</pre><h2 id="worked-example-35m-highway-improvement">Worked Example: £35M Highway Improvement</h2><span class="ge-worked-label">Worked Example</span><div class="ge-callout ge-anim"><p><strong>Scenario:</strong> A £35M NEC4 Option C highway improvement in West Yorkshire. The commercial manager presents the monthly CPI trend at the month 12 review.</p><div class="ge-table-wrap ge-anim"><table class="ge-table"><thead><tr><th>Month</th><th>CPI (cumulative)</th><th>CPI (period)</th><th>3-Month Avg</th><th>6-Month Avg</th></tr></thead><tbody><tr><td>1</td><td>0.91</td><td>0.91</td><td>--</td><td>--</td></tr><tr><td>2</td><td>0.93</td><td>0.95</td><td>--</td><td>--</td></tr><tr><td>3</td><td>0.95</td><td>0.98</td><td>0.947</td><td>--</td></tr><tr><td>4</td><td>0.97</td><td>1.01</td><td>0.980</td><td>--</td></tr><tr><td>5</td><td>0.98</td><td>1.00</td><td>0.997</td><td>--</td></tr><tr><td>6</td><td>1.00</td><td>1.02</td><td>1.010</td><td>0.978</td></tr><tr><td>7</td><td>0.98</td><td>0.94</td><td>0.987</td><td>0.983</td></tr><tr><td>8</td><td>0.96</td><td>0.90</td><td>0.953</td><td>0.975</td></tr><tr><td>9</td><td>0.94</td><td>0.87</td><td>0.903</td><td>0.953</td></tr><tr><td>10</td><td>0.92</td><td>0.85</td><td>0.873</td><td>0.930</td></tr><tr><td>11</td><td>0.90</td><td>0.82</td><td>0.847</td><td>0.900</td></tr><tr><td>12</td><td>0.87</td><td>0.80</td><td>0.823</td><td>0.863</td></tr></tbody></table></div><p><strong>Trend reading:</strong></p><ul><li>Months 1-6: Steady improvement. Project mobilising well. CPI crossed 1.0 at month 6.</li><li>Month 7: Inflection point. Period CPI dropped from 1.02 to 0.94. Something changed.</li><li>Months 7-12: Continuous decline. Six months of deterioration. 3-month average fell from 1.01 to 0.82.</li></ul><p><strong>Root cause:</strong> Steel erection subcontractor mobilised at month 7. Productivity significantly below programme rates. Labour shortages, plant breakdowns, rework on connections. The subcontractor issue isn't resolving.</p><p><strong>The EAC implications:</strong></p><ul><li>EAC using cumulative CPI (0.87): £35M / 0.87 = <strong>£40.23M</strong> (+£5.23M overrun)</li><li>EAC using 3-month CPI (0.82): £35M / 0.82 = <strong>£42.68M</strong> (+£7.68M overrun)</li><li>The gap between these two EACs (£2.45M) represents the difference between assuming the historical average will continue and assuming the current deterioration rate continues</li></ul><p><strong>The commercial manager's recommendation:</strong> Present the range. The 3-month trend is more representative of current performance than the cumulative average, which is diluted by the healthy early months. The honest forecast is closer to £42M than £40M. If the subcontractor doesn't improve by month 14, replace them.</p></div><h2 id="trend-analysis-for-spi">Trend Analysis for SPI</h2><p>Everything above applies to <a href="/en/earned-value/definitions/schedule-performance-index">SPI</a> too. But SPI trend has an additional nuance: the convergence problem. SPI always trends towards 1.0 as the project approaches completion, regardless of actual delays. See the <a href="/en/earned-value/definitions/schedule-performance-index">SPI page</a> for the full explanation.</p><p>For schedule trend analysis beyond 60-70% completion, switch to SPI(t) from earned schedule. SPI(t) doesn't converge and gives you an honest trend right through to project completion.</p><h2 id="the-confidence-test-for-eac-forecasting">The Confidence Test for EAC Forecasting</h2><p>Trend stability determines which <a href="/en/earned-value/definitions/estimate-at-completion">EAC</a> formula to use:</p><div class="ge-table-wrap ge-anim"><table class="ge-table"><thead><tr><th>Trend Pattern</th><th>EAC Formula</th><th>Rationale</th></tr></thead><tbody><tr><td><strong>Stable CPI (3+ months)</strong></td><td>EAC = BAC / CPI</td><td>Current efficiency will continue</td></tr><tr><td><strong>Declining CPI</strong></td><td>EAC = BAC / (3-month CPI)</td><td>Recent performance is more representative</td></tr><tr><td><strong>Volatile CPI</strong></td><td>Bottom-up ETC + AC</td><td>Data is unreliable for formula-based forecasting</td></tr><tr><td><strong>Step change</strong></td><td>Recalculate from break point</td><td>Pre-change data no longer relevant</td></tr></tbody></table></div><p>Don't use the cumulative CPI for EAC if the trend is declining. The cumulative is dragged up by historical good months that are no longer representative. The 3-month average is closer to reality. That's an unpopular message, but it's an honest one.</p><h2 id="common-mistakes">Common Mistakes</h2><ol><li><strong>Reporting a single CPI without context.</strong> "CPI is 0.92" is incomplete. "CPI is 0.92, down from 0.98 three months ago, with a 3-month average of 0.90 and declining" is useful. Context turns data into intelligence.</li><li><strong>Ignoring early warning signs.</strong> A CPI that drops from 1.02 to 0.96 in one month doesn't look alarming. It's still above 0.95. But if you're tracking the trend, you see the first break after five months of stability. That's the moment to investigate, not after three more months of decline.</li><li><strong>Averaging across unlike phases.</strong> If Phase 1 was groundworks with CPI of 1.05 and Phase 2 is steel erection with CPI of 0.82, the cumulative of 0.93 is meaningless. It doesn't represent either phase. Segment your trend analysis by project phase or work package where the nature of work changes significantly.</li><li><strong>Not plotting the trend visually.</strong> Numbers in a table are hard to read. A chart showing CPI over 12 months with the 3-month rolling average is instantly understandable. I've seen project directors glance at a trend chart and immediately ask the right question before anyone's opened the slide deck.</li></ol><div class="ge-product-note ge-anim"><p><strong>How Gather helps.</strong> Gather's AI reads your site diaries daily and maps progress against your cost-loaded programme, giving you accurate earned value data without manual spreadsheet updates. <a href="https://gatherinsights.com/contact">Book a demo</a> to see it working on a live NEC4 project.</p></div><h2 id="frequently-asked-questions">Frequently Asked Questions</h2><h3>How many data points do I need before trend analysis is reliable?</h3><p>Minimum six months of data for any meaningful trend. Below that, you're looking at mobilisation noise. The <a href="/en/earned-value/eac-etc-tcpi">EAC formula page</a> recommends waiting for three to four months of CPI stability before using formula-based forecasts, but for trend direction you want at least six.</p><h3>Should I track period CPI or cumulative CPI?</h3><p>Both. Cumulative CPI shows the overall project health. Period CPI shows what happened this month. The period values reveal trend changes faster because they're not smoothed by historical data. But period values are noisier, so always look at them alongside the rolling average.</p><h3>What causes a sudden CPI step change?</h3><p>Common causes: new subcontractor mobilising, phase transition (e.g., from substructure to superstructure), major scope change, re-baselining after compensation events, or a change in the way costs are reported. When you see a step change, investigate the cause before assuming the new level will persist.</p><h3>Can AI automate trend analysis?</h3><p>Yes. Plotting CPI/SPI over time, calculating rolling averages, and flagging threshold breaches is exactly the kind of repetitive analysis that benefits from automation. Gather's <a href="/en/ai-construction/quantity-surveyors">AI tools for quantity surveyors</a> can process monthly EVM data and surface trend anomalies automatically, freeing the commercial team to focus on interpretation and action.</p></article></div>