Phase 1 — Validating and cleaning the XLSX export
Before any analysis, you have to look at the data. A crucial step, often skipped.
First check: consistency of the fill-in. Sort by column and spot the anomalies. An alumnus on a permanent contract but with no industry sector? To handle manually. A €5,000 net monthly salary for an alumnus on an internship? Probably a data-entry error confusing gross and net. An alumnus doing a PhD coded as "employed"? Needs correcting.
Second check: duplicates and triplicates. An alumnus who answers twice (sometimes across two classes or two tracks) must be consolidated. Keep the latest response, which is generally the most complete.
Third check: open-ended fields. "Other, please specify…" type questions can contain answers that need recoding. "Ind. advisory" and "Consulting" and "Consulting firm" are the same category — to harmonize before producing sector charts.
This phase typically takes half a day on a class of 400. In pure Excel, it's laborious. A platform like Terrilink Surveys automatically applies consistency checks at data entry (a €5,000 net salary triggers an "are you sure?" message) that reduce this work to almost zero.
Phase 2 — Benchmark CGE indicators to calculate first
The CGE expects precise indicators. Four families are non-negotiable and must be calculated first.
Employment rate. CGE definition: alumni in professional activity (permanent contract, fixed-term contract, self-employed, civil servant) over alumni "employed or seeking employment" (so excluding further study, excluding volunteering). For engineering schools covered by the CGE and the CTI, be careful: the definitions differ slightly. See our CTI vs CGE comparison. Healthy target: 85-95% at 6 months for a business school, 90-97% for an engineering school (based on our field observations).
Share of permanent contracts. Of alumni in employment, the percentage on a permanent contract. A structural indicator of professional autonomy. Healthy target: 70-85% at 6 months, 85-95% at 30 months.
Median gross salary. Prefer the median to the mean, which is artificially pulled upward by outliers (investment banking roles, expatriate salaries). Always specify: gross annual, bonuses included or not, France or international, at 6 months or at 30 months. Compare to earlier classes using rigorously identical wording.
Satisfaction with the program. Often optional on the CGE side, but useful for the internal story. A 0-to-10 scale, with a possible NPS calculation ("would you recommend this school?"). It lets you cross-reference satisfaction and placement.
Other secondary indicators (average job-search duration, hiring sector, geography, employer size) add richness, but these 4 are the backbone of the report. Calculate first, publish first.
Phase 3 — Comparison against external benchmarks
A number on its own tells you nothing. "Our 6-month employment rate is 91%" — good or bad? It depends on context, sector, and school segment.
The most useful comparison is against the 8-12 neighboring schools: comparable size, comparable profile (generalist, specialized, engineering, management), comparable accreditations. The CGE publishes aggregated benchmarks by segment each year (top 5, top 10, generalists, etc.) that let you position yourself.
Also compare against the internal benchmark: the same class the previous year, the n-2 and n-3 classes. A stable school is one that first measures itself against itself. On class-to-class gaps, watch out for cyclical biases: a 91% employment rate in a recession year is worth more than a 93% rate in a growth year.
For response-rate benchmarks (which are not a placement indicator but do condition the reliability of the reporting), see our dedicated article on CGE survey response-rate benchmarks by school type. A response rate under 60% doesn't allow reliable conclusions at the median-salary level — to be stated explicitly in the report.
Phase 4 — Storytelling: from raw number to board slide
The most neglected phase. A board is not a technical committee: it looks at 8 to 12 slides in 20 minutes and wants a story, not volume.
Recommended structure for the board slide. Page 1: the 4 key numbers (employment rate, share of permanent contracts, median salary, satisfaction). One slide, one number per quadrant, a reference color. Page 2: the 3-year trend. Same indicators, in bars or lines. The board immediately sees the trajectory — stable, rising, falling.
Page 3: the external comparison. "Our school vs our peer group". Segmented bench, never aggregated.
Page 4: the qualitative "stories". 3 to 5 chosen verbatim quotes (anonymized) that tell a journey, a sector, a mobility. The board absorbs these stories more easily than numbers.
Page 5: the blind spots. Under-represented classes or segments, indicators falling behind, known sampling biases. This transparency strengthens the credibility of the report.
Page 6: the 2-3 decisions to make. A report with no actions is a dead report. Increase the career-center budget? Reinforce mentorship on the n-1 class (link to mentorship program)? Re-target communications on a given sector?
This 6-slide structure replaces the "thirty-odd pages with tables" that many alumni offices still produce — and that ends up printed but unread.
Classic interpretation mistakes to avoid
Three recurring traps drag down the quality of the analysis.
Sampling bias. When the response rate is low (under 60%), respondents are not representative. Alumni in a good situation answer more willingly than those in transition or unemployed. Consequence: a 92% employment rate calculated on 50% responses is probably overestimated by 4 to 8 points. To mention in the report, not to hide.
Misleading class-to-class comparisons. Comparing the 2025 class (engineering graduates with a dual curriculum) to the 2024 class (without a dual curriculum) amounts to comparing apples and oranges. Any change must be deseasonalized or compared at constant structure.
The use of the mean salary. The arithmetic mean on distributions pulled up by very high salaries (banking, top-tier consulting) inflates the result. Always use the median and specify the interquartile range (P25-P75) to show the spread. A school reporting a €56,000 mean salary with P25 = €38k and P75 = €65k tells a very different story from a school with a €56k mean and P25-P75 = €50-62k.
Automating this reporting on Terrilink
Board reporting takes on average 3 to 5 person-days a year when you start from a raw Excel export. That's time taken away from setting up mentorship, from driving dues, from engaging mobile alumni via the Network Radar.
On the Terrilink side, the reporting module automatically calculates the 4 benchmark CGE indicators, compares them to earlier classes and to external benchmarks, and exports a board-ready PDF in 8-12 pages. Plus the raw detail in XLSX if needed. To understand the full stack, see Terrilink Surveys. To get started, the fastest route is to book a demo. The full methodology is also described in our CGE survey guide.