Skip to main content
other_ratings uses two scales — always check the dimension type before displaying Three dimensions (business_outlook, ceo, recommend_to_friend) are approval ratios between 0.0 and 1.0. The remaining five are scores on a 0–5 scale. Build a lookup dict and apply the correct formatting for each type: 
RATIO_DIMS = {"business_outlook", "ceo", "recommend_to_friend"}

dim_lookup = {r["type"]: r["value"] for r in emp["other_ratings"]}

for dim, value in dim_lookup.items():
    if dim in RATIO_DIMS:
        display = f"{value * 100:.0f}%"        # ratio → percentage
    else:
        display = f"{value:.1f} / 5.0"         # score → star scale
    print(f"  {dim}: {display}")
ceo_rating on individual reviews is a string enum, not a float The aggregate ceo entry in other_ratings is a 0–1 float. The ceo_rating field on each individual review object is a string ("APPROVE", "DISAPPROVE", or null). These fields measure the same concept but have completely different types — never compare or mix them: 
# Aggregate — float ratio
ceo_approval_pct = dim_lookup.get("ceo", 0) * 100         # e.g. 84.0%

# Per-review — string enum
ceo_sentiment = review.get("ceo_rating")                   # "APPROVE", "DISAPPROVE", or None
Guard all per-dimension review ratings against null All seven per-review dimension fields (career_opportunities_rating, ceo_rating, compensation_and_benefits_rating, etc.) can be null when the reviewer did not provide a score. Always check before accessing: 
# ❌ TypeError when rating is null
avg = sum(r["culture_and_values_rating"] for r in reviews) / len(reviews)

# ✅ Filter nulls first
rated = [r["culture_and_values_rating"] for r in reviews if r.get("culture_and_values_rating") is not None]
avg   = sum(rated) / len(rated) if rated else None
review_date and location can be empty strings — not null Unlike most APIs where absent values are null, these fields return an empty string "" when not provided. Use truthiness checks rather than is not None: 
# ❌ Won't catch empty strings
if review["review_date"] is not None:
    ...

# ✅ Correct — empty string is falsy
if review.get("review_date"):
    ...
Use limit=0 or limit=1 when you only need aggregate scores The overall_rating and other_ratings aggregate fields are always returned regardless of limit. When you only need the overview scores for a benchmark or monitoring check, set limit=1 to minimise response size and credit consumption: 
# Aggregate scores only — minimal reviews payload
r = requests.get(url, headers=HEADERS, params={"company": "canva.com", "limit": 1})
emp = r.json()["data"]["employee_reviews"]
# overall_rating and other_ratings are always present
Use is_current_employee and years_of_employment to segment reviews Separating current from former employees, or filtering by tenure, can surface significantly different sentiment patterns — particularly useful when assessing post-acquisition or post-layoff cultural impact: 
current_reviews = [r for r in reviews if r["is_current_employee"]]
veteran_reviews = [r for r in reviews if r["years_of_employment"] >= 3]
Use helpful_count to surface the most credible reviews Reviews with a higher helpful_count have been up-voted by the community and tend to be more detailed and representative. Sort by helpful_count descending for qualitative diligence or executive briefings: 
top_reviews = sorted(reviews, key=lambda r: r.get("helpful_count", 0), reverse=True)