How It Works

The three inputs

Every number on the site is derived from one of three sources, in roughly this order of trust:

1. Anonymous submissions from couples. The default and most-weighted source. Couples submit what they actually paid, with a date, guest count, market, and category-specific context.
2. Vendor-published pricing. When a vendor publishes a price list, a menu, or a starter package on their own site, we may use it as a secondary signal. Vendor-published prices are labeled distinctly from couple submissions.
3. Editorial roundups and directory listings. Used to discover vendors and bootstrap a new market. Never used to fabricate a price point.

What gets normalized

Raw submissions are messy. A photographer quoted for six hours of coverage is not the same data point as one quoted for ten. Before a submission contributes to a vendor’s published median, we normalize on a few axes so that comparisons mean something:

• Day of week. Saturday weddings price differently from Friday or Sunday. We normalize to Saturday-equivalent where the day affects price (mostly venues and catering).
• Guest count. Per-person categories (catering, bar, rentals) are normalized to a 100-guest equivalent for cross-market comparison. Per-event categories (photography, hair and makeup) are not.
• Seasonality. Peak, shoulder, and off-peak buckets are computed from market-specific demand curves, not from raw submission counts (couples tend to skew toward cheaper months, which would otherwise mask the real curve).
• Scope. Category-specific. Photography hours-of- coverage, venue rental versus package, catering style (plated, buffet, family-style, stations), beauty service (bride-only versus party).

Submissions that fall outside reasonable bounds (a $100 venue, a $500,000 cake) are flagged for human review before they enter the public dataset.

How medians are calculated

Where we have at least three normalized data points for a vendor, we publish a median. Below three points, we show the individual submissions but do not show a summary statistic. We use median rather than mean throughout because the distribution of wedding prices is skewed and a small number of outliers will pull the mean somewhere unhelpful.

For market-level statistics, we publish the median and the interquartile range (25th to 75th percentile). The IQR is more informative than the average alone because it tells you what “typical” actually means in a given market.

What we do not do

• Sponsored listings.No vendor pays to appear, rank higher, or suppress data. There is no “featured” tier.
• Lead-gen referrals. We do not sell your contact information to vendors. We are not in the business of routing couples to vendors who paid the most.
• Estimated prices presented as real ones.Every published price is traceable to either a couple’s submission or a vendor’s own public materials. We do not backfill with industry averages.

Anonymity and verification

Submissions do not require an account. We do not ask for your name, email, or any other identifying information. We hash the submitter’s IP for rate limiting and discard the raw value.

Because submissions are anonymous, we cannot perfectly verify any individual data point. We mitigate this in three ways: (1) category-specific sanity checks at submission time, (2) outlier detection against the existing distribution before publication, and (3) human review of flagged rows by a real person before they go public.

Vendors who believe a published submission is inaccurate can dispute it via the takedowns page. We respond promptly and, where the dispute is credible, we will remove the row and audit related submissions.

How the data grows

Coverage is uneven by design. We launch a market with a curated seed list of marquee vendors and grow outward as couples in that market submit. Until a market has enough submissions to be useful, we say so on the market page rather than smoothing thin data into a confident-looking average.

Open methodology

This page is the canonical version of the methodology. When we change how we normalize, weight, or summarize data, we update this page and note it in the changelog.

Have a question we did not answer here? Email us. Methodology questions from researchers and journalists are always welcome.