How we calculate these numbers
Every wedding cost page on the internet is generated from a model. Almost none of them publish it. Here is ours, including the parts we are least confident about.
Where the base numbers come from
The model is anchored on the average US wedding: $34,200 at 117 guests. We call that the reference wedding. Every category cost in the table below is what that category costs for the reference wedding in a market at the national average price level, and the column sums to exactly $34,200 — a unit test enforces it, so the pages cannot silently drift off the published average.
- The Knot 2026 Real Weddings StudyNational average spend and guest count. Survey of 10,474 US couples married in 2025.
- Zola — how much does a wedding costCross-check on category shares of total budget.
The two coefficients that do the work
A national average has to be adjusted for two things: how many people you are feeding, and where you are. The common shortcut is to apply one multiplier for each, uniformly, to every line item. That is wrong in two specific and predictable ways.
Headcount elasticity is how much a category tracks guest count, from 0 to 1. Catering is 0.95 — nearly every dollar is per-head. An officiant is 0, because the fee is identical whether fifty people watch or two hundred and fifty. A photographer is 0.05: a day rate, barely moved by the guest list. Models that scale everything by one factor typically hand a 150-guest wedding a 40% per-plate discount, which no caterer offers.
Local intensity is how much a category tracks the metro price level, also 0 to 1. A venue is 0.95 — local real estate, fully exposed to the market. Attire is 0.15, because a dress is a national retail good that costs about the same in Houston as in Boston. Apply one city multiplier to everything and you end up claiming Manhattan officiants cost 60% more than Orlando ones, in lockstep with ballroom rental.
The exact formulas, with no rounding hidden in them:
headcount factor = (1 − elasticity) + elasticity × (guests ÷ 117) metro factor = 1 + (city index − 1) × local intensity category midpoint = base × headcount factor × metro factor
At 117 guests in a 1.00-index market both factors are exactly 1 for every category, which is what keeps the whole model tied back to the published national average.
Every coefficient, published
| Category | Base | Headcount | Local |
|---|---|---|---|
| Venue & site rental | $6,500 | 0.30 | 0.95 |
| Catering (food & bar) | $8,700 | 0.95 | 0.80 |
| Photography | $3,200 | 0.05 | 0.65 |
| Videography | $2,000 | 0.05 | 0.65 |
| Flowers & decor | $2,700 | 0.55 | 0.70 |
| Music (DJ or band) | $2,300 | 0.05 | 0.60 |
| Attire & rings | $2,100 | 0.00 | 0.15 |
| Rentals & production | $1,800 | 0.75 | 0.75 |
| Planner or coordinator | $1,500 | 0.15 | 0.80 |
| Transportation | $900 | 0.60 | 0.70 |
| Hair & makeup | $600 | 0.35 | 0.70 |
| Stationery & signage | $600 | 0.70 | 0.25 |
| Cake & desserts | $450 | 0.85 | 0.55 |
| Favors & gifts | $350 | 0.80 | 0.20 |
| Officiant & licence | $350 | 0.00 | 0.45 |
| Insurance & permits | $150 | 0.10 | 0.50 |
| Total | $34,200 |
Why the ranges are lopsided
Each category has a spread, and the low and high are not symmetric around the midpoint. Wedding pricing is right-skewed: the floor of any category is set by what it physically costs to deliver, while the ceiling is set by taste and has no real limit. A symmetric band would understate how far the top end runs. We publish the downside at 62% of the upside for that reason.
Where the model is weakest
We would rather say this ourselves than have you find it:
- Metro indexes are single numbers. Each of our 24 markets carries one price index, softened per category by local intensity. That is a real improvement on a flat multiplier, but it is still not per-category market data. A city where venues are unusually cheap and florists unusually dear will not show that.
- The total assumes you buy everything. It includes videography and a coordinator, which many couples skip entirely. Read the category rows, not the total, if you are building a real budget.
- Season and day of week are not modelled. A Friday in February and a Saturday in September are different prices in the same room — often by 30% or more. That swamps the difference between several of our cities.
- The elasticities are estimates. They are informed by how vendors actually price, not derived from a regression on transaction data we do not have. We publish them so you can disagree with a specific number rather than the whole page.
Where we do have real data, we say so
On some city pages you will find a block titled “what vendors actually list”. That is not modelled — it is published price ranges from vendors in our own directory, shown as a median with the sample size next to it. We only show a category when we hold at least three priced vendors for it, and we omit hand-entered illustrative rows entirely. Most markets do not have that block yet. When they do not, we show nothing rather than presenting a national estimate as local evidence.
Corrections
If you have quotes that contradict these numbers, we want them — that is the fastest way this model gets better. Write to support@ammore.ai.