Why our donation values may look lower than you remember

If you tracked donations for a decade or two with ItsDeductible before it shut down in October 2025, there is a good chance our numbers will read low to you. A shirt you are used to seeing valued at eight dollars may come out at five. That gap is real, it is deliberate, and you deserve a straight explanation of it rather than a marketing sentence about accuracy.

The short version: the old product priced against a data source we cannot license and would not copy, and we price against published charity resale ranges that we can show you. Both are defensible readings of the same IRS guidance. Ours is the more conservative one, and we think that is the right trade for a product whose entire promise is that the number survives a question.

Where the old numbers came from

ItsDeductible described its methodology publicly for most of its life. It combined comparable-sales data from completed online auction listings with thrift-store sales data, refreshed annually and reviewed before each tax year. Whatever else was true of that product, the auction feed was a genuine licensed data asset and a large part of why the values felt generous.

We cannot replicate that. Not because we did not think of it, but because a closed-auction data license of that kind is not available to an independent app at any price we could justify, and because building on it would mean building on a dependency that could be withdrawn. The product you are reading about exists precisely because a valuation service that depends on one company's continued interest can disappear on a Tuesday.

Two markets, two prices, and the question of which one you donated into

Publication 561 tells you to look at sales of comparable properties. The interesting question is comparable to what.

A completed online auction is a national market. The item was photographed, described, searched for by someone who specifically wanted it, and shipped. A thrift shop shelf is a local market with no search, no description, no selection, and a buyer who happened to walk past. Those two markets clear at genuinely different prices for the same jacket, and neither price is fictional.

But Publication 561 does not leave the choice entirely open. In its discussion of used clothing it says that "the price that buyers of used items actually pay in used clothing stores, such as consignment or thrift shops, is an indication of the value," and its worked example is a garment given to a thrift operator, valued at "the price that typical buyers actually pay for clothing of this age, condition, style, and use."

If you dropped four bags at a thrift store, the comparable sale is the thrift shelf. That is the market your property actually entered. Valuing it at auction prices is valuing a transaction that did not happen, and requires you to assume the charity will list, photograph, and ship every garment individually, which no thrift operation does.

How a DeductiBee value is actually built

There is no black box here, so here is the whole method on one item. Three charities publish their own donor valuation guides covering sofas, at these ranges:

We take the lower median of the lows and the lower median of the highs, not the mean. Lows are 30, 36, 40, so the consolidated low is $36. Highs are 150, 207, 395, so the consolidated high is $207. The median discards the outlier at both ends: one region's unusually high ceiling cannot drag a national figure upward, and one region's unusually low floor cannot drag it down.

That published span is then split into the three condition tiers at fixed fractions, 35% and 70% of the span, and rounded once by a published rounding rule. The sofa comes out at $36 to $95 Fair, $95 to $155 Good, and $155 to $205 Like new.

Two rules make that reproducible rather than rhetorical. We never extrapolate outside the published envelope, in either direction, so no tier can exceed what a publisher actually printed. And every leaf in the dataset carries the source ids behind it: the generator refuses to emit a value with no citation, so an uncited number cannot reach the app by accident.

Where we are actually lower, and where we are not

"Lower across the board" is not accurate, and it is worth being precise about the shape of the difference.

• Mid-condition everyday items read lower. This is the bulk of a normal donation and the source of most of the felt gap. Our Good band sits astride the middle of what charities themselves publish.
• Genuinely unworn items may read higher. After 2006 the old product had only two enterable tiers, so a coat with the tags still on had nowhere to go above the top tier. Our Like new band occupies the top 30% of the published range, which is often the more generous answer for items that deserve it.
• Worn-but-usable items read lower, because our Fair band reaches down to the published floor rather than stopping at a mid tier.
• Items nobody publishes a range for get no value at all. Roughly a third of the leaves in our catalog are custom-entry: we tell you what evidence the IRS expects and let you supply the figure, rather than manufacturing one.

That last point is the one most likely to annoy you, and it is the one we are least willing to change. A catalog leaf is backed by a published range for that item or it does not exist as a value. Some of the published guides offer a fallback heuristic for unlisted items (30% of the original price), and we surface that as attributed guidance on the custom-entry screen. We do not use it to generate catalog values, because a heuristic applied at scale produces a table of numbers nobody ever published.

Why the conservative number is the one that survives

The uncomfortable case for conservatism is in Publication 561's own penalty section. It describes a 20% penalty on the underpayment of tax related to a valuation misstatement where the claimed value is 150% or more of the correct amount and the resulting underpayment exceeds $5,000, and a 40% penalty where the claimed value reaches 200% or more of the correct amount with the same underpayment threshold.

Most household donors will never approach those numbers. The point is not fear, it is direction: on a spectrum from "defensible" to "generous," the errors that cost something are all on one side. And a claim backed by three named publishers' printed ranges is a fundamentally different conversation from a claim backed by a number that appeared in an app.

Publication 561 also cautions, about published price guides generally, that "these sources are not always reliable indicators of FMV and should be supported by other evidence." That is an argument for consolidating across several independent publishers rather than trusting one, which is exactly what the median step does, and an argument for keeping a photo alongside the number, which is what the app is for.

What to do when you think a number is wrong

Sometimes it will be. A published range describes the typical item, and your item may not be typical: a premium brand, an unusually well-kept piece, a regional market that prices differently.

1. Override it. DeductiBee lets you set your own value on any item, which the old product notably did not allow on catalog items.
2. Expect the override to be flagged. Manual values are marked as such in exports, so you and your preparer can see instantly which figures are sourced and which are yours. That is a feature, not a scold: it is the difference between a record and a guess.
3. Write down why. One line about the comparable sale, brand, or condition that justifies the figure, saved with the item, converts an assertion into evidence.
4. Tell us. The dataset is refreshed as a versioned annual edition, and a category with a bad range is a bug we want reported.

What you get in exchange

Visible sourcing on every range, an override that is honest about being an override, a dataset that is refreshed annually as a dated edition rather than silently mutated, and a record that lives on your own device instead of on a server that can be switched off. If you want to inspect the numbers before you trust them, start with the value guide, which shows the sources per row, or run a bag through the donation value calculator. If you are comparing directly against what you used before, the side-by-side comparison lays out the differences without hedging.

We would rather tell you a number is $5 and be able to show you three charities who printed it than tell you it is $8 and be unable to say where that came from. If that costs us on the first-impression comparison, it is a cost we are choosing on purpose.