> ## Documentation Index
> Fetch the complete documentation index at: https://docs.propilotapp.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Selection & Quality

> How comps qualify through the tier ladder, how outliers and off-price sales are removed, and what the quality grade and discount badge mean.

ProPilot qualifies comparable sales through a **tier ladder**: a cascade of matching rules that
starts strict and loosens step by step until enough comps are found. The result is a comps
value plus a quality grade that tells you how strict the underlying matches really are.

<Info>
  **Key things to know**

  * Tier 1 is the strictest match; the search only falls to looser tiers when it cannot find enough comps.
  * Two cleanup rules run on top of the ladder: price-per-sqft outlier removal and a list-price band that drops sales priced more than 50% away from the subject's list price.
  * The grade (Excellent to Poor) comes from the average tier of the selected comps, not from the comp count.
  * If too few comps qualify, you get a clear message: "Insufficient comps: found X, need at least Y."
</Info>

## How comps qualify

Each candidate sale is tested against the tiers in order. A tier defines:

| Criterion     | What it checks                                                                                                                           |
| ------------- | ---------------------------------------------------------------------------------------------------------------------------------------- |
| Property type | The comp must be the same property type as the subject.                                                                                  |
| Radius        | Maximum distance from the subject, in miles. Tighter in early tiers.                                                                     |
| Size window   | How far the comp's square footage may differ from the subject, as a percentage.                                                          |
| Beds / baths  | The allowed difference in bedroom and bathroom counts.                                                                                   |
| Similarity    | A minimum neighborhood similarity score, built from census tract data, so comps come from comparable areas rather than just nearby ones. |
| Sale age      | The sale must be within your org's **Max Days Old** setting.                                                                             |

The search collects qualifying sales tier by tier and stops as soon as it has enough comps to
meet your org's minimum. It does not keep descending into weaker tiers once the minimum is met,
which keeps the result anchored to the best available matches.

Admins can edit every tier's radius, size window, bed/bath allowance, similarity floor, and
grade in [Advanced Settings > Comps](/settings/advanced-settings/comps).

## Cleanup after qualification

* **Outlier removal:** sales whose price per square foot sits far from the local median are
  dropped so a single unusual sale cannot skew the value.
* **List-price band:** comps with sale prices more than 50% above or below the subject's list
  price are excluded. This filters out data errors and sales that are clearly a different
  class of property.

Both rules can occasionally remove a sale you consider valid. You can always re-include it
manually with **Use comp** in the [comps selector](/features/comps/view-adjust).

## Quality grade

The grade pill summarizes match quality:

| Grade                    | Meaning                                                                          |
| ------------------------ | -------------------------------------------------------------------------------- |
| **Excellent**            | Comps came from the strictest tier: very close, very similar sales.              |
| **Very Good** / **Good** | Solid matches with minor differences in distance, size, or beds/baths.           |
| **Fair**                 | Looser matches; the search had to descend the ladder. Treat the value with care. |
| **Poor**                 | Weak matches. Consider adding manual comps or reviewing the selection.           |

The grade is computed from the **average tier** of the comps actually used. If you hand-pick a
selection, the grade updates to reflect the tiers of the comps you kept. Each comp also shows
its own tier grade in the table and map popup so you can see which sales are dragging the
overall grade down.

## Insufficient comps

If fewer comps qualify than your org's minimum, no value is produced and the property shows
"Insufficient comps: found X, need at least Y." Your options, roughly in order:

1. **Recalculate** to pull the latest sales data.
2. **Add Comp** manually for sales you know about.
3. Ask an admin to widen the tier ladder or raise **Max Days Old** in
   [Advanced Settings > Comps](/settings/advanced-settings/comps).
4. Lower **Min Comps** (last resort - fewer comps means a less reliable value).

## Discount badge

For **Active** listings, a badge compares the comps value to the list price:

* **12% under** - the property is listed about 12% below its comps value (potential discount).
* **5% over** - the property is listed above its comps value.

The same percentage appears as **Discount** in the comps selector header and updates live as
you change the selection.

## Common questions

<AccordionGroup>
  <Accordion title="Why did a comp disappear?">
    Four possibilities: it was removed as a price-per-sqft outlier, it fell outside the
    list-price band, it stopped passing tier rules (for example a settings change or the sale
    aged past Max Days Old), or **Use comp** was toggled off and saved.
  </Accordion>

  <Accordion title="Why is the grade low even though there are plenty of comps?">
    The grade measures match strictness, not quantity. Ten comps from a loose tier still grade
    lower than five comps from the strictest tier. Check each comp's tier in the table to see
    where the matches came from.
  </Accordion>

  <Accordion title="Can two similar properties get different grades?">
    Yes. The grade depends on what sold nearby recently. A property in an active market can
    grade Excellent while an identical one in a thin market grades Fair because its comps came
    from looser tiers.
  </Accordion>

  <Accordion title="Why was a higher-priced valid comp excluded?">
    Most likely the list-price band: if the subject is listed well below market, legitimately
    higher-priced comps can land outside the plus or minus 50% band. Re-include them manually
    with **Use comp** if you trust them.
  </Accordion>

  <Accordion title="Does excluding a comp change the grade?">
    It can. The grade is recomputed from the average tier of the comps that remain selected, so
    removing a strict-tier comp and keeping looser ones lowers the grade, and vice versa.
  </Accordion>

  <Accordion title="How is this different from Neighborhood Benchmarks?">
    Comps are deal-specific for one address, built from individual sales you can inspect.
    [Neighborhood Benchmarks](/features/neighborhood-benchmarks) are area-level statistics for
    a census tract.
  </Accordion>
</AccordionGroup>

## Limitations

* Neighborhood similarity depends on census tract data, which is missing in some areas; where
  unavailable, matching relies on distance and property filters alone.
* MLS coverage varies by market. Off-market sales are not always captured, and distressed
  sales can skew values in heavily affected areas.
* A grade describes match quality, not market certainty. Even Excellent-grade values deserve
  scrutiny in fast-moving or volatile markets.