What's new.

The on-screen report never showed the subject home's photo, only the comparables' photos. The report now opens with it.

When a home had no listing photo, the PDF left out its street view photo, on the cover for the subject and on the cards for comparables. They are included now.

Each comp card on Results now shows the year the home was built, when the record has it, next to its sale date or days on market.

A follow-up question in the support chat often failed with “The assistant couldn't finish that reply”, and trying again failed the same way. Follow-ups now get an answer.

When you searched a manufactured or mobile home, its comparables were manufactured homes but were labelled “Single Family” on Results and in the report, and each lost points on its property-type match. They now show as Manufactured and score as a match, and the occasional site-built house or lot that slipped into the set is left out.

In some areas manufactured homes are listed as houses, so a manufactured home could come back with “No Properties Found” while manufactured homes had sold nearby. The search can now find the ones listed as houses and use them as comparables, without adding site-built houses.

On every paid plan, set your own list price on any seller report with “Set your own price” on the Suggested List Price tile. The time to sell and a short note on where your price sits are rewritten for it, net proceeds add a “Your list price” column, and the PDF uses your price. The comps, the estimated value range and the three scenarios stay as the evidence, and “Reset to suggested” goes back.

Closing costs in Estimated Seller Net Proceeds can now be edited, like commission. They were fixed at 1.5%.

Add the seller's mortgage payoff to the net proceeds table. It is subtracted from every column, so the estimated net is what the seller walks away with.

Leaving the report page while an analysis was generating threw the finished report away, because it was only saved once it reached your screen. It now finishes and saves to Your reports on the search page, so you can open it later.

The Support page showed a “Manage or cancel” button to accounts with no subscription, and it could only fail. It now shows only when there is a subscription to manage.

A Rent Analysis on a multi-family building now prices it unit by unit. The public record describes the whole building, so the report starts from its totals split evenly over its units, clearly marked, and prices each unit from apartment, condo and small multi-unit rentals nearby. The building's monthly rent is the sum of its units, and the yield and cash flow use it.

Under Edit property & units, describe the building's real units: bedrooms, bathrooms, size and how many of each (a four-plex of identical 2-beds is one row × 4). Updates are free.

Before this, multi-family buildings never got a rent estimate: the building's total bedrooms and size were matched against single rentals.

Rent estimates with an even number of rentals could set aside a close, well-matched rental as out of line with the rest, because it was compared with the lower of the two middle rents instead of their average. It's now compared with the average, which keeps it.

A house's rent estimate could use half of a duplex or a single-floor flat (addresses like “Unit 1/2” or “Floor 2”) as a comparable rental. These are now recognised as units and left out for houses.

On a Rent Analysis map, rentals close together could hide one another, so some had no pin. They are all drawn now (units in the same building share one spot, so zoom in to tell them apart).

For condos with few rentals nearby, units looked up inside nearby buildings could include units for sale with their sale price. Only units listed for rent are used now.

Some condo units are filed on the public record as multi-family buildings. They were compared with multi-family properties, so a downtown condo's Rent Analysis could find no similar rentals at all. A unit with a condo association is now treated as a condo, for comps and rentals alike.

If you type an address you searched in the last 24 hours and click Search, we now ask first: open the results you already have, free, with any changes you made, or search again for a credit. Before, typing it again always ran a new search.

Clicking an address under Recent on the search page always ran the search again and used a credit, even when you only wanted to get back to Results you had a few minutes earlier. When those Results are still open (they last 24 hours), the address now says “Open results” and takes you straight back to them, with any Edit details changes, at no credit.

Click Rent analysis on the Results page for a monthly rent estimate and range built from similar homes listed for rent nearby, each adjusted for size. Each rental shows its photo and asking rent and, where its listing or record has them, its days listed, any rent cut, and its earlier rental listings: the asking rent when it came off the market and how long that took.

The report also covers the local rental market (how many homes of its type are for rent nearby, median rent, median rent per sq ft, how long current listings have been up, how many have cut their rent, and rent by bedroom count) and a gross yield on the asking price or the automated value. It has its own branded PDF, uses no credit and doesn’t need any comps selected. Every figure is calculated, not written by AI.

Rent or sell, for owners: switch a Rent Analysis to “The current owner” under Property & units and it compares what selling at today’s value would put in hand (after commission, closing costs and any mortgage payoff) with what renting it out would bring in a year, using the owner’s own tax bill (when the record has one) and mortgage.

Cash flow on every Rent Analysis with a price: rent less vacancy, taxes, insurance, HOA, management and maintenance, then the mortgage, with cap rate and cash-on-cash return. Taxes are estimated on the purchase price, since they’re usually reassessed after a sale, and every assumption is shown and editable under Property & units.

The Rental Market section now shows HUD’s Small Area Fair Market Rent for the ZIP by bedroom count — a benchmark many housing agencies use for Housing Choice Voucher rents — and the report states plainly that its rents are long-term (monthly lease) rents, not short-term or vacation rentals.

Rent estimates now cap each rental’s size adjustment at 10% of its rent. Where the nearby rentals were much smaller than the home, large adjustments could lift the estimate above most of their asking rents.

When the rentals nearby support an estimate but the public record’s automated rent figure is far from it, the Rent Analysis now gives the estimate with a clear note on how far apart they are, instead of withholding it. Both have been wrong before, and the report shows every rental behind the estimate so you can judge.

Where too few similar homes are for rent nearby to support an estimate (common for large homes and rural areas), the report says so plainly, lists the nearest similar rentals as they are, and shows the public record’s automated rent figure for reference, instead of giving no report at all.

Edit property & units on any Rent Analysis: correct the home’s bedrooms, bathrooms and size, add an efficiency, ADU or garage apartment that rents separately (each priced from small rentals nearby, with a total monthly rent for the whole property), and enter the purchase price your client is considering for the yield. Updates are free.

Only need the rent? The Rent analysis tab on the search page builds the Rent Analysis straight from an address, without pulling sale comps. Like a search, it uses one credit, and nothing when the home can’t be found.

For condos and apartments with few homes for rent nearby, the rent estimate (in the Rent Analysis and in Pro seller reports) now also looks inside the nearest apartment buildings and compares individual units: rent, bedrooms, bathrooms and size.

The Rental Estimate now looks at a wider set of homes listed for rent around the property, and when few similar homes are for rent nearby (often the case for larger homes) it also searches the whole city for rentals of a similar size. In a check on 40 recent searches it appeared in 27 reports, up from 12. The rules for what counts as comparable are unchanged.

Townhouses never got a Rental Estimate: the search for townhouses listed for rent failed every time, and the section was left out. They now get one like any other home.

The support assistant only looked at your five most recent searches, so after a busy week it could say it couldn’t see a home you had searched a few days earlier. When you name the home, it now finds that search anywhere in the last 14 days.

Click Edit next to the commission line in Estimated Seller Net Proceeds to set the total commission for this report (0–10%). Every scenario’s net updates, and the PDF uses the same rate. Reports without a rate you set still use 5.5%.

In Edit details, lot size can be entered in acres or square feet. A lot of an acre or more opens in acres.

A comp whose listing description says it was a short sale, foreclosure, bank-owned or auction sale is tagged “Distressed sale” on its card. Estate and trust sales, homes sold as-is or as a fixer, and sales between related parties get their own tags. Hover a tag to see why it is there.

Tagged comps are never selected automatically, because these sales often closed below what the home would fetch on the open market. They stay in the list, and you can still select one if it is the right comparison.

More comps now show a story count: the listing description is read for the home’s own count, ignoring rooms like a “two-story foyer”. A comp that differs from your subject by a full story is ranked accordingly, as before.

On large searches the page numbers under the comp cards ran off the screen (a search with 230 comps showed all 26). They now show the first and last page, the pages around the one you’re on, and “…” for the rest.

Some comps showed a story count that described a room or a building amenity rather than the home — “two-story keeping room”, “two story light-filled foyer” or “two-story fitness center” read as a 2-story home, and “one-level living” read as 1-story in homes that also have an upstairs. Those comps could be marked as a story mismatch with your subject and lose points they shouldn’t have. They now show the story count as unknown, which is scored neutral. A count the listing states for the home itself still reads as before.

Seller reports on Pro and Premium include a Rental Estimate: a monthly rent and range from up to five homes of similar size and layout listed for rent nearby, each shown with its asking rent and a rent adjusted for size. It helps with the “sell or rent it out?” conversation. Rents are asking rents on current listings, and every figure is calculated, not written by AI.

When an area has too few comparable rentals (often the case for large homes), or their rents are too far apart to support a figure (common where vacation rentals mix with long-term ones), the section is left out rather than showing a number you couldn’t defend.

Reports showed “Garage: No” for many homes that have a garage, and the written analysis could repeat it — one agent’s report said a home with a two-car garage had none. Public records often leave the garage out, and that gap was being read as “no garage”. A garage now appears only when the record shows one; otherwise the report says nothing about it. To describe a garage in the analysis, add it under Fine-tune AI before generating.

When the home you search is listed for sale, Results shows a “Buyer offer” button next to Generate Analysis. It builds a report for your buyer with three offer prices — Strong, Competitive and Value — each shown against the asking price and backed by your selected comps.

The report shows how the selected sales closed against their last asking price, the listing’s days on market and price reductions, and other active listings a buyer could consider. Offer prices are calculated from the comps, never written by AI, and the report describes the listing’s record — not the seller.

Comp adjustments sometimes showed cents (“$-25,264.5”, “+$5,687.50”). They are now always whole dollars, and report text writes dates as “Sep 29, 2026” instead of “2026-09-29”.

In a hot market (under 3 months of inventory), a buyer offer report’s Strong offer is never below the asking price. A couple of soft sales could otherwise put every offer under asking in a seller’s market.

The Competitive Position paragraph could explain a comparable's adjustment with different figures than its comp card — one report gave a comp a −$37,759 adjustment in the text and −$27,987 on the card. The paragraph now compares the comps by their size, layout, location and sale prices and leaves the adjustment figures to the comp cards.

Buyer offer reports leave out the public automated estimate, which sat beside the asking price as a third, unexplained value. Seller CMAs still show it.

Like every report, a buyer offer report uses no credit — credits are only spent on searches and imports. It saves to Your reports labeled “Buyer offer” and exports as a PDF titled for your buyer.

Some detached houses on their own lot are legally set up as condominiums (“site condos”, common in Michigan subdivisions), so the public record lists them as condos. The search then looked only for condo sales, and in an area with none it showed “No Properties Found” even though plenty of house sales were nearby. A condo record with no unit number on at least a quarter acre is now treated as a house, and comps come from house sales. Condos with a unit number are unchanged.

In “Add a comp by address”, the map pin icon sat on top of the address you typed and the placeholder text. The text now starts after the icon.

The support assistant can now see searches that came back with no results. Before, a search that showed “No Properties Found” left no trace, so the assistant answered “I don’t see that search” while you were looking at it. It now sees the address, why nothing came back, and which property type was searched.

Starting a search made the background illustration on the search page jump, because your recent searches disappeared while the search ran. They now stay in place, dimmed, until results open.

Some saved reports couldn’t be reopened from “Your reports”. Clicking one showed “Invalid Request — Please select 3-6 comparables” and sent you back to Search, even though the report was saved — about one in six recent reports, mostly ones that included a comp added by address or comps from certain listing sources. Every saved report now reopens; nothing needs to be generated again.

Picking a city, ZIP code or area from the address suggestions ran a search anyway, pulling comps around the middle of town for no particular property. The address box now asks for a street address instead. A street name with no house number still works, for vacant lots and new builds.

Row houses recorded as single-family homes were only compared against detached houses. Most sales in row-house neighborhoods are listed as townhouses, so a search in a city ZIP code could return one recent sale among a page of active listings. A row house is now treated as a townhouse for comps.

Comps are now compared on story count. A one-story home was being matched against two-story homes with nothing on the card to say so. When both counts are known and differ by a full story, the comp scores lower, its card leads with “2-story vs subject’s 1-story”, and it isn’t preselected if other comps can fill the top three. Many listings don’t state a story count; those are treated as unknown, not as a mismatch.

Story count shows on comp cards, the subject card, the report and the PDF, and Edit details now has a Stories field for when the record is missing it or has it wrong. Condos and apartments are not compared on stories, since the record usually counts the building’s floors.

When you search a street address without a unit number and the public record comes back for one specific unit, Results now says so: “The public record for this address is unit #324.” The facts, property type and comps all describe that unit, and nothing on the page used to point it out — so a search could quietly price the wrong home. The notice links to a new search (add the unit with “+ Add unit / apt #”) or to Edit details.

The support assistant could occasionally finish without replying, and the chat showed nothing — no answer and no error — so it looked like the message had been ignored. It now always finishes with an answer, and if a reply does fail the chat says so with a Try again button.

The support assistant can now look at your own recent searches, so when you report a problem with an address, the property facts or the comps, it answers about that search instead of giving general steps. It asks which address you mean when it isn't clear.

When fewer than three recent sales of the subject's own type are on record, the search also pulls sales of the closest type — townhouses for a house, houses or condos for a townhouse, townhouses for a condo. Those comps score lower on property type than same-type comps, so the closest matches still rank first, and the Results header says which types were added.

The comp radius widens past 5 miles when fewer than three recent SALES sit inside it, even when there are plenty of active listings. Every adjusted value in the report is built from closed sales, so a busy 5-mile radius with one sale in it produced an estimate resting on a single property — most often in dense city ZIP codes.

When the public record has no bedroom or bathroom count, it is no longer treated as a home with zero of them. Reports were opening with wording like “a 2,520 sqft single-family home with 0 bedrooms”, and every comp lost the full bedroom match score against a count nobody had.

A missing count now shows as a dash on screen and in the PDF, comps are not adjusted for it, and comp match scores treat it as unknown rather than as a mismatch — so strong comps are no longer pushed down the ranking.

Results shows an amber notice when the record has no bed or bath count, with an Edit details button. Generate Analysis stays available — it is a prompt, not a block — and entering the count re-scores the comps at no extra cost.

Results now says so when fewer than three recent sales are on record near the address: the estimate then leans on asking prices rather than closed sales, and the notice says to treat the value band as indicative. Pro and Premium can add a sale they know of with “+ Add a comp”.

The comp header no longer says “within 15 miles (widened)” when every comp shown is a couple of miles away. It reports the tightest distance that still covers all the comps on the page.

Vacant land is now scored as land. A lot has no living area, bedrooms, bathrooms or year built, and matching a comp’s zeros on those was paying out up to a third of the match score — so a four-acre parcel could outrank a similar-sized lot next door. Lots are now scored on lot size, distance, price, sale recency and property type, with lot size carrying the most weight.

Land adjustments are calculated from local land prices per acre instead of a price per square foot of living area the lot doesn’t have. The report’s market tile shows average price per acre for land, and no longer prints “$0 per sqft”.

A comp lot whose size isn’t listed is treated as unknown rather than as a poor match, and says “Lot size not listed — verify acreage” on the card.

The estimated-value summary can no longer explain a comp’s adjustment with different figures than the comp’s own breakdown shows. Both came from the same total but were split differently; the summary now repeats the published breakdown, and any figure that doesn’t match it is replaced with a plain computed summary.

The report shows the subject's federal flood zone, whether it sits in a Special Flood Hazard Area, and the base flood elevation where one is published. Properties in a high-risk zone are highlighted.

The flood wording reports what the official map shows and asks you to verify it against the current map; it is never a flood determination or insurance advice. Where no map covers the property, the report says so rather than implying there is no risk, and imported comp sets have no verified location so no zone is shown.

Market Analysis now includes the published price index for the subject's metro area, county or state, with its year-over-year and quarter-over-quarter change and the period it covers.

That price measure sits beside the existing trend label, which reflects inventory and days on market. They can disagree — prices can flatten while supply still looks tight — and the report shows both instead of merging them into one verdict.

A comp that closed before the latest published quarter is restated to today's market using the same price index, and the adjustment appears as its own line in that comp's breakdown. In a market that has fallen, the sale is adjusted down.

That matters most for imported comps, which often include sales from the past year; comps from a normal address search are recent enough that no adjustment is applied. Reports that use the index say so, including what it does and does not cover.

Market Analysis also includes the week's average 30-year fixed mortgage rate and its change from a year ago, credited to its published source. It is national context for buyer demand and says nothing about the property's value.

If the rate cannot be retrieved or has not been updated recently, the report leaves it out rather than showing a figure that may no longer be current.

When you generate a report, the comps you selected are checked against the property record for their exact sale date, plus any missing lot size, year built or living area. Most sold listings arrive without a sale date, so recency was previously estimated from the sold window.

Comp cards and the report show the real month and year of sale where the record has it, and comp scoring uses the exact recency. Uses no extra searches, and facts you entered yourself are never overwritten.

Lot sizes are now read from more listing formats, including abbreviated units, lot-size labels that come before the number, and acreage written with a tolerance. Previously these properties were treated as having no lot size, which kept them out of acreage matching.

A lot size written with decimals is no longer read as zero.

The days-on-market metric in Market Analysis is labelled "Avg DOM · Comps" because it covers the comps you selected, while the written analysis may cite a days-on-market figure for the wider set of properties pulled. Both are accurate, and an unlabelled figure could read as a contradiction.

Excluded-row messages now use the row number from your own file. Exports with title rows above the header, or blank rows between listings, previously pointed a few rows off.

Reports built from an import no longer show an "Unknown Market" label in slow-market orange next to the estimated value. The trend now reads "Unavailable" on screen and in the PDF, matching the days-on-market figure beside it.

The Distance sort is hidden for imported comps, which have no distances to sort by, and adding a comp by address now explains that it is unavailable for an imported set instead of reporting an impossible distance.

Homes on at least two acres now get a deeper comp pull on every plan. Match order prioritizes comparable acreage, and small-lot or unknown-lot properties are no longer automatically selected to fill an acreage report. Thin coverage is explained on Results.

Material lot-size corrections now remove the public estimate based on the old parcel facts. Reverting restores the original estimate.

Completed analysis appears while the report saves separately. Loading has time limits and a Stop waiting action; a failed save offers Retry save without hiding the report.

Support now shows a direct Manage or cancel button and live subscription status, including scheduled cancellation and access end dates when available.

Export PDF is now at the top of the report. Prepare a PDF and share it through supported devices, or download it to attach to an email or text.

The pricing page now explains MLS imports, supported file types and limits as a Pro and Premium benefit.

Import Excel workbooks as well as CSV, TSV and TXT files. Choose a worksheet, header row and text encoding when needed.

Combine split addresses and bathroom counts, convert square meters and acres, and choose date and number formats. Map unfamiliar listing statuses and property types explicitly, then review the normalized facts before importing.

Save a named mapping on this browser for your account. Reapply it by column name when your next export changes column order; missing columns are flagged for review.

Choose Import MLS comps on the Search page, enter the subject details, match your export's columns and review the properties before importing. Select 3–6 comps in Results and generate your report as usual. Available on Pro and Premium; a completed import uses one search credit.

MLS imports accept exports with optional columns omitted, recognize quoted delimiters, and prefer specific property subtypes. Subject-address checks recognize common abbreviations. Missing market data stays unavailable, and imported land details can be edited without entering living area.

The preview explains excluded rows, including missing closing prices, missing facts, duplicates and unsupported statuses. Imported comps retain their actual closing dates. Reports disclose the uploaded source; unavailable distances are not treated as nearby, and selected exports do not produce a market-wide inventory estimate.

A home whose public estimated value sits far above its neighbourhood could come back with a single comp. The price-outlier filter anchored on that estimate and discarded every nearby sale below about a third of it. When that anchor would leave fewer than 12 comps, the filter now anchors on the comp set's own median instead, so the neighbourhood's listings and sales stay in the set and match scoring ranks the closest-priced ones first.

When the public record was found but listed no living area, Generate Analysis was disabled with only a hover tooltip saying why, and re-running the search returned the same record for another credit. Results now shows an amber banner explaining that the square footage is missing, with an Enter property details button that enables the analysis once it's saved.

Report prose could describe a comp as "smaller" while its adjustment line said "larger" with the same dollar figure. The generation instructions now fix the sign convention (an inferior comp adjusts up, a superior comp adjusts down) and require prose to repeat each adjustment's direction word exactly.

The home page opens on a new illustration and headline, and every section below it is rebuilt in the same style: what the product does, how it works, what the report contains, how it compares to legacy CMA tools, what agents say, and where it works. The sign-up and sign-in pages share the look.

The product screenshots are gone. The home page now shows a drawn rendition of the search results and the report's actual pricing strategy section, fed with sample data, so what you see there is exactly what the app produces.

Pricing, FAQ, the blog, this changelog, support, the seller net sheet calculator, the guides, and the terms and privacy pages share the same look, so the site reads as one piece wherever you land.

Inside the app, the search page, results, and the report use the same palette and type. Light and dark modes both still work from the toggle in the header; the colours are warmer in both.

A "See a sample report" button on the home page opens a finished report, so you can see what you get before creating an account.

A subject with no living area (a lot, or a record we couldn't find) could still generate a report against house comps. With no square footage to adjust, the comps got no deduction for the house the lot doesn't have. Generate Analysis now stays disabled until the facts are entered, and the report explains why if you get there another way.

When the public record identifies the address as a lot, the comps pulled are land listings and sales, and the report prices the lot against them. A lot the record doesn't recognise still can't be priced — the banner now says so instead of asking for beds and baths.

Some subjects showed a raw code ("HOME_TYPE_UNKNOWN") as their property type. The type is now read from the record's facts when the label is missing.

The support assistant now checks your plan before answering questions about Pro and Premium features, so it won't send a Starter user looking for a button they don't have.

Same-street listings were being dropped. When the public record placed a subject miles from where it actually is, the 5-mile comp filter threw out everything nearby — in one case all 40 recent sales and a listing three doors down. The verified map location now anchors comp distances.

The comp radius widens automatically in thin markets: 5 miles, then 10, then 15, until at least a dozen comps are in view. The Results header says when it widened, and listings on the subject's own street rank first with a "Same street" tag.

Some searches returned a subject with 0 beds, 0 baths and 0 sqft plus an estimated value that was really the area's typical price. Those searches are now retried, and if the record still can't be found you get blank facts, no estimate, a clear banner into Edit details — and the search is not charged.

Sold comps with no recorded closing price (common in non-disclosure states and some townhouse feeds) are no longer preselected or fed into any price figure. Results shows a notice when most recent sales are unpriced, and the report discloses the count.

Pro and Premium: "+ Add a comp" on the Results page adds a sale or listing by address — the one you know about that the pull missed. It's scored like any other comp, tagged "Added by you", and the report notes it. No credit used.

Implausible year-built values from the public record (one home showed 1111) now display as unknown instead of skewing the age adjustment.

"Edit details" on the Results subject card lets you correct beds, baths, living area, lot size and year built. Comps re-score and every pricing figure recomputes instantly; no credit is used. The report discloses which details you corrected, and "Restore record" undoes it.

If the living area changes by 10% or more, or the bedroom count changes, the public estimated value is removed — it was calculated for the wrong facts — and the analysis anchors on the comparable sales instead.

Coverage claims for non-disclosure states, new-construction builders and rural markets were rewritten to say what the data actually supports.

If a renewal payment fails, a banner now explains the grace period and links to Billing, instead of access silently ending.

A support chat is available from the floating button on every page and at /support. It knows the product, your plan and your remaining searches, and it's the fastest way to reach us.

Estimate a seller's net proceeds from sale price, payoff, commission and closing costs at /seller-net-sheet-calculator.

If your subscription has lapsed, the app now says so and offers to reactivate it in one step, rather than showing the sign-up pricing page as if you were new.