How to Do a CMA That Wins the Listing (2026)

Almost every listing presentation guide teaches the room: what to wear, what to bring, how to close. Very little of it teaches the thing the appointment is actually decided on, which is whether your number is right and whether you can defend it. A seller can tell within a few minutes whether you built a price or borrowed one. This guide covers the analysis itself, from selecting comparables through deriving adjustments to reconciling a final opinion, using the same standards appraisers work under. It is the most requested topic in our real estate listing agent coaching sessions, and it is where most agents are quietly weakest.

What the CMA is actually for

Most agents treat the CMA as a document they produce on the way to an appointment. It is a page in the packet, printed the night before, assembled from whatever the MLS auto-comp tool suggested. That framing is why so many listing appointments come down to personality.

The CMA is the only part of the appointment where you demonstrate a skill the seller cannot perform themselves. They can look up their own address on a portal. They can read your reviews. They can compare commission rates. What they cannot do is look at eleven recent sales in their neighborhood and explain which four actually competed with their house and why the other seven are noise.

That is the whole job. Everything else in the presentation is packaging.

There is a second reason the analysis matters more than the presentation, and it is commercial. The list price you recommend is the single decision in the transaction with the largest dollar consequence. Get it wrong on the high side and you spend three months earning nothing while the listing goes stale. Get it wrong on the low side and you have cost your client real money. Pricing skill is what makes the rest of the pipeline work, which is why it sits underneath most of the tactics in our guide to how to get listings.

National market data as of mid 2026 gives you the backdrop. NAR reported the median existing home price at 440,600 dollars in June 2026, with inventory at 1.56 million units and 4.6 months of supply, and a median 28 days on market. Redfin's national figures for April 2026 showed a sale to list price ratio of 98.4 percent, with 21.2 percent of homes taking a price drop and median days on market at 49.

Read those two data sets together and the picture is a market where pricing errors get punished but not instantly. Roughly one listing in five is cutting price. The sale to list ratio sitting just under 99 percent means the average seller is landing close to asking. Sellers who priced correctly are getting paid. The ones who did not are the ones showing up in the price drop column.

The gap between those two columns is the argument for doing the work yourself. A national median tells you almost nothing about what a specific house on a specific street will bring. Northern Virginia in June 2026 was running at under two months of supply with an average of 19 days on market. That is a fundamentally different pricing environment than the national 4.6 months and 28 days.

If the national number is misleading at the metro level, it is far more misleading at the neighborhood level, and neighborhoods are where you list houses.

CMA, appraisal, BPO, and AVM are four different things

Agents use these terms loosely and it creates real exposure. They are separate products with separate legal standing, and the differences are worth knowing precisely.

An appraisal is prepared by a licensed or certified appraiser working under the Uniform Standards of Professional Appraisal Practice. A broker price opinion is prepared by a licensed agent or broker, commonly for a lender, servicer, or asset manager, and it usually follows a required form. A comparative market analysis is prepared by an agent for a consumer, most often in the context of listing or purchasing. An automated valuation model is a statistical estimate produced by software with no human inspection.

The federal line between them is specific. Under 12 U.S.C. section 3355, enacted through the Dodd Frank Act in July 2010, a broker price opinion may not be used as the primary basis to determine the value of a property. That restriction applies to loan origination for a residential mortgage secured by a consumer's principal dwelling. The same section defines a BPO as an estimate prepared by a real estate broker, agent, or salesperson, and expressly excludes automated valuation models from the definition.

What that means in practice is simple. Your CMA is a marketing and advisory tool. It is not a substitute for an appraisal in a lending decision, and you should never let a client believe otherwise.

The ethics side matters too. NAR's Code of Ethics, in the version effective January 1, 2026, addresses this directly. Article 11 requires that services conform to the standards of practice and competence reasonably expected in the specific real estate disciplines involved.

Standard of Practice 11-1 is the operative one for CMAs. It requires that when preparing an opinion of value the member be knowledgeable about the type of property and familiar with the area. It also requires access to the information and resources necessary to formulate an accurate opinion. Any lack of those must be disclosed to the client in advance.

It also requires that an opinion of value prepared other than as an appraisal include a statement that it is not an appraisal. It must also disclose whether a physical inspection of the exterior and interior was performed.

That last requirement is one most agents ignore. If you built the analysis from MLS photos without walking the property, say so on the document. It takes one line and it is the difference between a defensible opinion and an exposed one.

One more note on that last row, because it changed recently and most agents missed it. The federal banking agencies, together with the CFPB and FHFA, issued joint quality control standards for automated valuation models. The rule was published in the Federal Register on August 7, 2024 and took effect October 1, 2025.

It requires institutions using AVMs in certain credit decisions to maintain policies designed to ensure a high level of confidence in the estimates. Those policies must protect against data manipulation, avoid conflicts of interest, conduct random sample testing and reviews, and comply with applicable nondiscrimination laws.

That is a useful fact to have in your pocket. Federal regulators wrote a rule specifically because AVM output needed controls before a lender could lean on it. A consumer facing estimate on a portal is not operating under those controls at all.

Selecting comparables: the standard appraisers use

Comparable selection is where a CMA is won or lost. The adjustments get all the attention, but if you pick the wrong four houses no amount of adjustment arithmetic rescues the conclusion.

Borrow the appraisal standard, because it is objective and because it is the standard the lender's appraiser will later apply to the same house. Fannie Mae's Selling Guide section B4-1.3-08, in the version published June 4, 2025, states that a minimum of three closed comparables must be reported in the sales comparison approach.

Treat three as a floor, not a target. Three comparables that all needed heavy adjustment is a weaker analysis than five that needed light adjustment. The number is less important than the quality of the match.

The definition of market area in the same section is the part worth internalizing. Fannie Mae defines it as the geographic region, for a subject property, from which most demand comes and in which most of the competition is located.

Read that again, because it is not a radius. It is a behavioral definition. The market area is wherever the buyers who would consider this house are also looking.

This is why the one mile rule that circulates in agent training is wrong as a rule and only sometimes right as a habit. In a dense, uniform subdivision, a quarter mile may already be too far because it crosses into a different school assignment. In a rural market, the nearest genuinely competing sale may be four miles away and still be the best comparable available.

Fannie Mae's own appraisal FAQs, updated November 2023, address the distance question head on. Comparable sales located a considerable distance from the subject property can be used if they represent the best indicator of value.

What the guide does require is disclosure of the distance. The appraiser must be specific with respect to the distance in terms of miles and include the applicable directional indicator, in the format of 1.75 miles NW. Put that on your CMA too. It signals that you thought about geography rather than accepting whatever the software returned.

The fourth item on that list deserves emphasis because it is the one most often missed. A comparable that closed thirty days ago may have gone under contract ninety days ago. The contract date is when the price was negotiated. The closing date is just paperwork catching up.

When you are working in a market that moved, sort your comparables by contract date rather than settlement date and the pattern usually becomes clearer immediately.

The twelve month window and when an older sale is better

Fannie Mae's guidance on recency is a single sentence that agents routinely over read. Comparable sales that have closed within the last 12 months should be used in the appraisal.

The sentence that follows in the same section is the one nobody quotes. The best and most appropriate comparable sales may not always be the most recent sales.

Fannie Mae's appraisal FAQ, updated November 2023, expands on it. A sale more than 12 months old may be more appropriate when market conditions have impacted the availability of recent sales. The appraiser must account for any changes in market conditions.

That is permission to use judgment, and it matters in three specific situations that come up constantly.

The first is the unusual property. Say you are pricing the only five bedroom colonial on a street of three bedroom ranches. A fourteen month old sale of the genuinely similar house two streets over beats a fresh sale of a house that is not similar at all.

The second is the thin market. Some neighborhoods produce four sales a year. Insisting on a twelve month window there means insisting on four data points, some of which may not be comparable for other reasons.

The third is the market that has not moved. If prices in a submarket were flat for eighteen months, the age of the comparable carries much less risk than it would in a market that appreciated 8 percent over the same period.

What you cannot do is use an older comparable and then ignore what happened in between. That is the trade. Reaching back in time obligates you to support a market conditions adjustment, which is covered in its own section below.

What actives, pendings, and expireds tell you

Closed sales tell you what the market paid. They do not tell you what the market is paying right now, and in a moving market that difference is the whole game.

Fannie Mae's position on this is narrow and worth knowing. Section B4-1.3-08 permits contract offerings and current listings to be used as supporting data if appropriate. For new construction projects the guide allows the appraiser to use two pending sales in the subject project in lieu of one settled sale.

Supporting data is the operative phrase. Listings are not evidence of value, because nobody has agreed to pay those numbers. They are evidence of competition, which is a different and equally useful thing.

Each status category answers a different question, and treating them as interchangeable is a common error.

The expired column is the one agents underuse. When a seller pushes for a number above your recommendation, the strongest response is not an argument. It is three expired listings at that number with their days on market attached. For the follow up side of that story, our breakdown of expired listing scripts covers what those sellers say when a second agent calls.

Redfin's April 2026 national data put the share of homes with a price drop at 21.2 percent. That figure is a national average across every market condition in the country, so pull your own local equivalent before you quote a number to a seller. In a fast submarket the local share may be far lower, and in a slow one it may be much higher.

Pendings deserve a specific habit. Build a standing search for pendings in your farm area and check the sale prices as they record. In a market that is turning, pendings are the first place the change becomes visible, usually six to ten weeks before it shows up in closed sale medians.

Deriving adjustments the market can defend

Here is the uncomfortable part. Most agent CMAs contain adjustments that came from nowhere. A round number for a bedroom, a familiar figure per square foot, fifteen thousand for a finished basement because that is what everyone in the office uses.

The standard is not that. Fannie Mae's Selling Guide section B4-1.3-09 was published June 4, 2025. It requires that the appraiser's adjustments reflect the market's reaction, meaning market based adjustments, to the difference in the properties.

The market's reaction. Not your estimate of what the feature cost to build, and not what the seller thinks it is worth. What buyers actually paid differently because of it.

USPAP takes the same position from the other direction. Standards Rule 1-4(a) covers it. When a sales comparison approach is necessary, an appraiser must analyze such comparable sales data as are available to indicate a value conclusion.

Fannie Mae names the acceptable derivation methods in the context of time adjustments, and the same list applies generally. Those methods are home price indices, statistical analysis, modeling, paired sales, and other commonly accepted methods.

Paired sales analysis is the one you can actually run without software. Learn it properly, because it is the method you will cite when a seller or an appraiser challenges a number.

A practical warning about square footage adjustments. There is no authoritative published schedule of dollar per square foot adjustments anywhere, from Fannie Mae, the Appraisal Foundation, or any other standards body. The figure is market derived only.

More importantly, the correct adjustment is never the market's average price per square foot. If houses in a neighborhood sell around 300 dollars per square foot, the adjustment for a hundred extra square feet is not 30,000 dollars.

The reason is that the average price per foot includes the land, the location, the systems, the kitchen, and everything else. The contributory value of one additional square foot of similar space is a fraction of that blended figure, and the fraction varies by market.

Condition and updating are harder than size and generally require more judgment. Two houses with identical bedroom counts can be twelve years apart in effective age because one has a renovated kitchen and mechanical systems that were replaced.

The honest approach is to bracket. Find one comparable clearly superior in condition and one clearly inferior, then place the subject between them. Bracketing is more persuasive than a precise looking number you cannot support, and it is what a careful appraiser does with the same problem.

The 15 and 25 rule and other CMA folklore

Ask ten agents about adjustment limits and most will tell you no single adjustment can exceed 15 percent and total gross adjustments cannot exceed 25 percent. Some will add that it is a Fannie Mae rule.

It is not. Fannie Mae's Selling Guide section B4-1.3-09 states directly that Fannie Mae does not have specific limitations or guidelines associated with net or gross adjustments.

Fannie Mae's appraisal FAQ document puts it even more plainly, answering the question with a flat no. It repeats that there are no specific limitations or guidelines associated with net and gross adjustments for comparable sales.

This matters more than a trivia correction, for two reasons.

The first is that agents who believe the rule exists sometimes shave adjustments to stay under an imaginary threshold. That is backwards. It replaces a supported number with an unsupported one in order to satisfy a guideline that was never written.

The second is that large adjustments do carry real information, just not the information the myth suggests. A comparable requiring a 30 percent gross adjustment is telling you it is probably not a good comparable. The right response is to question the selection, not to trim the arithmetic.

Market conditions adjustments

A market conditions adjustment, often called a time adjustment, accounts for price movement between the date the comparable was negotiated and today. It is the adjustment agents are least comfortable making and the one that most often decides whether a CMA is right.

Fannie Mae strengthened its position here recently. Announcement SEL-2024-07, issued November 6, 2024, updated market area analysis requirements and added standardized glossary definitions for market area terms.

The current Selling Guide language in section B4-1.3-09 is direct. Time adjustments, or the lack thereof, must be supported by evidence.

Note the phrase "or the lack thereof." Deciding that no time adjustment is warranted is itself a conclusion that requires support. You cannot skip the analysis by declaring the market flat.

On what counts as support, Fannie Mae is specific and reasonably generous. Use of home price indices to support time adjustments is consistent with policy. Adjustment rates can also be determined through statistical analysis, modeling, paired sales, or other commonly accepted methods.

The documentation standard is that the report must at minimum summarize the supporting evidence and include a description of the data sources, tools, and techniques used.

Apply the same discipline to your CMA and you will be ahead of most of the appraisals that follow it.

A caution on direction. Time adjustments go both ways, and agents apply them asymmetrically. In a rising market almost everyone remembers to adjust older comparables upward. In a softening market the same agents leave the adjustment at zero because the conclusion is unwelcome.

That asymmetry is how listings end up overpriced by exactly the amount the market fell.

Build the grid

The adjustment grid is where the analysis becomes visible. Every comparable gets a row, every difference gets a line, and the adjusted value at the bottom is what that comparable indicates the subject is worth.

The tool below builds one. Enter your subject property, the adjustment rates you derived from paired sales rather than from habit, and three comparable sales. It returns each adjusted value and the gross and net adjustment percentages that tell you how good each comparable actually is. It also returns a reconciled range weighted toward the comparables that needed the least adjustment.

Two things in that output matter more than the reconciled number itself.

The gross adjustment percentage is a quality score for each comparable. A sale that needed 6 percent total adjustment is telling you it genuinely competed with the subject. One that needed 28 percent is telling you it did not.

The spread between the three indications is a confidence measure. When three well selected comparables land within a few percent of each other, you can present a tight range with conviction. When they scatter, that scatter is real information about the market and should be explained rather than averaged away.

Reconciling three answers into one

Reconciliation is the step agents skip. The grid produces three adjusted values, they get averaged, and the average becomes the recommendation.

Averaging is the wrong operation. It assumes every comparable is equally informative, which is exactly what the gross adjustment column just told you is false.

Reconciliation is weighted judgment. You are deciding which comparables deserve to drive the conclusion and saying why.

The second item is the one worth dwelling on, because it catches a specific and common failure.

Imagine a comparable that is 400 square feet larger than the subject and also missing a full bathroom. The size adjustment runs downward, the bath adjustment runs upward, and they roughly cancel. The net adjustment looks like almost nothing.

That comparable is not well matched. It is two significant differences that happened to point in opposite directions. If either adjustment rate is off, and rates derived from small paired sample sets often are, the error does not cancel. It compounds.

This is exactly why the gross figure exists as a separate line from the net figure. Net tells you the direction of the conclusion. Gross tells you how much the conclusion depends on your adjustment rates being right.

Price, range, or band: what to present

You now have a supported conclusion. The question is what shape to give it when it reaches the seller.

Three approaches are common and two of them cause problems.

Presenting a single number implies a precision the data does not support. Real estate does not price to the dollar, and a seller who hears one number will treat any offer below it as a failure, including offers that are entirely reasonable.

Presenting a wide range without a recommendation moves the decision to the seller. Sellers pick the top of the range every time. Handing over a range from 620,000 to 680,000 is functionally recommending 680,000, and you have surrendered the pricing conversation while appearing to be thorough.

The approach that works is a range plus a recommendation inside it, with the reasoning attached to both.

The last row is the version I use, and it is worth building into your template. It slots into the pricing section of the realtor listing presentation checklist. Rather than arguing about a number, present what each pricing strategy is likely to produce.

An aggressive price at the top of the supported range trades time for the chance of a premium. A price at the reconciled indication targets the most likely outcome. A price slightly below the indication trades a small amount of upside for speed and competition.

Each of those is a legitimate choice depending on the seller's situation, and framing it that way converts a disagreement into a decision. A seller relocating on a corporate timeline and a seller with no deadline should not price the same house identically, and they will recognize that immediately when you lay it out.

What none of the three scenarios includes is a price above the supported range, and that is the point of showing all three. The menu defines the boundaries without you having to say no directly.

Defending your number against an AVM

Every listing agent has had this conversation. The seller has an estimate from a portal, it is higher than your number, and it arrived before you did.

The instinct is to attack the estimate. That rarely works, because the seller heard it first and because attacking it sounds like self interest.

The better approach is to explain what an automated model can and cannot see, using facts rather than dismissal.

Start with what the models themselves publish. Redfin states that its estimate has a median error rate of 1.87 percent for homes listed for sale and 7.28 percent for off market homes. Redfin adds that its off market estimate will be within 7.28 percent of the eventual sale price half the time. That page carried an update notation of September 2025, so confirm the current figure before quoting it.

Zillow publishes its own accuracy figures for the Zestimate on its accuracy page, broken out by state and by on market and off market status. Pull the current national and state numbers directly from that page on the day you build the presentation rather than relying on a figure quoted in a blog post.

The phrase "half the time" is the one to slow down on with a seller. A median error rate of 7.28 percent means half of all estimates are off by more than that. On a 700,000 dollar home, an error of 7.28 percent is nearly 51,000 dollars, and half the time the miss is larger.

The federal regulatory angle is a second and underused argument. The CFPB, OCC, Federal Reserve, FDIC, NCUA, and FHFA jointly issued quality control standards for automated valuation models. The rule was published in the Federal Register on August 7, 2024 and took effect October 1, 2025.

The rule requires institutions to adopt policies designed to ensure a high level of confidence in AVM estimates. Those policies must protect against data manipulation, avoid conflicts of interest, run random sample testing, and comply with nondiscrimination law.

Put plainly to a seller: federal regulators wrote a rule requiring lenders to test and control these models before relying on them in a credit decision. The free version on a consumer website operates under none of those controls.

There is also a structural point worth making about medians generally. A median error rate says nothing about dispersion, and error is not evenly distributed. It concentrates in properties that are unusual for their area, in neighborhoods with few transactions, and in rural markets where the model has little to learn from.

FHFA working paper 18-03 by Bogin and Shui, originally published April 2018 and revised March 2019, is now eight years old but illustrates the geographic pattern. It found that approximately 25 percent of properties in rural areas were appraised at more than five percent above contract price, compared with 12.7 percent in urban areas.

The practical translation is that the more distinctive your seller's property is, the less an automated estimate is worth. That is a comfortable thing to tell a seller, because most sellers already believe their house is distinctive.

When the seller wants more than the data supports

The analysis is finished and the seller wants 60,000 dollars more. This is the moment the listing is actually won or lost, and no amount of additional data solves it by itself.

What usually fails is arguing. You brought evidence, they brought a number they have been carrying for two years, and evidence does not dislodge an anchor.

What works is separating the two questions that are getting tangled: what the house is worth, and what to list it at. Those are different conversations, and sellers conflate them constantly.

The sixth step is the one that saves listings, and almost nobody does it at the appointment.

Agree in advance that you will revisit price if the listing has fewer than a set number of showings in the first two weeks, or has no offer by a specific date. That agreement changes the nature of the later conversation entirely. You are not asking for a reduction. You are executing a plan the seller already agreed to.

Put it in writing in the listing packet. It costs nothing at the appointment and it removes the single most uncomfortable conversation in the business. When the trigger does fire, these price reduction conversation scripts cover the call word for word.

There is also a version of this where the right answer is to decline the listing. If a seller insists on a price well above the supported range and will not agree to a review trigger, look closely at what you are being asked to do. It is funding a marketing campaign for a price the data says will not happen.

Walking away from those is a skill, and it is one of the clearest markers of agents who have moved past chasing volume. If you want help building the judgment for when to hold a number and when to walk, that is a core part of our real estate coaching and training work with listing agents.

CMA mistakes that cost the listing

These are the failures I see most often when reviewing agents' work, ordered roughly by how much damage they do.

The last one deserves a closing note because it is the cheapest fix on the list and the one with the largest effect on outcomes.

Redfin's April 2026 data showed 21.2 percent of homes taking a price drop nationally. Price reductions are not rare events, they are a normal part of a meaningful share of listings, and the only question is whether the conversation was planned or improvised.

Planned, it is a scheduled review that both parties agreed to. Improvised, it is the agent's fault. Same reduction, entirely different relationship. The wording matters as much as the timing, and our listing presentation closing lines cover how to set that expectation at the appointment.

Frequently asked questions

What is a CMA in real estate?

A comparative market analysis is an agent's supported opinion of what a property would sell for in current market conditions. It is built by selecting recent sales of competing properties and adjusting each one for the ways it differs from the subject. Those adjusted figures are then reconciled into a single conclusion or a narrow range. It is not an appraisal and cannot be represented as one.

How many comparable sales should a CMA include?

Three closed sales is the working minimum, because that is the minimum Fannie Mae requires of appraisers in the sales comparison approach under Selling Guide section B4-1.3-08. Three is a floor rather than a target. Most credible analyses use three to six closed sales, supported by current listings and pending sales that show where the market is heading.

How far back can comparable sales go in a CMA?

Twelve months is the standard window. Fannie Mae's Selling Guide states that comparable sales closed within the last 12 months should be used. It adds that the best and most appropriate comparable sales may not always be the most recent sales. An older sale can be the better comparable when recent competing sales are scarce, provided you account for how the market moved in between.

Is there a 15 percent net and 25 percent gross adjustment rule?

No. This is the most widely repeated piece of misinformation in residential valuation. Fannie Mae states plainly in Selling Guide section B4-1.3-09 that it does not have specific limitations or guidelines associated with net or gross adjustments. Large adjustments are a signal to question your comparable selection, not a violation of a rule that does not exist.

Can a real estate agent's CMA be used as an appraisal?

No. Under 12 U.S.C. section 3355, a broker price opinion may not be used as the primary basis for determining the value of a property. That applies when the purpose is originating a residential mortgage loan secured by a consumer's principal dwelling. NAR's Code of Ethics also requires that an opinion of value prepared other than as an appraisal state clearly that it is not an appraisal.

How do you derive adjustments between comparable sales?

From the market, not from a rule of thumb. Fannie Mae requires that adjustments reflect the market's reaction to the difference between properties. The named acceptable methods include paired sales analysis, statistical analysis, modeling, and home price indices. A dollar per square foot figure someone gave you at a sales meeting is not a derivation.

Why is the Zestimate different from my agent's price opinion?

An automated valuation model has never been inside the property. It cannot see a renovated kitchen, a failing roof, a bedroom that is legally a den, or the fact that the house backs to a highway. Published AVM error rates are national medians, which means half of all estimates miss by more than the stated figure, and error concentrates in unusual properties and thin markets.

Should a CMA present one price or a range?

Present a range that reflects genuine uncertainty, then recommend a specific list price inside it. A range alone leaves the seller to pick, and sellers pick the top. A single number without a range implies a precision the data does not support. The combination shows your work and still gives a recommendation the seller can act on.

Do I have to disclose whether I inspected the property?

NAR's Code of Ethics addresses this in Standard of Practice 11-1. An opinion of value prepared other than as an appraisal must disclose whether an exterior and interior physical inspection was performed. Most agents omit this. Adding one line to your CMA template satisfies it permanently.

What if the appraisal comes in below my CMA?

It happens, and it is not automatically a failure of your analysis. The appraiser selects comparables independently and may reach a different conclusion from the same data. Keep your paired sales work and your comparable selection reasoning on file, because a documented analysis is what supports a reconsideration of value request.

Educational content only, and not legal, appraisal, or financial advice. Nothing here is an appraisal and nothing here should be represented as one. Appraisal and comparable selection standards cited come from the Fannie Mae Selling Guide sections B4-1.3-08 and B4-1.3-09 published June 4, 2025. Additional Fannie Mae sources are the appraisal FAQs updated November 2023 and Announcement SEL-2024-07 issued November 6, 2024. The USPAP citation is Standards Rule 1-4(a). Market figures come from NAR existing home sales data for June 2026, Redfin national data for April 2026, and NVAR market statistics for June 2026. The AVM error figure cited is Redfin's own published rate carrying a September 2025 update notation. The interagency AVM quality control rule was published in the Federal Register on August 7, 2024 and took effect October 1, 2025. The BPO restriction is 12 U.S.C. section 3355. The ethics citations are NAR Code of Ethics Article 11 and Standard of Practice 11-1 in the version effective January 1, 2026. All were current as of August 2026. Standards, guidelines, and published error rates change, and state licensing law governs what you may prepare and how you must label it.