Two of the most common ways to put a number on a home pull in opposite directions. An automated valuation model (AVM) gives you an instant figure from an algorithm; a comparative market analysis (CMA) gives you a considered figure from an agent's judgement. Both are useful, both are misused, and choosing the wrong one for the moment costs you either accuracy or hours you didn't have. This guide explains exactly how the two methods work, where each is reliable, where each fails, and a simple decision rule for which to reach for — with a worked example that shows why the two can legitimately disagree on the same house.
An automated valuation model is a statistical engine. It ingests public records, tax assessments, prior sale prices and the attributes of nearby homes, then uses a model — often a regression or a machine-learning method such as a hedonic or repeat-sales approach — to predict what a given property would sell for today. The whole thing runs in under a second and costs almost nothing per estimate, which is why Zestimate-style figures appear all over consumer property portals and why lenders use AVMs to screen low-risk loans. The catch is that no human ever looks at the home: the model knows the house has three bedrooms and 95 square metres, but not that the kitchen was gutted last year or that it backs onto a railway line. For the full mechanics and accuracy bands, see our deep dive on what an AVM is and how accurate it is.
What a CMA actually does
A comparative market analysis is the valuation an agent builds by hand. You select a handful of genuinely comparable recent sales, adjust each one for differences against the subject property — extra bathroom, smaller plot, better condition — and triangulate to a defensible price or range. You then layer in the live picture: what's currently on the market competing for the same buyers, what's gone under offer, and what failed to sell. The result is slower to produce than an AVM, but it reflects the specific property and the agent's read of the local market. If you want the step-by-step method, our guide on how to create a CMA walks through it, and finding the right comparable sales covers the comp selection the whole analysis rests on.
AVM vs CMA: the key differences
Both methods lean on comparable evidence, but they differ on almost every practical axis that matters in your day:
Speed. An AVM is instant; a CMA takes a human anywhere from twenty minutes to an hour or more.
Cost. AVMs are near-free at scale; a CMA costs your time.
Inputs. An AVM sees recorded data only; a CMA sees condition, renovations, layout, light and kerb appeal.
Judgement. An AVM applies a fixed model to every home; a CMA lets you discard a bad comp and weight a great one.
Defensibility. An AVM is a black box you can't argue in a listing appointment; a CMA shows the seller the evidence behind the number.
Best use. AVMs win on volume and screening; CMAs win on any decision a client will act on.
Where each one is accurate — and where it isn't
AVMs are accurate in aggregate and in markets that suit them: dense, uniform housing stock with plenty of recent arms-length sales — think a postwar suburb of near-identical homes. There, the model has rich data and little variation to misjudge. They struggle with unique, renovated, rural or thinly traded properties, where one missing fact swings the value by tens of thousands. A CMA is the mirror image: it shines exactly where the AVM struggles, because a human can find the three true comps in a thin market and adjust for the renovation the model never saw. But a CMA is only as good as the agent — weak comp selection or wishful adjustments produce a confident, wrong number. The deeper question of how far to trust any algorithmic figure is covered in how accurate online home value estimates really are.
A worked example: same house, two numbers
Suppose you're asked to value a three-bedroom terraced house that the owner renovated top to bottom eighteen months ago. All figures are illustrative, to show the method rather than any real market.
The AVM pulls the last recorded sale (pre-renovation) and nearby comps, and returns €340,000 with a wide confidence range. It has no way of knowing about the new kitchen, rewiring and loft conversion, so it anchors to the home's older, plainer profile.
The CMA starts from three genuinely similar recent sales at €352,000, €358,000 and €366,000. You adjust upward for the renovation and the converted loft (roughly +€18,000 of value the AVM ignored), and downward slightly for a smaller garden, landing on €362,000.
The gap — about €22,000, or 6% — is not error so much as information. It is almost entirely the renovation the model couldn't see.
That gap is the whole story. The AVM gave you a defensible starting point in one second; the CMA explained why the right number sits above it. If you'd listed at the AVM figure you'd have left money on the table; if you'd ignored the AVM entirely you'd have lost a fast cross-check. Turning that final number into an asking-price strategy is its own step — see how to price a listing for the strategy layer on top of the valuation.
The decision rule: which to use when
The rule is simple: match the method to the stakes. When you need speed or scale and a rough number is fine — qualifying a lead, powering a website valuation tool, screening which listings deserve a full workup, sense-checking a portfolio — reach for the AVM. When a real decision rides on the number — a listing appointment, an offer, a price reduction, a seller who needs convincing — build the CMA, because condition and judgement are what move the figure and what a client can challenge. And the best operators don't choose at all: they run the AVM first for an instant baseline, then build the CMA to refine it, treating any large gap between the two as a prompt to investigate rather than a number to average. A close-up of how a CMA differs from a lender's formal value opinion is in broker price opinion (BPO) vs CMA.
How software lets you run both at once
The reason agents historically picked one method was friction: the AVM was a portal you visited, the CMA was an hour of pulling and adjusting comps by hand. Property-analysis software collapses that. With Biedradar, you enter an address and it gathers comparable sales, recent listings and market signals, then generates a clean, branded valuation report in minutes — giving you the instant baseline of an AVM and the comp evidence a CMA is built on, in one place. You keep the judgement that matters: which comps are truly comparable, how to adjust for the renovation the data never recorded, and which number the situation actually calls for. The tool does the data assembly at AVM speed; you supply the human read that turns it into a CMA a client will trust. For an agent juggling lead qualification and listing appointments in the same week, that single workflow is what makes running both methods practical instead of a choice between fast and accurate.
Frequently asked questions
What is the difference between an AVM and a CMA?
An AVM (automated valuation model) is an algorithm that estimates a property's value instantly from public records, past sales and statistical models — no human looks at the home. A CMA (comparative market analysis) is a valuation an agent builds by hand: selecting genuinely comparable sales, adjusting them for differences, and weighing condition, location and current competition. The AVM is fast and cheap but blind to specifics; the CMA is slower but applies human judgement to the exact property.
Is an AVM more accurate than a CMA?
Usually not for an individual home. AVMs are accurate in aggregate and in dense, uniform markets with lots of recent sales, but they miss condition, renovations, layout quirks and micro-location, so any single estimate can be well off. A well-built CMA accounts for exactly those factors. The honest framing: an AVM is a fast starting point and sanity check; a CMA is the defensible number you put in front of a client.
When should an agent use an AVM instead of a CMA?
Use an AVM for speed and scale: an instant ballpark on a lead, a website valuation widget that captures seller enquiries, a quick screen before deciding whether a listing is worth a full workup, or a portfolio-level read. Use a CMA whenever a real pricing decision is on the table — a listing appointment, an offer strategy, a price reduction conversation — because that is where condition and judgement decide the number.
Can you use an AVM and a CMA together?
Yes, and that is the strongest workflow. Run the AVM first for an instant baseline and to flag obvious outliers, then build the CMA to refine it with comps you have actually vetted. If the two land close together, your confidence rises; if they diverge sharply, that gap is a signal to dig into why — a renovation the model can't see, a bad comp, or a thin local market.
Do lenders accept an AVM or do they need a CMA or appraisal?
It depends on the loan and the risk. Lenders increasingly use AVMs for low-risk decisions — refinances, low loan-to-value loans, portfolio monitoring — because they're instant and cheap. Higher-stakes lending still relies on a full appraisal by a licensed appraiser. A CMA is an agent's pricing tool, not a lender's underwriting document, so it sits alongside these rather than replacing the appraisal.
