Two readings from the same benchmark, months apart, tell you whether movement is active or historic. One reading cannot.
2 readingsthe minimum for a trendA second reading months later shows whether movement is active or historic. It is a notebook, not an elevation survey.
This log compares two floor elevation readings, in inches from a benchmark you can find again, and the months between them.
It returns the change and the rate, then says whether that looks like little change or real movement.
It is not an elevation survey, it does not name a cause, and it never says you need piers.
Two elevations, then the interval
One reading is not a trend. Not a survey. Nothing is emailed.
—Difference
—What two points can say
—What this page will not do
What this assumed—
Your two numbers, months apart. Change is the question. Not engineering. Not a survey. HyreFoundation does not survey houses.
The mechanism
Out of level is a condition. Still moving is a problem
Being out of level and still moving get conflated constantly, and the conflation is worth money to somebody. They are different measurements answering different questions.
Level is a snapshot. Change is a process. A house can be an inch and a half out across a room and have been exactly that since the Eisenhower administration. Another can be nearly true today and be moving a sixteenth of an inch a month.
The second one has a problem; the first one has a history. A single reading cannot tell them apart, and neither can a photograph, a marble on a floor, or a forty-minute visit from anyone.
Why buildings tolerate being out of level. Materials accommodate slow strain. Plaster, framing, brickwork and finishes all creep, and given years they redistribute stress rather than concentrate it.
Concrete does the same thing at the material level — ACI notes that "The tensile stress induced by restraining drying shrinkage is reduced with time due to creep or stress relaxation. Cracks develop only when the net tensile stress reaches the tensile strength of concrete."
It is the rate that determines whether a structure accommodates movement or cracks in response to it.
Our analysis — what this makes the elevation log. A rate meter. Not a levelness meter.
Its output is inches per month, and the only way to get that number is to have taken a reading before you needed one.
That is the awkward property of this whole subject: the most useful measurement is the one you take when nothing is wrong.
Which is why this is the cheapest thing on the site. It costs a flashlight, a pencil and twenty minutes twice a year. Nobody sells it, and by the time a repair company is standing in your kitchen it is far too late to start.
And why it changes the conversation later. When somebody tells you the house is actively failing, "here are seven readings from the same four points over three years" is a different kind of reply from "it looks the same to me".
It moves the discussion from impressions to evidence, and you are the only person who can have started collecting it.
Instruments
Four ways to read a floor, and what each one is really good for
The last column matters most. Each method is genuinely useful for something and useless for something else, and the useless part is never mentioned in the product description.
Method
Roughly what it can resolve
What it is good for
What it cannot do
Phone level app
Slope, coarsely. Sensitive to how the phone is held and to the surface it sits on.
A free first look and a rough sense of which way a room falls. Fine for deciding whether to bother with anything else.
Repeatability. The reading changes with the case, the spot, the temperature of the sensor and the operator. Two phone readings months apart are not comparable enough to establish a rate, which is the only thing you actually want.
Long spirit level and shims
Fractions of an inch over the length of the level.
Checking a specific short span — a door threshold, a section of floor. Cheap, repeatable if you mark the exact position.
Anything across a whole house. Errors accumulate as you leapfrog the level, and by the fourth move you are measuring your technique.
Water level (clear tube)
Very good over long distances, because water finds its own level regardless of the path between the ends.
Comparing distant points in a house cheaply and surprisingly accurately. Works around corners and through doorways, which is exactly where optical methods struggle.
Being convenient. It is slow, needs two people or great patience, and is affected by bubbles and by temperature differences along the tube.
Rotary or laser level with a rod, or an engineer’s survey
Hundredths of a foot across a whole building, recorded against a fixed benchmark.
The real thing. This is what an elevation survey means, and it produces the drawing that a structural dispute will actually turn on.
Being free, and being yours. It is a professional service, and its value depends on who commissioned it — which is a question about incentives, not accuracy.
Our analysis. Nothing on this list makes a homeowner’s readings into an elevation survey, and the tool above does not pretend otherwise. What homeowner readings do is establish whether there is a trend worth paying a professional to measure properly.
Reading the number
What a rate actually means, and why one reading can never produce one
The calculator divides a difference by an interval. Almost everything that can go wrong with that arithmetic is about the interval, not the difference.
Illustrative only — drawn curves, not data. The point is the sampling trap: the same house yields opposite conclusions depending on when you happened to measure.HyreFoundation illustration; the seasonal mechanism is the wet-dry cycling described by the Colorado Geological Survey
Source fact — why the seasonal curve exists at all. The Colorado Geological Survey states that expansive soils can "expand up to 20% by volume when exposed to water" and can exert "Expansive soils can exert a force of up to 30,000 pounds-per-square-foot".
Soil that swells when wet and shrinks when dry moves the building above it up and down with the seasons, and in some regions that annual cycle is larger than a decade of real settlement.
Our analysis — the first sampling trap. Two readings three months apart in a clay region are mostly measuring the weather.
To see a trend rather than a season you want readings spanning at least a full year, and preferably taken at the same point in the cycle — late summer against late summer, spring against spring. That single discipline removes most of the noise for free.
The second trap: the benchmark moved. A rate is a difference between two points, and it is only meaningful if the reference did not move.
Take your reference off something that is unlikely to shift and record precisely where it is — a photograph of the exact spot with a mark on it is worth more than a written description.
If the reference is on the part of the house that is moving, the log will be quietly and confidently wrong.
The third trap: it is not the same spot. Half an inch along a floor is a different reading. Mark the measurement points physically and photograph them. Four marked points beat twenty unmarked ones, every time.
Our calculation, and its honest label. The tool divides the change in your two readings by the months between them and reports inches per month. That is arithmetic on numbers you supplied.
It is not a measurement we made, not a survey, and not an opinion about your foundation. If your inputs carry any of the three errors above, the output carries them too — a calculator cannot detect a moved benchmark.
What to do with a rate that looks real. Do not act on two readings. Take a third. A trend confirmed across three observations spanning a year is the point at which paying a licensed engineer for a proper elevation survey stops being cautious and starts being obvious.
The document that settles arguments
Whose elevation survey is it, and why that question decides its value
An elevation survey is the closest thing to objective data in this trade. Who commissioned it changes what it is for.
What an elevation survey actually is
Measured relative floor heights across the whole building, recorded against a fixed reference and drawn on a plan. It converts "the floor feels wrong in the dining room" into a contour map of a house, and it is the single document that a later structural argument turns on.
It is not a topographic survey of the lot, and it is not a level reading on one floor. It is the whole building, at once, from one reference.
Why the commissioning matters
The Texas Board of Professional Engineers held in February 2018 that "Analyzing the cause of a condition, recommendations for repair, or providing any other expert engineering opinion associated with a foundation … would be considered the practice of engineering."
Our analysis. A home inspector flags a symptom, and that is a good job. Naming the cause and specifying the repair is engineering.
So when a free inspection hands you a diagnosis and a pier count on one visit, ask who signed it.
This is Texas law; every state regulates engineering, so check the wording in yours.
So a survey produced by a company that sells piers is not necessarily wrong — the readings are readings.
But the interpretation, the scope and the count sit downstream of an invoice, and there is no arrangement in which that is not true.
A survey you commissioned from someone who sells nothing has the same numbers and none of that pressure.
What your own log adds that a professional survey cannot
Time. An engineer arriving in March produces a superb snapshot of March. Your log has last March in it, and the March before. No professional can retroactively supply that, at any price.
Independence of the sales moment. Readings taken before anyone was trying to sell you anything are unusually hard to argue with, precisely because there was no argument in progress when you took them.
A test of the repair. Continue the same readings from the same marks after work is done. That is how you find out whether it worked, and it is the half of the file almost nobody collects. A warranty conversation two years out is a documentation conversation.
Method
How to keep a log that will still be useful in five years
Seven steps, none of which cost anything. The discipline is in the repeatability, not the instrument.
1
Choose a reference point and photograph it
Something unlikely to move, clearly identifiable, and reachable without moving furniture. Mark it. Photograph the mark in context, so a different person could find it in five years — including you, who will not remember.
2
Choose three or four measurement points and mark them too
Corners of the building, either side of a suspected problem, and one somewhere you believe is stable as a control. Small pencil crosses, photographed. Consistency beats coverage.
3
Record the method along with the number
Which instrument, which direction you faced, whether the floor covering was in place. A reading taken on bare subfloor and one taken on carpet are not comparable, and in three years you will have no way to tell which was which unless you wrote it down.
4
Note the date, the season and the recent weather
Especially in expansive clay regions. "Late August, after six weeks with no rain" is context that makes a reading interpretable. Without it, the number is just a number.
5
Read again at the same point in the next annual cycle
Same season, ideally similar conditions. This is what separates a trend from a season, and it is the single highest-value discipline in the whole method.
6
Keep the crack photographs alongside the numbers
Same dates, same file. The crack identification page explains what to photograph; elevations and crack widths moving together is a much stronger observation than either alone.
7
Store it somewhere it will survive a phone upgrade
Dates in filenames, not only in metadata — metadata does not survive being emailed or messaged. One folder, backed up. The file is worthless if you cannot produce it in the year you need it.
The questions
What to ask about anyone else’s elevation readings
When a survey is put in front of you, these five questions establish what kind of document it is. The last two are the sentences to be careful of.
"Where is the reference, and how do you know it has not moved?"
Every elevation figure is relative to something. If the reference is on the part of the structure under discussion, the whole map is telling you a story about itself.
"When was this read, and is there an earlier one?"
A single survey is a snapshot, however precise. Two surveys a year apart are a rate — and a rate is what determines urgency.
"Who paid for it, and does the same firm do the repair?"
Not an accusation. It is the single most useful piece of context for reading any recommendation attached to the numbers.
"What is the instrument and its stated accuracy?"
If the movement being claimed is smaller than the instrument’s resolution, the claim is noise. This is a fair, technical question with a short answer.
"Can I have the raw readings and the point locations?"
So a future reading can be taken from the same points. A survey that cannot be repeated is a one-off impression, however beautifully drawn — and repeatability is the whole value.
"Your floor is out of level, therefore your foundation is failing"
Out of level is a condition, not a diagnosis. Almost every older house is out of level somewhere. Whether it is changing is the question, and one visit cannot answer it.
"We can do the survey free as part of the estimate"
Often competent, and it may cost you nothing. But it arrives with a proposal, and the FTC advises getting it in writing and getting more than one bid. Treat the numbers as data and the conclusion as a sales proposal.
Vocabulary
The measurement words, defined
Benchmark
The fixed reference every reading is taken against. If it moves, every number in the log moves with it and nothing in the log will reveal that. Choose it carefully, photograph it, and never change it silently.
Relative elevation
Height compared with the benchmark rather than with sea level. It is the only kind of elevation that matters here, because buildings are damaged by internal differences, not by altitude.
Differential movement
Parts of a building moving by different amounts. Uniform movement damages almost nothing; the difference across the structure is what cracks it. This is the reason a survey covers the whole building rather than the worst room.
Deflection
Bending of a floor or beam under load, as distinct from the foundation beneath it moving. A springy floor can be a joist question rather than a foundation question, and the two get confused routinely.
Rate
Change divided by time — what this tool computes from the two readings you enter. Our arithmetic on your numbers, not a measurement we made, and only as good as the repeatability of the two readings.
Seasonal cycle
Annual rise and fall driven by soil moisture. In expansive clay regions it can exceed years of genuine settlement, which is why like-for-like seasonal comparison is the core discipline of a useful log.
Elevation survey
Measured relative floor heights across the whole building, against a fixed reference, drawn on a plan. Not the same as a lot survey, and not something a homeowner’s readings become.
Creep
Slow deformation of a material under sustained stress. ACI notes that restrained shrinkage stress "The tensile stress induced by restraining drying shrinkage is reduced with time due to creep or stress relaxation. Cracks develop only when the net tensile stress reaches the tensile strength of concrete."
It is why slow movement is tolerated and fast movement cracks things.
Frost heave
Seasonal ground movement from freezing, distinct from expansive clay. Exterior footings are to be placed not less than 12 inches below the undisturbed ground surface, and, where applicable, below the frost line specified for the locality.
Both conditions apply and the deeper one governs. Twelve inches is a floor, not a target: in a locality with a 42-inch frost line, 12 inches is nowhere near legal.
Control point
A measurement point you expect to be stable, kept deliberately so that a change everywhere at once can be distinguished from a change in your method. The cheapest quality check available.
Limits
What this tool cannot do
Specific blind spots. Two numbers and an interval is a narrow input, and it is worth being precise about how narrow.
It is not a survey and cannot become one. It performs arithmetic on two figures you typed. It has not seen your house, your instrument, your benchmark or your floor covering.
It cannot detect a moved benchmark, a different measurement point, or a change of method between readings. All three produce a confident, wrong rate, and the arithmetic looks identical in every case.
It cannot separate season from trend. Two readings six months apart in clay may be measuring rainfall. The chart above shows the shape of that trap; the fix is a third reading and a year, not a better calculator.
It has no threshold, and will not invent one. There is no honest universal number of inches per month above which a house needs piers. That judgment depends on structure, soil, span, materials and history, and it belongs to a licensed engineer looking at the building.
It cannot tell you which direction the movement means. Settlement, heave and lateral pressure all show up as elevation change. The repair method selector explains why that distinction decides everything, and it too takes the answer as an input rather than producing it.
It knows nothing about your soil or your climate. No lookup here reaches your address. The USDA NRCS Web Soil Survey gives general soil context free, and even that is context rather than a reason your particular floor has dropped.
HyreFoundation does not survey houses, does not inspect foundations, does not perform repairs, holds no dataset of jobs or bids, and takes no referral fee. That is exactly what allows this page to conclude that the right next step is to write a number in a notebook and read it again next August.
How this calculator works
Subtraction, then a sentence about the interval. No soil model. No pier count.
delta = reading 2 − reading 1 (inches)
if months is 0:
trend = enter months so the interval is part of the sentence
else if |delta| < 0.1 in:
trend = little change on these two numbers; keep logging
through a wet season and a dry one
else:
trend = a change; still not a cause;
show an engineer the log, not a pier company the panic
never = two phone-level readings are not an elevation survey
Cracks are evidence and elevations are data. A second survey months later shows whether movement is active or historic. This engine is that idea reduced to two points. It will not promote those two points into a survey, a cause, or a repair.
HyreFoundation does not survey houses, does not inspect them, and does not repair them.
What each input means
Inputs on this tool, in the order they appear on the form.
Input
What it is actually asking
Reading 1 (in)
First elevation at a marked location, inches relative to a benchmark you can find again. Sign is yours: pick a convention and keep it.
Reading 2 (in)
The later reading at the same location, same method, same benchmark. Mixing methods is how people invent movement.
Months
The interval. 0 leaves the trend sentence unfinished, on purpose. A week is usually weather. A season is the useful unit on expansive clay, in general, not as a rule about your lot.
Worked examples
Including one where the naive answer misleads, which is the example most calculators leave out.
Two close numbers, six months apart
Reading 1: 0.00 in. Reading 2: 0.04 in. Months: 6. Delta: +0.04 in over the interval you typed. Trend: little change on these two numbers. Keep logging through a wet season and a dry one.
Never: two phone-level readings are not an elevation survey. That is a notebook entry, not a pass, and not a reason to skip the engineer if the wall is horizontal or bowing.
A change, interval entered
Reading 1: 0.00 in. Reading 2: −0.45 in. Months: 8. Delta: −0.45 in over the interval. Trend: a change. This still is not a cause.
Show an engineer the log, not a pier company the panic. Direction on one point is not a grid. Heave versus settlement is still an elevation-survey question across the footprint, plus water.
The one where one reading misleads
A pier company "surveys" the house in an afternoon, shows a single contour, and quotes 12 piers because a corner is low. One reading is not a trend. Original construction can be out of level.
Historic movement that stopped can look the same as active movement until you come back. Type the same number in both fields and months 0: the trend sentence asks for the interval.
Type only one real reading and you do not have a log. Pay for a real grid if the decision is a purchase or a repair, and get the second grid from someone who does not sell the piers.
What changes the result
The two readings only change the delta. Months only change the trend sentence. A large delta with months at 0 is still "enter months." A tiny delta with months at 12 is still "little change on these two numbers," not "the house is fine."
The 0.1 inch threshold is an editorial gate on this form so two phone numbers that wiggle by a hair do not become a panic. It is not a code tolerance. It is not a structural opinion. ICC Chapter 4 is not a tolerance budget for your floor.
Nothing on the form asks where in the house the point is. One kitchen corner is not the footprint. That is why this is a log and not a survey.
Local considerations
Who is licensed to produce an elevation survey, whether a building department will look at one, and how expansive the soil is are local. NRCS Web Soil Survey is a general lookup for shrink-swell as a mapped soil property, not a reason your two numbers moved. Onsite investigation is needed for engineering applications.
We do not ship a city table of "allowable" floor deviation. Older houses in particular carry movement that stopped decades ago. The goal is a stable house, not a perfectly level one, wherever you are.
When not to use this
Do not use this as a survey, a structural opinion, or a reason to authorize piers. Two points are not a grid. A delta is not a cause.
Do not use a single listing measurement, or a single free-inspection contour, as a trend. One reading is not a trend. That sentence is the point of the tool.
Do not mix methods, locations, or units and then treat the subtraction as movement.
Do not skip an engineer on a bowing wall or a wide offset crack because this log looks quiet. Pattern is a different file. The crack identification tool is still not an opinion on that wall.
HyreFoundation does not survey houses. Matching is still being built. The inquiry form is not a dispatch line.
It is a set of floor heights measured across the whole building on a grid, drawn as a contour map.
It shows where the high and low areas are and whether the pattern looks like settlement, heave or original construction. A licensed surveyor or engineer produces it. This log is a two-point notebook, not that document.
Why do I need two elevation readings?
Because one number is not a trend. A floor can be out of level because it was built that way, because it moved long ago and stopped, or because it is moving now. A second reading, ideally after a wet and a dry season, separates those stories.
How do I take floor elevation readings myself?
Pick a benchmark you can find again, such as a nail or a mark on a pipe. Measure at the same spot, with the same method, and write the date. A phone app gives a rough relative number, not a survey. If a purchase or repair depends on it, pay for the real grid.
What units should I use?
Inches relative to your benchmark, because that is what the tool subtracts. If your notes are metric, convert them first. Never mix a first reading in one unit with a second in another.
The change is small. Is my foundation fine?
Not necessarily. A small change means little change on these two numbers, so keep logging through a wet and a dry season. Seasonal clay movement can hide in a short interval. A horizontal crack, bowing, or several doors out of square still need an engineer.
The change is large. Do I need piers?
Not from this number alone. A change is not a cause. Settlement, heave and original construction are corrected differently. Show the log to an engineer who does not sell repairs. See why foundations move and the repair method selector.
Can I use this instead of hiring a surveyor?
No. Two points are not a grid, and a grid is how you see a pattern across the whole footprint. This log tells you whether one point moved between two dates. It is a notebook, not a substitute for the survey.
Why did a blank field come out as zero?
Because empty number fields count as 0 on this form. If you have only one measurement, you do not have a trend yet. Leave the second field until you have a second date, then type both real numbers.
How do elevation readings relate to crack photos?
Cracks are evidence; elevations are data. Date both. The crack identification tool names a pattern, and this log names a change. Together they make a file an engineer can use.
Does HyreFoundation survey houses?
No. We do not survey, inspect or repair, and matching with companies is still being built. For what a paid inspection includes, see foundation inspection.
Sources and methodology
Figures dated 26 August 2026. Last reviewed .
Engineering Advisory Opinion Request 43 (February 2018) (Texas Board of Professional Engineers, retrieved 2026-09-05. Read directly as PDF. Diagnosing a foundation and recommending repair is engineering under Texas Engineering Practice Act §1001.003(c)(1). Texas law, cited as Texas law.)
Expansive Soil and Rock (Colorado Geological Survey, retrieved 2026-09-05. Swelling pressures up to 30,000 psf and 10–20% volume expansion on wetting — the mechanism behind the seasonal component of any elevation log kept in a clay region.)
International Residential Code, Chapter 4 Foundations (International Code Council, retrieved 2026-08-26. Model code for foundation construction, not a survey standard and not a local rule until adopted. This page does not certify a floor as in or out of tolerance.)
Web Soil Survey (USDA Natural Resources Conservation Service, retrieved 2026-08-26. General soil context, including shrink-swell as a mapped idea. Labeled general: a map unit is not a reason your two numbers changed. Onsite investigation is needed for engineering applications.)
How to avoid a home improvement scam (Federal Trade Commission, retrieved 2026-08-26. A panic number shown to a one-method company is how people buy a product. Get the diagnosis from someone who is not selling the repair.)
Building America program (U.S. Department of Energy, retrieved 2026-08-26. Cited for moisture and durability as the usual drivers of seasonal movement. Not a survey method.)