Speech Therapy Data Collection: Methods, Probes, and a Free Data Sheet
How to count trials, turn them into percent accuracy, tell probes from teaching trials, and keep data manageable in groups and at home, with a printable articulation data sheet.
Data collection is the part of therapy that happens while you are also modeling a sound, managing a game, and keeping three students on task. It is also how you decide what to do next session and how you show progress at report time. Done well, it takes a few seconds per attempt and a minute at the end. Done badly, it either eats the session or produces numbers nobody can interpret.
This guide covers the most common methods, how to calculate percent accuracy, the difference between probes and treatment trials, how to sample instead of scoring everything, how to take data in a mixed group, and what to do with data from home practice. There is a free articulation data sheet to print further down.
What your data needs to do
Therapy data has three jobs. It tells you what to change next session: move up a level, fade a cue, or try a different approach. It shows progress on each goal when the report is due. And it documents the service itself: who attended, for how long, and what they worked on.
ASHA’s practice portal on documentation in schools spells out all three. Treatment notes should give both a qualitative and a quantitative description of the student’s performance in each session, confirm the student was present, and record the minutes of service. Check marks and other informal session data help you compare performance from one session to the next, and your written interpretation of them becomes part of the record (ASHA’s example is a student completing cloze sentences with 70 percent accuracy with minimal cues). ASHA also advises keeping an ongoing record of your data, because parents, advocates, or a district audit may ask for it. It notes that IDEA requires regular progress reports to parents, and that the SLP should give both qualitative and quantitative information in them.
For SLPs who supervise assistants, a tutorial in Language, Speech, and Hearing Services in Schools (Schultz, Powell, and Ross, 2022) walks through the same process for SLP and SLPA teams: planning what data to collect, recording it, and reporting it to the supervising SLP. The authors conclude that competence with a wide variety of data collection methods, data types, and documentation tools helps assistants communicate results accurately, which in turn helps the SLP make well-informed clinical decisions. The same holds when you take the data yourself.
Speech therapy data collection methods
Most session data is some combination of the methods below. For articulation, the everyday combination is a trial-by-trial tally, the percentage it produces, the cueing level, and a line of context.
| Method | What you write down | Good for | Watch for |
|---|---|---|---|
| Trial-by-trial tally | A mark for every attempt: + for correct, 0 for incorrect | Articulation drills and any skill practiced in separate tries | Marking every attempt can slow a fast-paced session |
| Percent accuracy | Correct attempts divided by total attempts, times 100 | Goals written as accuracy, and comparing one session with the next | Small numbers of trials make the percentage jumpy |
| Cueing level | How much support a correct response needed, and what kind | Showing progress while accuracy holds steady, and planning to fade cues | Only useful if every level is defined and applied the same way |
| Opportunities in conversation | Correct productions out of the times the target came up in a sample | Carryover goals | The number of opportunities changes from sample to sample |
| Rating scale | A score on a short scale with written anchors | Skills that are not simply right or wrong on one try, like using a strategy | Subjective unless the anchors are written down |
| Anecdotal notes | A line on what helped, what got in the way, attention and behavior | Context that explains the numbers | Not a measure on its own |
Whatever you choose, match it to how the goal is written. A goal that says 80 percent accuracy needs a tally and a total. A goal about using a strategy in conversation needs opportunities or a rating. Our guide to writing measurable speech therapy IEP goals has examples of both.
How to collect data for speech therapy, step by step
- Start from the goal. Note the skill, the level it is practiced at, and the criterion. For articulation that means the sound, its position in the word, and the step on the articulation hierarchy, from syllables through conversation.
- Decide what counts as correct. For example, only the target sound is scored, not the rest of the word. Write the rule down so it is the same next week and for anyone else who takes data.
- Decide what you will sample. Every trial, a short probe, the final ten trials, or a mix. More on this below.
- Record as you go with simple codes. A plus or a zero per attempt, with a letter for the cue when one was needed.
- Total up before the student leaves, or right after. Numbers written from memory at the end of the day drift.
- Write the note. The number plus one sentence of context: what cue worked, what got in the way.
- Look back weekly. Three or four sessions side by side tell you whether to move up, hold, or change the approach. A single session rarely does.
Trial-by-trial tallies and percent accuracy
A tally is one mark per attempt at the target: a plus for correct and a minus or zero for incorrect. Many SLPs add a small letter for the cue a correct response needed, such as M for a model or V for a verbal reminder, so one row of marks carries both accuracy and support. Only count attempts at the level you are scoring. If the goal is S in sentences, the warm-up words are practice, not data.
Percent accuracy is the number correct divided by the total, times 100. The total matters as much as the percentage. With 5 attempts, every single attempt moves the result by 20 points, so 3 out of 5 and 4 out of 5 look like very different sessions when they may be the same skill on a different day. With 20 attempts, each one moves it by 5. Write the total next to every percentage so you and anyone reading the report can see how much a number can be trusted.
Illustration only, not real student data
J. is working on S at the start of words. Ten of the treatment trials looked like this:
That is 7 correct out of 10, or 70 percent, with two of the correct attempts needing a cue (one model, one verbal reminder). Across the whole session J. made 25 attempts and 18 were correct: 72 percent.
The total number of attempts is worth keeping for a second reason: it tells you how much practice happened. Speech sound research treats the amount of practice as an ingredient of therapy in its own right. Williams (2012) describes intensity in terms of dose, frequency, duration, and cumulative intervention intensity, and across three intervention studies found that greater intensity yielded greater treatment outcomes. A session with many attempts and a session with a handful are not the same session, even at the same percentage.
On mastery: many IEP goals use a criterion such as 80 percent accuracy across three consecutive sessions. That is a common convention, not a research-derived rule. Choose the criterion that fits the student and the goal, and apply it to the same kind of data every time, which for many SLPs means probes.
Probes versus treatment trials
This is the distinction that makes the biggest difference to what your data means.
- Treatment trials are the teaching part of the session. You model, cue, shape, and give feedback. Their accuracy tells you what the student can do with support.
- A probe is a short set of trials with no cueing, no models, no feedback, and no teaching on the target beforehand. Many SLPs run it at the very start of the session, before any practice, which is why it is often called a cold probe. It tells you what the student can do on their own.
The two numbers can disagree, and that is normal. A student may get most treatment trials right with a model before every attempt, then miss most of the probe at the start of the next session. Motor learning research draws the same line. Summarizing that research, Austermann Hula and colleagues (2008) note that immediate or frequent feedback tends to boost performance during practice but can interfere with retention and transfer. That does not make feedback a problem: a tutorial on R treatment lists frequent feedback among the principles for the earliest stages of establishing a sound (Preston and colleagues, 2020). It is a reason to keep the two kinds of data apart. A strong treatment-trial percentage shows the skill is within reach. A probe shows whether it stayed.
A few habits keep probes useful:
- Keep the probe list separate from the practice list, so the probe measures the skill and not memory of the same ten words.
- Use the same number of items each time, so percentages are comparable across weeks.
- Add untaught words now and then to check generalization, and a short conversation sample once words and sentences are solid. Our carryover tips cover how to structure that step.
- Pull probe words from a list sorted by position, such as the R word list, S word list, or TH word list, and keep a separate set for practice.
End-of-session sampling: you do not have to score everything
Scoring every attempt is the most precise option, but it slows the pace, and in articulation practice the number of attempts is part of the treatment. The alternative is to teach at full pace and score a planned sample: a probe at the start, the final ten trials of the session, or both.
How much precision you lose has been studied in discrete trial teaching, which, like articulation drill, is built from repeated trials. In a study of 11 children with autism, Lerman and colleagues (2011) reanalyzed therapists’ every-trial data several ways. For most children and skills, scoring only the first trial of each session gave a rough estimate of performance, but it often signaled mastery too early and missed early changes. Also recording the prompt level each response needed made it more sensitive. Scoring a larger subset of trials matched the every-trial data fairly well. The study comes from behavior analysis rather than speech therapy, but the practical lesson carries over: a planned sample of trials is usually enough, a single trial is not, and the level of support is worth writing down.
Whatever sample you use, name it in the note, for example “final 10 trials, no cues.” A percentage means little if a reader cannot tell which trials it came from.
Cueing levels and cue types
Accuracy sometimes holds steady for weeks while the support a student needs is quietly shrinking. Recording the cueing level captures that progress. Definitions vary between clinics, so write yours down. A common four-step scale looks like this:
- Independent: no help.
- Minimal: a brief reminder or gesture.
- Moderate: repeated cues, or a specific cue about placement.
- Maximal: a full model to imitate, often with more than one kind of cue.
The cue type is worth a letter too: verbal (a spoken reminder or instruction), visual (a picture, a mirror, a gesture), tactile (a touch cue), or a model the student imitates. Over a few weeks this shows which cues are doing the work, which is exactly what you need to plan fading them. ASHA’s school documentation guidance names the level of prompting a student needs as one of the details that treatment notes help document.
Collecting data in mixed groups
School caseloads often mean a group with three students working on three different goals. When you can, group students by target: a tutorial on R distortions lists heterogeneous groupings in group therapy among approaches that currently lack scientific support (Preston and colleagues, 2020). When the schedule does not allow it, these habits keep data manageable:
- One sheet for the group. A row per student with each target written at the top, so you are not flipping pages mid-game.
- A steady rotation. Short turns in the same order keep the number of attempts similar across students and keep everyone listening.
- Probes while others work. Run one student’s short probe while the others do something independent, such as a game move, coloring, or a quick written task.
- A focus student. Some SLPs rotate one student per session for detailed data, such as cue types and error patterns, and keep simple tallies for the rest.
- Attendance for everyone, every time. Mark who was absent and why, so make-ups and service minutes are easy to account for.
That last point matters beyond compliance. In a study of 126 kindergarten to second grade students with speech sound disorders receiving school-based services, the total number of therapy sessions a child received over the year predicted gains in speech sound accuracy (Farquharson, Tambyraja, and Justice, 2020). Missed sessions that are logged can be made up. Missed sessions that are not logged simply disappear.
Data from home practice
Home practice is a routine part of speech sound intervention. In a survey of 288 Australian SLPs working with speech sound disorders, 96.4 percent involved parents in intervention, most commonly through home practice, which was recommended on average five times a week for 10 minutes (Sugden and colleagues, 2018). The question is what to do with what comes back.
The useful information from home is simple: the level the student practiced at, how the attempts went, roughly how much practice happened, and, if you can get them, a few recordings to listen to. Treat it as practice data, not probe data. At home a student usually has support and feedback, from a parent or an app, which makes it closer to treatment trials. Keep it separate from the session probes you use to judge a goal, and report it alongside them.
The point of looking at home practice is planning, not checking up on families. It tells you which level to assign next, which words are sticking, and whether a student needs a different cue before the next session. It also helps to tell families why, not only what. In a small qualitative study of nine parents of young children with speech sound disorders, one of the main themes was that clear communication is key to effective home practice, and the authors conclude that parents need to understand not only what to do but why they are doing it (Pritchard and colleagues, 2026). Our guide to speech therapy homework that families follow through on covers how to set it up.
Free printable articulation data sheet
This sheet fits one session for one student on one printed page, with room for three targets. Tick probe or treatment for each target, mark each box with a plus or a zero (and a cue letter when one was needed), then total the row before the student leaves. For a group, print one per student, or give each student one of the three target blocks.
Articulation Data Sheet
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 |
Key: + correct, 0 incorrect. Add a letter for the cue a correct response needed, for example +M after a model or +V after a verbal cue. Probe: no cues, models or feedback, and no teaching on the target first. Treatment: teaching trials with cues and feedback.
From session data to the progress report
At reporting time, each goal needs the same few things: where the student started, where probe accuracy is now, how the level of support has changed, attendance for the period, and what home practice looked like. If your session data already records those, the report is mostly a matter of reading down the column and writing a sentence about what it means. Our speech therapy progress report template walks through each section. For the note you write after each session, see our speech therapy SOAP note examples.
Where LumaSpeech fits
LumaSpeech is a practice app for SLPs, built for the days between sessions. You assign practice, and students practice at home with feedback on their target sound, at word, sentence, and motor chaining levels, for any sound and position. Students who are ready for carryover can practice in a real conversation with Luma, where their target sounds are counted and misses are modeled back naturally rather than corrected mid-conversation, and you see a tally for each sound along with the transcript. Practice reminders help students keep going between sessions, and optional summary emails can keep parents in the loop.
What comes back to you saves time rather than adding it: recordings of each attempt you can re-listen to, progress pages for each student, and a weekly email summarizing your caseload.
For the sessions themselves, the session log does the job of the sheet above, and the data does not need retyping at report time. Each therapy session gets one entry. Add the students in a group, mark anyone absent with a reason so make-ups and service minutes are easy to account for, and for each goal record trials as correct out of total, the cueing level and cue type, and whether they were probe or treatment trials. That data appears on each goal and feeds the progress report, which you can export for any reporting period as a PDF or Word file, as an IEP progress report or a clinical progress note, with a section on home practice this period. You confirm when a goal is mastered, and you can edit the narrative before it goes out.
Frequently asked questions
What are the main data collection methods in speech therapy?
Common methods are trial-by-trial tallies (a mark for each attempt, correct or incorrect), percent accuracy calculated from those tallies, the cueing level a correct response needed, correct productions out of the opportunities in a conversation sample, rating scales with written anchors, and short anecdotal notes. Many SLPs combine a tally with the cueing level and a sentence of context, which together give the quantitative and qualitative description of performance that ASHA asks for in a treatment note.
How do you calculate percent accuracy in speech therapy?
Divide the number of correct attempts by the total number of attempts and multiply by 100. For example, 18 correct out of 25 attempts is 72 percent. Write down the total as well as the percentage, because with 5 attempts every single attempt moves the result by 20 points, while with 20 attempts each one moves it by 5.
What is the difference between a probe and treatment trials?
Treatment trials are the teaching part of the session, where you model, cue and give feedback, so their accuracy shows what the student can do with support. A probe is a short set of trials with no cueing, no models, no feedback and no teaching on the target beforehand, so it shows what the student can do on their own. Keep the two apart on the data sheet. Many SLPs rely on probe data when judging progress toward a goal.
How do you take data in a mixed speech therapy group?
Use one sheet for the whole group, with a row per student and each student's target written at the top. Take turns in a steady rotation so everyone gets a similar number of attempts, run each student's short probe while the others do something independent, and consider picking one focus student per session for detailed data such as cue types while you tally the rest. Mark attendance and minutes for every student at every session.
How often should you collect data in speech therapy?
Every session, in some form. ASHA's guidance for school-based SLPs says treatment notes should give a qualitative and quantitative description of the student's performance during each intervention session and record the minutes of service. That does not mean scoring every single attempt. A planned sample, such as a short probe or the final ten trials, is often enough.
Should home practice data count toward IEP goals?
Home practice usually happens with support and feedback, so it is closer to treatment trials than to a probe. Many SLPs keep it separate from the session data they use to judge a goal and report it alongside, for example as a home practice section in the progress report. It is most useful for planning: which level to assign next, which words are sticking, and whether a student needs a different cue.
Sources
- ASHA Practice Portal, Documentation in Schools: treatment notes, session data, and progress reports.
- Schultz, Powell, and Ross (2022), LSHSS: tutorial on data collection and documentation strategies for SLP and SLPA teams.
- Williams (2012), International Journal of Speech-Language Pathology: intensity in phonological intervention.
- Austermann Hula, Robin, Maas, Ballard, and Schmidt (2008), JSLHR: feedback frequency and timing in speech motor learning in adults, with a summary of motor-learning research on feedback.
- Lerman, Dittlinger, Fentress, and Lanagan (2011), Behavior Analysis in Practice: comparing methods of collecting data during discrete trial teaching.
- Preston, Benway, Leece, Hitchcock, and McAllister (2020), LSHSS: tutorial on motor-based treatment strategies for R distortions, including practice principles and approaches that lack support.
- Farquharson, Tambyraja, and Justice (2020), LSHSS: child and therapy factors in speech sound gains in school-based services.
- Sugden, Baker, Munro, Williams, and Trivette (2018), International Journal of Speech-Language Pathology: survey of parent involvement and home practice.
- Pritchard, Stojanovik, Titterington, and Pagnamenta (2026), IJLCD: a qualitative study of parents’ perspectives on supporting home practice.
Keep students practicing between sessions, and your data in one place
LumaSpeech gives students feedback on their target sounds at home, gives you a session log organized by goal, and brings both together in a progress report you can edit. Try it free for 21 days with up to 2 students, no credit card needed.
See how LumaSpeech works for SLPs →