Skip to page content

Free form and downloads for house managers, program supervisors and QIDPs

Behavior support plan implementation checklist for DSPs

A free observation form, in Word and PDF, for watching a direct support professional carry out a person’s existing behavior support plan: mark each step as written, give feedback the same shift, and track it over time. The plan’s own author supplies the steps; the form gives them a place to be marked, counted and talked about, so the supports a person agreed to are really there.

Special Learning · November 16, 2026

Free download

Download the checklist

The Word file has the observation form (who, when and what routine, then steps before, steps for teaching, steps for response and data recording, each with blank rows), the percent as written, a box for when a step cannot be done as written, a same-shift feedback page, a tracking grid, a how-to page and a checklist. It is a standard .docx file; we opened it in LibreOffice to check it. The PDF has the same pages. No sign-up, and nothing you type is sent to us.

The first page of the fillable Word checklist and its same-shift feedback page. Page 1: boxes for the date, setting, observer, DSP, person supported, plan and its author, and the routine observed, then a table with a column for the step as written in the plan, Y, N and NA marks, and notes. Page 4: boxes for what went well, one step to practice, whether it was practiced on the spot, and the next observation.
The observation form and the feedback page. Open the full-size image.

Short answer

What is a behavior support plan implementation checklist?

Read

A behavior support plan implementation checklist is a form an observer uses to watch a direct support professional (DSP) carry out a person’s existing plan: the plan’s own steps are listed in its author’s words, each is marked done as written, not done as written or not applicable, and the percent done as written is worked out and shared with the DSP the same shift. The form never writes the plan. The BCBA, psychologist or behavior specialist who wrote it supplies the steps, and calls to that author settle any step that cannot be done as written. How often to observe depends on your state, funder and agency, so check yours; this page gives you the form, a tracking grid across observations and a way to give feedback that helps.

We say plainly that rules vary and do not give one frequency for every program. This page is practice guidance built on the sources below; it does not replace the plan’s author, your licensing rule or your funder’s contract.

Why it matters

Why check that a plan is carried out as written

Four cards on why to check that a plan is carried out as written: a plan only helps if it is carried out, steps drift without anyone noticing, training is rarely enough on its own, and staff are meant to implement each person's plan. The same text is in the list above.
Why to check. The federal rule shown is for ICF/IID, and the research is mostly from schools.

What the sources say

Rules vary by state and funder: examples

These are examples of how different the sources are. None of them sets one frequency for observing DSPs, and we found no national rule that does. Use your own licensing rule, your funder’s contract and the plan author’s instructions; our state finder links each state’s DSP training rule and agency. Find your state’s DSP training rule and agency in our free state finder, then read your own licensing rule. The federal ICF/IID rules are all in 42 CFR Part 483, Subpart I, on eCFR.

Six cards on what six sources say: three federal ICF/IID rules, a Minnesota example, an Ohio example and the NADSP Code of Ethics. The same text is in the list above.
Six sources. Use your own rule.

The form

What is on the form

The nine parts of the form. The layout is ours; the steps are the plan author’s.
PartWhat it holdsWhy
1. Who, when and what routineObservation date, time, setting, the observer, the DSP observed, the person supported (initials or your own ID), the plan and its date, the plan author and their phone, and the routine or situation observed.Every mark means something only for one routine on one day, and the plan author is named before anything goes wrong (our practice).
2. Steps before: proactive and antecedentBlank rows for each step the plan lists to do before or at the start of the routine, in the plan’s words, marked Yes, No or N/A with a note.Vollmer and colleagues note that looking at individual components may be as important as the overall figure and can allow more focused feedback. (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice)
3. Steps for teaching: replacement skillsBlank rows for each teaching or replacement-skill step in the plan, in the plan’s words, with the same marks.Teaching steps are where a plan changes what a person can do; they get their own section so they are not lost.
4. Steps for responseBlank rows for each step the plan gives for how to respond, in the plan’s words, with the same marks. If the situation did not come up, mark N/A.N/A keeps the percent honest: a step you did not get to see is not counted for or against anyone (our practice).
5. Data recordingBlank rows for each data step the plan asks for, such as a tally or a note at the end of the routine.The plan’s data tell its author whether it is working; a missed data step is worth a kind, specific word (our practice).
6. Percent as writtenA line for each section and one overall: steps marked Yes, divided by Yes plus No, times 100. N/A steps are left out.Vollmer and colleagues note that integrity is typically calculated as the “total percentage of opportunities for which the procedure was implemented correctly.” (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice)
7. Same-shift feedbackWhat went well, with detail; one step to practice, rehearsed on the spot; the date of the next observation; initials.Feedback has two forms, positive and corrective, and the research behind it says to praise correct steps and correct errors explicitly. (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice; Fiske (2008), Behavior Analysis in Practice)
8. When a step cannot be done as writtenA box with the plan author’s name and number, and a line for what you did: mark it, call the author, tell the supervisor.A change to a plan is the plan author’s to make, with the person and their team (our practice). The NADSP Code asks DSPs to “seek guidance from supervisors, managers, or policies when unsure of the correct course of action.” (NADSP Code of Ethics)
9. Tracking gridOne row per observation across weeks, with date, observer, DSP initials, the percent for each section and overall, whether feedback was given the same shift, and the next date.Fiske’s review recommends data review that includes graphs of staff performance; a grid is the plainest version. (Fiske (2008), Behavior Analysis in Practice)
A table of the nine parts of the form with what each holds: who, when and what routine; steps before; steps for teaching; steps for response; data recording; percent as written; same-shift feedback; when a step cannot be done as written; and the tracking grid. The same text is in the table above.
What is on the form.

The form’s layout, the marks and the tracking grid are ours. The steps are never ours: they come from the person’s plan, in its author’s words. For a person’s wider picture, see the one-page profile template; for the data a plan asks for, the ABC data sheet; and for a plan to follow in a crisis, the crisis plan template.

Seven steps

How to run an observation

  1. Before you start, get the current plan from the plan author, the person’s record or the place your agency keeps it, and copy its steps for the routine into the blank rows in the plan’s own words. Do not shorten, reword or add steps. If a step is unclear, stop and ask the plan author before you observe.
  2. Tell the DSP what you will do and why: you are checking that the plan is clear and works for the team, not catching anyone. Where you can, tell the person supported and ask what they would like you to know; write what they say. Follow your agency’s policy on how people are told about observations.
  3. Choose a routine where the plan’s steps will come up. Vollmer and colleagues suggest scheduling observations “during periods when the behavior is most likely to occur” to get more chances to see the steps.
  4. Watch quietly and mark each step as it happens: Yes if it was done as written, No if it was not, N/A if the situation did not come up. Add a short note for every No and N/A. Do not coach in the middle unless someone’s safety needs you to step in; if it does, follow your emergency procedures first.
  5. Add up each section and the overall percent. Count Yes divided by Yes plus No, times 100, and leave out N/A.
  6. Give feedback the same shift, in private, after the routine (see below), fill in the feedback box and put the next observation on the calendar before you leave.
  7. Enter the result on the tracking grid. If the same step is a No across several observations, ask the plan author whether the step is clear, possible in this setting and still right for the person.
A filled example, with invented people: seven steps from an invented plan for a morning shower routine, each with its section, the step in the plan's words, a Yes, No or N/A mark and a note. Four are Yes, two are No and one is N/A, so the result is 4 divided by 6, or 67 percent.
An invented example: what “as written” means and how the percent is worked out.
An invented example: one observation of the morning routine. Result: Yes 4, No 2, N/A 1: 4 ÷ (4 + 2) × 100 = 67%.
SectionPartStep as written in the plan (invented)MarkNote
ABeforeShow Lee the visual schedule and say what comes next.YesLee looked at the schedule.
ABeforeOffer the two-choice board for the towel color.Yes
ABeforeGive a two-minute warning with the timer.NoTimer was in the office. Dana used her voice. Call author?
BTeachPrompt the break card when Lee touches his ears.Yes
BTeachPraise within five seconds of using the card.YesSaid “nice job asking for a break.”
CRespondIf Lee says “no” to the shower, wait one minute, then offer the choice again.N/ADid not come up.
DDataMark the tally on the data sheet after the routine.NoFilled in at 9:30 from memory.

Lee, Dana and Priya are invented, and the steps are an invented illustration of what a plan’s author might write, not a clinical suggestion. Your steps come from your plan. Count the steps marked Yes and divide by Yes plus No, then multiply by 100: Yes 4, No 2, N/A 1: 4 ÷ (4 + 2) × 100 = 67%. Vollmer and colleagues describe integrity as typically calculated as “the total percentage of opportunities for which the procedure was implemented correctly,” and add that looking at each component can matter as much as the total, so the form gives a percent for each section too. (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice)

Same shift

How to give feedback the same shift

Parsons, Rollyson and Reid describe behavioral skills training (BST) as “instructions, modeling, practice, and feedback until mastery is achieved.” The same four moves can be used on the job in a few minutes after an observation. (Parsons, Rollyson and Reid (2012), Behavior Analysis in Practice)

Four cards for same-shift feedback: instruct, model, rehearse and feedback, each with one line on what to do. The same text is in the list below.
The same-shift feedback loop.

How to say it

The same example: the feedback box, filled in (invented).
BoxWhat was written
What went wellYou showed the schedule and offered the choice board, and Lee looked at the schedule. You praised the break card right away.
One step to practiceThe two-minute warning. We practiced it with the timer on the hook by the door. I will ask the plan author whether voice is OK when the timer is missing.
Next observationMarch 15, morning routine.

How often

How often to observe

How often to observe: what the sources say, with no frequency set by this page.
WhenWhat the sources say
Right after training or a plan changeObserve often at first. Fiske’s review: “Initially, frequent observations may be required following teacher training,” and “initially, daily feedback may be necessary to attain high levels of treatment integrity.” (Fiske (2008), Behavior Analysis in Practice)
When the percent is high and steadyFade. The review adds that observations “can be faded if treatment integrity remains high and stable,” and that fading “to weekly and biweekly feedback may result in greater maintenance.” (Fiske (2008), Behavior Analysis in Practice)
Keep feedback goingThe review says “performance feedback is necessary to maintain high treatment integrity levels over time.” Observation does not end when the percent reaches your target. (Fiske (2008), Behavior Analysis in Practice)
Watch for drift, and add a refresherIf the percent slips, go back to more frequent observation and a short refresher. Vollmer and colleagues suggest “periodic booster training sessions.” (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice)
Check your own ruleYour state, funder or agency may set a minimum, or may say “periodic.” Minnesota’s example asks for “periodic performance evaluations” based on direct observation. This page does not set one. (Minnesota Statutes 245D.09, subdivisions 3 and 4a (example))
Five cards on how often to observe: right after training or a plan change, when the percent is high and steady, keep feedback going, watch for drift and add a refresher, and check your own rule. The same text is in the table above.
How often to observe. No frequency is set by this page.

The ranges above come from school-based research and are not a rule for any home or program. Vollmer and colleagues also say monitoring should be simple, “using efficient methods such as intermittent sampling rather than continuous monitoring.” (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice) For where observation fits with supervision more widely, see the DSP supervision log template and the DSP competency checklist.

When a step cannot be done

When a step cannot be done as written, call the plan’s author

Five cards for when a step cannot be done as written: mark it honestly, do not improvise a new step, call the plan's author, tell your supervisor the same day, and if anyone is in danger follow your emergency procedures first. The same text is in the list above.
When a step cannot be done as written.

The person at the center

The plan belongs to the person it supports

If a check turns up a problem that involves someone’s safety, it may also belong in your incident report.

Over time

Track it across observations

One line per observation, with the percent for each section and overall. Fiske’s review recommends data review that includes “graphs of staff performance”; a grid is the plainest version, and the Word file has room for sixteen lines. (Fiske (2008), Behavior Analysis in Practice)

A tracking grid with invented initials and numbers.
DateObserverDSPRoutineABCDOverallFeedback same shiftNext
Mar 12PLDKMorning67100N/A067YesMar 15
Mar 15PLDKMorning100100N/A100100YesMar 22
Mar 22PLDKMorning100100100100100YesApr 5
Mar 13PLCREvening3350010038YesMar 16
Mar 16PLCREvening671005010075YesMar 23
Mar 23AOCREvening100100100100100YesApr 6
A tracking grid with six invented observations of two DSPs by initials, with date, observer, routine, percent for sections A to D, overall percent, whether feedback was given the same shift and the next observation date. Overall percent rises from 67 and 38 to 100 for both.
A tracking grid with invented initials and numbers.

DK, CR, PL and AO are invented initials and the numbers are invented. Percent is Yes divided by Yes plus No, times 100, with N/A left out; “N/A” in a column means every step in that section was N/A. Your sheet, your rule and the plan’s author decide the target. If the percent slips, go back to observing more often and add a short refresher; if a step is missed across several observations, ask the plan’s author whether it is clear, possible in this setting and still right for the person.

The checklist

Around each observation

A checklist of nine items around each observation, from having the current plan and its author's number on the form to putting the result on the tracking grid and the next observation on the calendar. The same text is in the list below.
The checklist.

For the staff training roster across the year, see the DSP training and HR toolkit.

On paper

Prefer paper? Print the PDF

The printable PDF has the same seven pages as the Word file, with write-in space for the steps, feedback and tracking grid.

Thumbnails of the seven pages of the printable PDF, portrait US Letter: the first steps table, steps for teaching and response, data recording with the overall percent and the box for when a step cannot be done, same-shift feedback, the tracking grid, how to run an observation and how often, and the sources and next step.
The printable PDF. Download it.

Questions

Behavior support plan implementation checklists: common questions

What is a behavior support plan fidelity checklist?

It is a form for watching how a direct support professional carries out a person’s existing behavior support plan. You list the plan’s steps in its author’s words, mark each one done as written, not done as written or not applicable, work out the percent done as written, and share it with the DSP the same shift. Researchers call this treatment integrity: “the degree to which the treatment is implemented as planned.” (Fiske (2008), Behavior Analysis in Practice)

How often should I observe a DSP carrying out a plan?

It depends on your state, funder and agency, so we do not give one number. A school-based research review says frequent observations may be needed after training and can be faded “if treatment integrity remains high and stable.” (Fiske (2008), Behavior Analysis in Practice) Minnesota’s example, for HCBS license holders, asks for “periodic performance evaluations” based on direct observation. (Minnesota Statutes 245D.09, subdivisions 3 and 4a (example)) Ask your licensing agency and the plan’s author.

How do I work out the percent of steps done as written?

Count the steps marked Yes and divide by the steps marked Yes plus No, then multiply by 100; leave N/A out. Vollmer and colleagues describe integrity as typically calculated from “the total percentage of opportunities for which the procedure was implemented correctly,” and add that looking at individual components can matter as much as the total. (Vollmer, Sloman and St. Peter Pipkin (2008), Behavior Analysis in Practice) The form works out a percent for each section and overall.

What if a step in the plan is unclear or cannot be done?

Mark it No or N/A with a note, do not make up a replacement, and call the plan’s author the same day. A change to a plan is the author’s to make, with the person and their team. The NADSP Code of Ethics asks DSPs to seek guidance when unsure of the correct course of action. (NADSP Code of Ethics)

How do I give feedback the same shift?

After the routine, in private: say what went well, be specific, pick one step to practice and rehearse it right then, and say when you will observe next. Behavioral skills training is “instructions, modeling, practice, and feedback until mastery is achieved.” (Parsons, Rollyson and Reid (2012), Behavior Analysis in Practice) Supportive feedback describes exactly what was done correctly; corrective feedback specifies what was not. (Parsons, Rollyson and Reid (2012), Behavior Analysis in Practice)

Does federal law require me to observe DSPs?

For ICF/IIDs, the federal rule says staff must be able to demonstrate the skills to implement each client’s individual program plan, and that the facility must give initial and continuing training. (42 CFR 483.430(e)) We did not find a federal rule that sets how often to observe. Home and community-based services rules differ by state and funder; Minnesota’s example is one. (Minnesota Statutes 245D.09, subdivisions 3 and 4a (example)) Check yours.

Who writes the steps on the form?

The person who wrote the plan: a BCBA, psychologist or behavior specialist. You copy the steps into the form in their words. The form never adds, changes or ranks steps.

Should the person know they are being observed?

We did not find a rule that answers this for every setting. Our practice is to tell the person in a way they understand, ask what they would like you to know, and follow your agency’s policy. The NADSP Code of Ethics asks DSPs to apply person-centered practices. (NADSP Code of Ethics)

Can I use this for a plan that has restrictive procedures?

A fidelity check does not authorize any procedure. Restrictive procedures belong in a plan only as the person’s plan and your state allow, and the people who carry them out may need specific training first; Ohio’s example requires training in the strategy “prior to serving the individual.” (Ohio Administrative Code 5123-2-06(F) (example)) Use the form to check that the plan is carried out as written, and take any concern to the plan’s author.

Is this a clinical or legal review?

No. It is a form and practice guidance built on the sources listed. The plan’s author, your licensing agency and your funder decide what your program must do.

Sources

Sources read November 16, 2026 (the eCFR text is current as of November 12). The quoted lines are as published; the form layout, marks, percent convention, tracking grid, feedback box and checklist are ours, and the steps in any plan are its author’s. Fiske’s review is of school-based research; Minnesota and Ohio are examples only and are not a rule for any other state. None of these organizations has reviewed or endorsed this form.

  1. Electronic Code of Federal Regulations, Title 42, section 483.430, Condition of participation: Facility staffing, (e) Standard: Staff training program, ICF/IID; text current as of November 12, 2026. https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-C/part-483/subpart-I/section-483.430
  2. Electronic Code of Federal Regulations, Title 42, section 483.440, Condition of participation: Active treatment services, (c) Individual program plan and (d) Program implementation, ICF/IID; text current as of November 12, 2026. https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-C/part-483/subpart-I/section-483.440
  3. Electronic Code of Federal Regulations, Title 42, section 483.450, Condition of participation: Client behavior and facility practices, (b) Standard: Management of inappropriate client behavior, ICF/IID; text current as of November 12, 2026. https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-C/part-483/subpart-I/section-483.450
  4. Fiske, K. E. (2008). Treatment integrity of school-based behavior analytic interventions: A review of the research. Behavior Analysis in Practice, 1(2), 19-25; a review of school-based research, so its examples are teachers and classrooms. https://pmc.ncbi.nlm.nih.gov/articles/PMC2846589/
  5. Vollmer, T. R., Sloman, K. N., & St. Peter Pipkin, C. (2008). Practical implications of data reliability and treatment integrity monitoring. Behavior Analysis in Practice, 1(2), 4-11. https://pmc.ncbi.nlm.nih.gov/articles/PMC2846587/
  6. Parsons, M. B., Rollyson, J. H., & Reid, D. H. (2012). Evidence-based staff training: A guide for practitioners. Behavior Analysis in Practice, 5(2), 2-11. https://pmc.ncbi.nlm.nih.gov/articles/PMC3592486/
  7. National Alliance for Direct Support Professionals (NADSP), The NADSP Code of Ethics (revised 2026), as published on nadsp.org. https://nadsp.org/resources/the-nadsp-code-of-ethics/
  8. Minnesota Statutes, section 245D.09, Staffing standards, subdivision 3 (staff qualifications) and subdivision 4a (orientation to individual service recipient needs): home and community-based services license holders for people with disabilities; one state’s example only. https://www.revisor.mn.gov/statutes/cite/245D.09
  9. Ohio Administrative Code, rule 5123-2-06, Development and implementation of behavioral support strategies, (F) Implementation of behavioral support strategies with restrictive measures, as published by the State of Ohio; one state’s example only. https://codes.ohio.gov/ohio-administrative-code/rule-5123-2-06