TaxDome Automations: What to Automate First in a Small Practice

Automate three things first in TaxDome: onboarding (engagement, signed agreement and welcome), document requests tied to a checklist, and reminders for what is still outstanding. They are the same steps every client goes through, the wording rarely changes, and a mistake shows up quickly. Leave pricing, advice and anything sensitive to a person.

A small practice has no spare hours for rework, which is why the order matters. An automation that fires at the wrong moment costs more goodwill than the hours it saves, so each one below is introduced with a test, a number to watch and a point at which a person steps in. The setting names used here come from TaxDome's help centre in November 2026; if one has moved in your account, search the help centre for it by name.

Follow me on Instagram@sagnikteaches

Check what your plan includes before you build

TaxDome has changed how it sells the product. When checked on 4 November 2026, the pricing page showed no per-seat plans: one plan priced by active clients, starting at $1,500 a year for 150 active clients (a minimum that applies even to a firm with fewer) with 20 seats included, billed annually upfront, and extra clients from $15 each, cheaper in volume bands. An active client is one with any activity in the past 12 months, and the page says existing customers keep their current plan until their first renewal on or after 1 January. Ready-made automations, document request lists and automated reminders are listed in that plan, so nothing in this tutorial needs an extra purchase. Prices move, so read the live page and the tutorial on plans and what to ask on a demo before you decide what a pilot costs.

Connect on LinkedInSagnik Bhattacharya

The AI features are a separate question. TaxDome's help centre describes AI checklist generation and checklist matching as included in the subscription for eligible firms, with document renaming, tagging, AI help in tax return delivery and API and MCP access sold as an add-on: $200 a month per firm, every team member included, with a 14-day free trial once per firm. The add-on help page says that from 5 November 2026 those paid features need the add-on or an active trial wherever the add-on is sold, while checklist generation and matching stay free. None of the three automations below needs the add-on. Add it later if the document volume justifies it, and only after the manual process runs cleanly. If you want the wider picture of where AI sits in this software, the comparison of tax preparation AI software covers it.

Subscribe on YouTube@sagnikteaches

Map one client journey on paper first

Do this before opening the automation settings. Draw the journey of a typical returning client as a column of steps with an owner against each:

1. Engagement letter sent (owner: partner)
2. Letter signed and prepayment or deposit taken (owner: admin)
3. Organiser and document checklist sent (owner: admin)
4. Documents arrive; checklist ticked off (owner: admin)
5. Preparation (owner: preparer)
6. Review (owner: reviewer)
7. Return delivered for signature (owner: preparer)
8. Filing confirmation and invoice (owner: admin)

Mark each step with one of three labels: always the same, same except for details, or needs judgement. Steps one, two, three, four and eight are usually the first two kinds, so they are your candidates. Steps five and six are judgement, so they get task assignment and due dates but never an automatic outcome.

This one-page map is also your test script. Without it, you end up with an automation for every idea someone had in a meeting and no way to see the whole journey.

First: onboarding that does not depend on anyone remembering

The goal is that a signed engagement triggers everything else. In TaxDome this is a pipeline: each stage carries automations that fire when a job enters it (the help centre lists types such as Send proposal/EL, Send email, Create task, Create organizer and Send client request), and Automove jobs, switched on per stage, moves the job to the next stage once every item in the stage is done: the proposal signed, the task completed, the organiser submitted. Build one pipeline per service type and resist copying it for every client variation; conditional automations handle the variations.

What to include

• Welcome message. Short, with the next step, who the client's contact is and when to expect a reply. No long policy paragraphs.
• Engagement and consent. The engagement letter and any consent for AI or third-party tools, sent together so there is a single signing moment. If you use AI on client data, make sure the wording matches what you really do; the tutorial on consent for AI tools goes through what clients should sign.
• Internal tasks. A Create task automation that gives the admin a "check identity and prior-year return received" task, with a short due date (two days is sensible), so the later steps cannot start on a blank file.
• Hand-off trigger. With Automove on, the job leaves the onboarding stage only when the engagement letter is signed and the admin's task is marked complete, not before.

A filled-in example for a six-person practice (illustrative): signed engagement at day 0; welcome message at day 0; admin task due day 2 ("confirm prior-year return and ID received"); organiser sent automatically at day 3. In the first season the partner watches the first ten clients go through and writes down every point where a client wrote back confused.

Test before you trust it

Create a client record with your own email address and your colleague's, run the whole sequence, and read every message as the client sees it, on a phone. You will find at least two things to fix: usually a merge field that renders as a blank, and a message that appears twice because two triggers overlap.

Second: document requests that match the client

A generic request ("please upload your documents") produces generic confusion. The upgrade is a checklist per client, built from last year's return and this year's changes, with each line telling the client what to look for. TaxDome's help centre describes AI checklist generation from a prior-year return, for four named desktop packages (Drake, Lacerte, ProSeries and UltraTax), included in the subscription for eligible firms. If your firm does not use those packages or the feature is not available to you, do the same thing by hand with a template per client type: employee, employee plus rental, self-employed, small company owner.

Three rules make the checklist work:

1. Ask for the blocking items first. A client with ten lines and one that stops everything should see the blocking item at the top, marked as needed to start.
2. Say what the document looks like. "Annual summary from your payment provider, usually under Reports" gets answered faster than "processor statements".
3. Allow a phone photo. If you do not say a photo is fine, some clients will wait until they can scan.

Say a print shop owner has two checklists in the template library, one for a sole trader and one for a company owner. The sole-trader list has twelve lines. The preparer removes the two that do not apply (no rental income this year), adds a line for a van purchase and sends. Under five minutes, and the client sees something that looks like it was written for them. The tutorial on building an organiser that chases missing documents has the line-by-line wording.

What the AI features add, and what they do not

According to the help centre, checklist matching sorts each upload into the right checklist section, such as a wage statement into the wage statement section, so progress shows in real time. That helps the admin see a client's status at a glance. It does not tell you whether the document is the right year, the right person or readable. Someone still opens the first few uploads for each client, particularly in the first season. The add-on features for renaming and tagging tidy the file names ("IMG_30459.pdf" becomes a descriptive name), which is a convenience for the reviewer, not a quality control.

Third: reminders that read like a person wrote them

Reminders are the highest-return automation and the one most likely to annoy. In TaxDome they belong to the item that goes out rather than to a calendar: an organiser, client request, signature request, invoice or message sent with the Reminders toggle on repeats its email every Days until next reminder (default 3) up to the Number of reminders (default 1), and stops by itself when the client submits, signs or pays. That is the property you want, because a client who has finished is never chased. The help centre also says a reminder email looks like the original notification, flagged as a reminder, so the wording you control is the first message of each item. A pattern that suits a small practice:

Notice that the third row creates a task, not another email. Automations fire when a job enters a stage or a custom status, not after a set number of days, so the admin moves stragglers into the phone status after checking the list of pending organisers once a week. The point of automating reminders is to put human effort where it works: a phone call after three weeks recovers more files than a fourth email. Clients can also mark a checklist item they cannot provide yet with a reason and an expected date, and the help centre says they then get an email reminder on that date, so read those notes before you phone. Use template shortcodes for the client's name and nothing cleverer; the more a message is customised by an automation, the more places it can go wrong.

Write the reminder wording once, with AI as the drafter

This is the right place for a chatbot: drafting the three pieces of wording the sequence needs (the organiser's covering text, the client-request text for one blocking item and the message offering a call) in the firm's voice, which you then paste into the templates. Use a prompt like this on a general-purpose assistant, with no client data in it:

Draft three reminder emails for a small accounting practice to send
to clients who have not yet uploaded tax documents. Placeholders:
[first name], [outstanding items], [staff name]. Message 1 is warm
and lists the items. Message 2 names only the one blocking item and
says it is a quick upload. Message 3 offers a phone call instead of
a further email. Under 90 words each. No threats, no deadlines, no
penalties, no exclamation marks.

A sample draft (illustrative) of message 2: "Hi [first name], a quick nudge. We can start your return as soon as we have [outstanding items]. A phone photo uploaded to your portal is fine. If you can't find it, reply and [staff name] will help." What to fix: models love to add a closing flourish and an invented deadline. Strike both, and read the final template aloud before saving it.

A worked example: a five-person practice, one client journey

An illustration with invented numbers. A five-person practice has 140 clients, one administrator and two preparers. Before the change, the administrator sends welcome emails and organisers by hand and chases clients as they remember, spending about 9 hours a week at peak on admin around documents. Clients who respond late come in at the last minute.

The practice builds three automations in the first fortnight: engagement-to-organiser onboarding, the client-type checklist templates, and the five-row reminder pattern above. In the first season it measures one number each. Onboarding: days from signed engagement to organiser sent (from a hand-driven average of four days to the planned three). Checklists: lines clients marked as not applicable (starting at about 20%, brought under 10% by pruning templates). Reminders: files with a blocking item missing at day 21 (down from 31 to 12 across the 140 clients).

The quick sum: if the administrator's 9 hours a week fell to 5.5 for the six peak weeks, that is 21 hours back. The automations sit in the plan the practice already pays for, so the cost was setup time: about two days of a partner's time spread across a fortnight. The numbers are illustrative, but the method is the useful part: choose a metric before you build, measure for a season and then decide what to build next.

What to leave manual in the first season

• Fees and quotes. A price sent in error is a trust problem. Keep proposals human-reviewed.
• Anything with a deadline you have not verified. Automations repeat your mistake at scale. Put real dates in only after a partner has confirmed them.
• Replies to confused or upset clients. Route these to a person with a task, not a template.
• Review and sign-off. Assign and date these tasks; never auto-complete them.
• AI on client documents without a data agreement. Read the vendor's security and data-use page first; the help centre lists the AI providers TaxDome relies on and says it sends only the data needed, but your own client consent should match what happens.

Reporting: what to look at and when

What sits in each plan has changed, so do not rely on older descriptions of which plan carries which reports. The current single plan's feature list on the pricing page includes advanced performance reporting, profitability analysis and a business analytics dashboard, but firms still on an older plan keep it until renewal, so confirm in your own account which reports you can see. Whatever your view, three numbers from the task and pipeline data are worth reading weekly in season:

1. Tasks overdue by owner. Tells you where the bottleneck sits.
2. Clients by stage. A count of how many are waiting on documents, in preparation, in review and awaiting signature.
3. Median days in the document stage. The one automation is meant to move.

If your reporting is limited, a weekly screenshot of the pipeline by stage and a one-line note ("12 waiting on documents, 4 over 14 days") is enough for a small firm.

Two mistakes to expect in the first season

The reminder that went to a client who had already sent everything

A realistic case, described generally. A yoga studio owner uploaded her documents by email to a partner instead of the portal. The checklist still showed 20% complete, so the day-7 reminder arrived asking for items the practice already held. She replied, politely, that she had sent them a week ago. Nothing was lost, but the firm looked disorganised. The fix was a rule rather than a setting: anything arriving by email is uploaded to the portal the same day by the admin, and the checklist is ticked, so the status the reminders read is true. Audit this weekly in season by comparing the inbox against the checklist for any client who was sent a reminder.

The onboarding that waited for a person who never saw the task

Another typical failure: the "confirm prior-year return received" task was assigned to a former team member, so it sat unseen and the organiser never went out. Three clients waited eleven days. The fix was to assign onboarding tasks by account role in the task template (the help centre says role-based assignees follow whoever holds that role for the client), and to run an "overdue tasks by owner" check every Monday that includes people who have left.

Before and after: the same client, one message

To see what the automation is replacing, compare two versions of the first document request for a tutoring centre owner (illustrative).

Before (typed by hand on a busy afternoon): "Hi, please send everything for your return. Thanks." The client replies with last year's documents and a question about whether a laptop counts.

After (from the sole-trader checklist, with the lines pruned): "Hi [first name], here is your checklist for this year's return: card processor annual summary, hall rent statements for the year, tutor payment records, and the receipt for the laptop you bought in March. The first one is the one we need to start; a phone photo is fine for all of them. Questions? Reply here and [staff name] will answer."

The second takes about four minutes to produce from a template and removes at least one round of replies. Multiply that across 140 clients and you can see why this is the automation that earns its place first.

A small decision rule for every new automation

Before you build the next one, answer four questions in a line each: what triggers it, what does the client see, who is told when it fires, and what number will tell us it worked. If you cannot answer the last two, do not build it yet. This keeps the list short, which is the real defence against a firm that is automated but not any faster.

How to keep automations from rotting

Automations age badly: a template written for last year refers to a form that has changed, a team member leaves and tasks are still assigned to them, or a partner's wording changes and the templates do not. Put a 30-minute review in the calendar twice a year, once before the season and once after, and walk the journey as a test client. Check the assignee on every task, the dates in every message, and the checklist templates against last year's most common clients. Keep a short change log (a note in the practice's shared folder will do) with the date and reason for each change, so a new staff member can see why a reminder says what it says.

If you are still comparing platforms, Canopy against TaxDome for tax-heavy practices and TaxDome against Karbon cover the choice of platform.

Questions about TaxDome automations in a small firm

Do I need the AI add-on to automate onboarding and reminders?

No. Onboarding, document requests and reminders are workflow features, not AI features. TaxDome's help centre lists AI checklist generation and matching as included in the subscription for eligible firms, and AI renaming, tagging, return-delivery help and API and MCP access in a separate add-on. Check the live page for your region before buying.

How many automations should a five-person firm start with?

Three or four, covering one client journey from engagement to filed return. Each one needs an owner and a test client. Firms that switch on fifteen at once cannot tell which one sent the confusing email.

Should clients ever receive an automatic message with no human review?

For routine items such as a welcome message or a missing-document nudge, yes, once you have tested the wording on yourself. For anything involving advice, fees, deadlines you have not confirmed or a client in difficulty, route the task to a person first.

How do I know an automation is working?

Pick one number per automation before you start: days from signed engagement to documents received, tasks completed late, or reminders sent per client. Compare it for two weeks before and four weeks after, and read ten of the actual messages clients received.

Further reads

• AI for Small CPA Firms: Chasing Missing Client Documents in Tax Season — A deeper reminder schedule for missing documents.
• Tax Services AI for a Small Practice: Intake, Chasing and Review Notes — Intake, chasing and reviewer notes with AI.
• How Small Tax Practices Use AI Through the Busy Season — How the pieces fit across a whole season.
• AI Software for Tax Professionals: Run a Pilot and Measure Cost per Return — Run a pilot and measure cost per return.
• AI Tax Return Review vs Software Diagnostics: What Each One Catches — What AI review catches versus software checks.
• AI Tax Return Prep for Small Practices: Extract, Review, Sign Off — Extract, review and sign off a return.
• Accounting Software for CPA Firms: Automation Features Compared — How automation features line up across practice software.
• Canopy Pricing: Power-Ups, Client Credits and Add-On Costs — Canopy's tiers, tax workflow credits, Close Automation and payment fees priced out for four firm shapes, with the questions to ask before you sign.
• Karbon Pricing: Team vs Business Plans and Where the AI Sits — Karbon's Team and Business plans priced for 3, 6 and 10 users, with where Kai and the AI agents sit and which add-on fees to check.
• AI for Doing Tax Accounting: What to Hand Over and What to Review — A small-practice task split for document extraction, tax workpapers and client letters, with a named check before each output goes out.
• What Is Automation in Accounting? Bank Feeds, Capture and AI Agents — What accounting automation includes, which parts are plain rules and which need AI, with a month-end example and the checks that stay manual.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: TaxDome pricing page; TaxDome Help Center articles on TaxDome AI, the AI add-on, automations, jobs automove, reminders, document checklists and task templates (all read 4 November 2026). Setting names change between releases; confirm against your own account. Examples are illustrative.