Trending Articles
Trending Topics
Trending
As President and CEO of Accountalent, J.R. helps VC-financed startups across the United States to navigate modern tax compliance while continuing to scale rapidly. Thanks to years of experience in audit and financial controls, he offers a perspective deeply rooted in practicality, one that recognises the tension between automation, regulation and the ever-important demand for accuracy.
For J.R., despite the recent advances in automation, tax remains one of the most stubbornly manual areas of finance. “Tax compliance is much more manual than most founders think,” he explains, particularly when it comes to “classifying expenses, proving R&D work, and reconciling state level filing”.
This tension exists largely due to the fact that tax regulations “encourage accuracy and record keeping rather than fastness”. Even one misclassification may cost an individual thousands, meaning that manual review acts as more of a defence than a technical requirement.
Despite this manual burden, J.R. sees tax data as one of the most underutilised sources when it comes to strategic insight. “Data connected with tax compliance silently contains operational decisions long before the deadline to file,” he explains, potentially offering an early insight into financial performance. This data allows switched-on businesses to act with a critical head start to their decision making.
In an environment where acting on insight can make a significant financial difference, the role of automation is also becoming increasingly important. Yet, as J.R.’s own experience bears out, there are many critical processes that still remain stubbornly manual.
It’s these issues that we decided to ask about first, to find out exactly which processes are causing problems – and what it is that’s getting in the way of automation.
Tax compliance is much more manual than most founders think, particularly with respect to classifying expenses, proving R&D work, and reconciling state-level filing. Practically, big data is still transferred across spreadsheets, emails and documents in general. That tension continues due to the fact that the tax regulations encourage accuracy and record-keeping rather than speed.
In jurisdiction rules, filing position rules, and rules based on credit eligibility, automation has a hard time, since it brings about some trepidation on the part of the teams that worry about rework or punishment. Even one misclassification may cost a person 10,000 dollars, and thus, individuals intentionally take their time. As it happens, manual review is a defensive practice rather than a technical requirement. Actually, according to a report prepared in-house, close to 45 per cent of the time spent on early-stage tax work is devoted to verification, not to calculation.
Automation fails as the tools seldom reflect the actual audit activity of the tax authorities on filings. Compliance is in grey areas, and systems like black and white are welcomed. That mismatch keeps human beings in a prolonged loop than anticipated. Anyway, adoption is slower where automation is unable to articulate its logic to be able to stand in the event of an audit. Even in the case of a technology that appears prepared, trust is built gradually in tax.
Data connected with tax compliance silently contains operational decisions long before the deadline to file. Trends shown by expense classification, payroll timing and patterns of R&D expense show that burn occurs long before cash problems start to exist. Indeed, the consolidated tax data usually uncovers trade inefficiencies several months before the financial statements are presented to the board.
Analytics can be used to identify whether the rate of hiring by a startup is in tandem with credit eligibility requirements or even exceeds safe filing assumptions. That is important since missing a credit window can be a missed opportunity of leaving $250,000 on the table.
Incidentally, founders take quicker action to dollar influence as compared to abstract predictions. Tax information gives ground for that. In the vast majority of cases, expansion sequencing is directed by analytics regarding amended returns or state exposure. Compliance overhead can be reduced by 18 per cent per year with the opening in three states instead of five. That wisdom determines the rate of growth, though no one may tag it as strategy.
Tax information, in a sense, is a stress test for the decisions of operation. The same applies to the timing of R&D, which moving spend a quarter can have significant changes in cash recovery. Analytics transform tax as a retrospective activity into a planning input, although such teams seldom express it as such.
In the tax process, AI has already cut down document collection, data normalization, and screening of eligibility. That effect shows up quietly. Activities that used to be 3 hours are now done in approximately 40 minutes. Pattern recognition is the realistic value, and not judgment. AI raises red flags in the categories of expenses, potential lack of substantiation and establishes an estimate of the risk of an audit through a past history of filing behaviour. And that risk scoring will reduce review cycles by almost 30 per cent.
Value in the short run is inconclusive but realistic. AI is most effective when the rules recur and the documentation is in a familiar format. The advantages of filling out R and D credit are that the narrative and cost categories have well-known structures every year.
With that said, AI does not go as far as interpretation. The tax law does not frown on ambiguities, and regulators want human beings to provide reasons. In this case, moderation is more important than aspiration. AI assists teams in providing cleaner inputs to ensure accountants use their time to defend positions as opposed to building them. The compensation is increasing over time through consistency. The reduction of revisions, amendments, and late notices. Over time, such reliability is more important than measures of speed.
Finance systems hardly have a common language. The time of the payroll is determined by the employee, the revenue of the billing system is determined by the customer, and the tax filing requires activity classification. Plotting of such data sets adds tension to each handoff. Even minor inconsistencies spread downstream as it is. One miscode in the payroll can transform into a wrong credit claim or financial statement.
The most frequent disintegration occurs in timing mismatch. Systems will close books at varying rates, and this compels a delay in reconciliation. The result of that delay to agility is that decisions are based on outdated figures. Precision is compromised since manual bridges can reappear in order to bridge the gaps. The compliance deadlines in any event never shift, and thus, the teams compromise among themselves.
It is reported that almost 60 per cent of the adjustments on taxation are a result of misalignment of the upstream system. Founders are surprised by such a number. The problem of integration still exists due to the optimization of vendors to their own processes, and not regulatory processes.
Tax is the final destination, which takes in-bound noise. Failing to coordinate more closely, finance teams swap speed with safety on a quarterly basis. Such a tradeoff has the disadvantage of reducing responsiveness in the presence of improved tools.
Audit readiness cannot be surpassed by speed in tax. It is only when the outputs can be traced that automation makes any sense. The control exists in documentation, version history and decision logs. In the absence of them, risk increases due to faster processing.
Practically, work-revealing systems are more readily accepted compared to black-box outputs. Auditability is based on the reproducibility of decisions on the same data months later. Such a need influences the decisions of automation. Regulators have a preference toward consistency, rather than novelty, as far as compliance is concerned. Automation succeeds when it rationalizes in and out, but not the thinking.
Indeed, internal standards indicate that filings that are prepared on the basis of documented automation have twenty-two per cent fewer follow-ups. That matters operationally.
Control systems slow down the early implementation, but cushion the confidence in the downstream. With time, work groups understand the safe places of automation. Workflow in the tax world requires time. The speed increases gradually with the building of trust.
Formatted data absorption presented silent yet beneficial proceeds. The use of automation in the intake of receipts, payroll feeds, and project logs drastically eliminated human sorting errors. The change reduced the time spent on rework by approximately a quarter every year.
Consistency brought in the unexpected value. Standardized data had made downstream review and audit preparation easier. It happens that reduced format variations are more important than accelerated processing.
The advantage accumulated in filing periods. The time spent on reconciling and validating positions of teams was reduced. That efficiency came out slowly. No single feature drove it. It was a system which merely made things less frictious. Over time, confidence grew. That trust was converted into less challenging filings and corrections.
