An outside review of the ZFCI: Here is what held up and what didn't.

In late October I sent a tax specialist something we had never published: the scoring grid behind the Zenbooks Financial Clarity Index. I attached our data workbook, our methodology and the 2026 report, and asked him to find what doesn't hold up.

He found several things.

The ZFCI is in its first year. It surveyed 565 Canadian small business owners and decision-makers in the spring of 2026 and scores each business from 0 to 100 on 25 questions about how it understands and manages its finances. We want it to become a benchmark people cite year after year. That only happens if the work can be checked by people who don't work for us, including people who might not like what they find.

An index produced by a company that sells financial services starts with a credibility problem. Our methodology discloses our commercial interest, and disclosure is necessary. It isn't enough on its own. The better test is to hand over the materials and let someone else publish what they conclude.

What Luc Dubé examined

Luc Dubé, M.Fisc., M.Tax., is a Quebec tax specialist (fiscaliste). His 41-page review, Les PME québécoises : mieux gérées ou mieux encadrées ?, is published in French on his own site under his name alone.

He starts from one of our results. Quebec businesses posted the highest regional score in the 2026 index, 59.8 against 56.7 nationally. His question is whether Quebec owners are better at managing their finances, or whether Quebec's tax and regulatory environment shapes the behaviours we measure. He works through that question using tax law, case law, valuation practice, owner compensation rules, record-keeping obligations and the research on financial literacy in Quebec.

I should be clear about how this came about. I asked Luc to do the review. Zenbooks supplied the report, the methodology, the data workbook and the scoring grid. He had full editorial control over the angle, the content and the conclusions. No compensation or other consideration was exchanged. His review is independent expert commentary. It is not peer review, an audit, or a validation of the index, and he doesn't present it as one.

Four scoring choices he challenged

Luc singles out four of our questions where tax law or valuation practice bears directly on the answer we reward.

Q12: Knowing what your business is worth. We gave 4 points to an owner who is "pretty confident" about the value of the business and 1 point to an owner who answers with a range. Luc points out that the CBV Institute's Practice Standard No. 100 allows a valuation conclusion to be expressed as a range. A range is how professionals state value.

He is right, and I don't think the original choice can be defended. Confidence is not evidence of knowledge. A range was also the most common answer in our sample, given by 44.7% of businesses, so the scoring penalized the answer a careful owner is most likely to give.

Q13: CRA contact. Any CRA review, audit request or compliance notice in the last three years cost points, whatever the outcome. Luc makes two points. First, an audit closed without a reassessment is evidence that the books held up. Second, the question names only the CRA, while in Quebec Revenu Québec administers the QST, the GST and provincial corporate tax.

Our own data adds a third problem. Reported contact rises steeply with size, from 5.9% of businesses with no employees to 43.6% of those with 20 to 99. In practice the question measures exposure to tax administration more than financial clarity. I agree this is a genuine weakness. The options are to move it to an unscored descriptive block, score it on the outcome of the review, or both, and to ask about Revenu Québec explicitly.

In our own practice, the businesses that hear from the CRA most often are frequently the ones that file late or don’t understand an obligation until a notice arrives. The data supports part of that: businesses that delayed payments to vendors or tax authorities were far more likely to report CRA contact, even after accounting for size. But across all businesses, those reporting contact scored no worse on the rest of the index. The question as written can’t tell a notice for a late filing from a routine review of a well-run company. That is the case for scoring the reason and the outcome of the contact, not the contact itself.

Q5 and Q25: Cash reserves. We give the most points to six or more months of cash runway and to holding both an emergency fund and a line of credit. Luc notes that cash sitting in an operating company beyond its reasonable working capital needs is generally not treated as used in the business, which can put the lifetime capital gains exemption at risk when the owner sells.

This is a real planning issue for owners approaching a sale, but it is a different issue from the one the question measures. Six months of runway is a liquidity question: can the business absorb a bad quarter? The exemption problem arises when surplus cash builds up beyond operating needs. Our question can't see where reserves are held, whether the business is incorporated, or whether a sale is planned, and a line of credit carries no purification risk at all. Capping the reward is worth considering. Rewarding a cash cushion isn't itself an error.

A business run purely for tax efficiency is not necessarily a business that survives a bad year. Clarity comes first: knowing your runway and having a cushion. Minimizing tax is something you do within that, not instead of it.

Q6: Delayed owner pay. We give 4 points to owners who did not delay paying themselves in the last 90 days. Only 16.8% of Quebec owners reported a delay, against 27.6% nationally. Luc suggests Quebec's 5,500-hour rule, which ties access to the small business tax rate to paid hours, may push some owners toward a steady salary. He also notes that the incentive is weak for a majority shareholder who keeps a record of hours.

The distinction that matters is between choosing how to pay yourself and delaying pay because the business can't afford it. Our question asks whether the owner delayed pay "in order to manage cash flow," which targets the second case. If you are making payroll for your staff but stop paying yourself because you didn’t see the shortfall coming, that is a planning failure, and the index should register it. Choosing a dividend over a salary is a different decision, and the question doesn’t ask about it. Luc's valuation point supports the finding: skipped pay inflates earnings, and a buyer's valuator will deduct a market salary. A question on how owners are paid would sharpen this, but I don't see a scoring error.

Does this change the 2026 findings?

This is the question that matters, and it has more than one answer, because the ZFCI makes four different kinds of claims.

• Direct survey findings. For example, 27.6% of owners delayed their own pay in the last 90 days. Scoring has no effect on these.
• Composite scores, such as the national score of 56.7. These depend entirely on the scoring choices.
• Associations between practices and scores, such as the gap between integrated and manual systems. These depend on scoring.
• Causal claims or claims about business performance. The 2026 index makes none. It measures reported practices, not profitability or survival.

To test this, we reran the index on the respondent-level data with alternative rules for each challenged question, one change at a time and then in combination, rescaling each affected dimension to its original maximum. Survey weights and regression controls (revenue, employees, sector, region and years in business) were unchanged. The exact scenarios and results are in a technical summary. We will share the code with researchers who want to reproduce it.

The ZFCI uses a non-probability online panel, so we don't report margins of error. Our intervals come from resampling the weighted sample. They show how much an estimate moves with the sample, not whether the panel represents all Canadian businesses. The 2026 index has also not been psychometrically validated or linked to business outcomes, and nothing below shows that the score predicts performance.

"Remains substantial" means the result keeps its direction and its resampled interval excludes zero in every scenario.

The other behavioural findings, on accountant contact, formal ROI analysis, KPIs and DIY bookkeeping, stay large and clearly above zero in every scenario. Business rankings barely move: individual scores correlate at 0.96 or higher with the originals under every scenario.

The confidence gap. 77.2% of owners rate their financial management good or excellent, but only 43.5% score 60 or more. Fixing Q12 narrows the gap from 33.7 to 29.7 points, the largest effect of any single change. Removing Q13 widens it. Under every alternative, a large share of owners still rate themselves well above what their practices support.

Manual vs integrated systems. The gap between businesses with fully integrated systems and those that are mostly manual stays between 26.2 and 30.3 points under every single change, and at 29.1 with all challenged questions removed. None of Luc's scoring concerns weakens it.

Quebec. Quebec has the highest point estimate under every scoring rule we tried. Its lead over Ontario, the next region, is small relative to the uncertainty in a subsample of 79 businesses. Quebec ranked first in 77% of resampled draws. Its controlled advantage holds under each single change. With all challenged questions removed it falls to 4.0 points, and the interval reaches zero. "Highest score in the country" should be read as the highest estimate, not a statistically established ranking.

Luc proposed tests of his own explanations, and we ran the ones the data allows. Removing Q13 barely moves Quebec's advantage. Quebec owners delay pay less often in every size band, not mainly where the 5,500-hour rule would apply. None of this settles why Quebec scores higher.

Where I agree, and where I don't

I agree with Luc on Q12 and Q13. Both are methodological weaknesses, not matters of taste.

On Q5, Q25 and Q6, my disagreement is about what the index is for. Financial clarity, tax compliance, tax optimization, profitability and business value overlap, but they are different things. The ZFCI tries to measure whether an owner understands and manages the business's finances: current books, known margins, a forecast, a cash cushion, numbers reviewed regularly. A well-run business can still carry a tax structure that costs its owner at exit. A tax-efficient structure can sit on top of books nobody reads. Luc's review is strongest where those two things meet, and it shows us where our questions are blind to tax consequences. It doesn't follow that a question about holding a cash cushion should score tax optimization.

That distinction has limits. It doesn't rescue Q12 or Q13, which fail on their own terms.

What we are looking at for 2027

Nothing below has been implemented, and the final decisions aren't made. The second edition fields in March 2027. Between now and then we are examining:

1. Scoring a valuation range at least as highly as stated confidence on Q12.
2. Either moving Q13 out of the score or scoring it on the outcome of the review, and naming Revenu Québec in the Quebec version.
3. Whether to cap the reserve reward, and whether to ask how owners are paid.
4. Recording the survey language, so French and English wording can be compared, which Luc recommends and our current file can't support.
5. Adding unscored questions on profitability, revenue change and financing, so we can test whether the practices we score relate to outcomes.

Changing the scoring also breaks comparability between editions. Our plan is to publish 2027 results under both the old and revised rules and document every change, with 2026 comparable series and updated methodology. We will publish a full account of the outside feedback we've received and the decisions we made before fieldwork begins.

Why this matters

Small and medium-sized businesses employ most private-sector workers in Canada, yet we know little about how they manage their finances day to day. Most of what we know comes from self-reported confidence, which our own data suggests is a poor guide.

The ZFCI will be worth something if it identifies measurable practices, tests whether they relate to outcomes, and gets better each year. Rankings and headline statistics don't do that on their own. Luc's review made the 2026 index easier to understand and will make the 2027 index better. That is the point of inviting criticism. The industry needs this information.