Benchmarks / SurveyBench / Why Tallyroom trials don't convert / Qwen4 Max

Measured by Spring Prompt

Qwen4 Max: Why Tallyroom trials don't convert

The decision: where to put next quarter's product and sales effort to lift trial-to-paid conversion: a new accounting integration, assisted setup, a cheaper plan, or a shorter trial. The brief and the data →

Verdict
✗ Not sound
Research score
94 of 100
Analysis rating
1,373
Head to head, this task
won 9 of 14

Why it is not sound

The analysis

  • Reported a material finding the data does not support

Findings the judges found unsupported

  • Do not give the finance director 61.3% as a conversion forecast. It is hypothetical intent at a stated price from a self-selected group, and only a quarter of them are 'very likely'. Plan on a fraction of it.No evidence supports the proposed 'standard' stated-intent haircut or the resulting estimate of paying teams. Discounting hypothetical intent by arbitrary percentages does not produce a validated conversion forecast. For existing Xero/QuickBooks users, the connection premise is already satisfied, not counterfactual.
  • The price objection is mostly a pricing-model objection, not a price-level objection. A cheaper per-user plan is the wrong instrument; a per-company or active-user tier is the hypothesis worth testing.The evidence supports testing pricing structure, but does not establish that price objections are mostly about structure rather than level, or that a cheaper per-user plan is the wrong instrument. Q2 and Q3 measure different things, and their aggregate percentages cannot establish this distinction.
  • About a fifth of lapsed trials are not recoverable by any of the four options, and for the already-integrated majority the listed fixes miss their blocker entirely. Set a realistic ceiling on what next quarter can deliver.Choosing another tool or reporting no need does not establish that these teams are unrecoverable. Moreover, Q3 omitted assisted setup, a cheaper plan and trial changes, so 'None of these' cannot show that the four decision options miss the blockers of the already-integrated majority or establish a recovery ceiling.
  • FreeAgent is a small but cheap second integration that skews to the core small-business segment — its demand rises under weighting, unlike Sage's. Worth a scoped estimate, not a quarter.Nothing supplied establishes that a FreeAgent integration would be cheap. Its relative build cost and feasibility require technical scoping.

Scorecard

Numbers 11 of 11 right

QuestionIts answerTrue
What percentage of respondents said they would need a Sage integration before paying?✓23.7%23.7%
Weighted to the company-size mix of all lapsed trials, what percentage would need a Sage integration?✓18.1%18.2%
Among businesses with 1-9 employees, what percentage gave 'didn't have time to set it up' as the main reason?✓27.4%27.4%
How many of the open-text answers mention not having time to set Tallyroom up, or setup taking too long?✓2424
What is the 95% margin of error, in percentage points, on the unweighted share of respondents needing a Sage integration?✓3.9 points3.89 points
Is the difference in the share giving 'too expensive' as their main reason between businesses with 1-9 employees and those with 50-249 statistically significant at the 95% level? (1 for yes, 0 for no)✓NoNo
Weighted to the company-size mix of all 1,640 lapsed trials, about how many teams' main reason was not having time to set it up?✓355355
How many of the open-text answers from businesses with 1-9 employees mention not having time to set Tallyroom up, or setup taking too long?✓1414
Weighted to the company-size mix of all lapsed trials, what percentage would give 'no integration with our accounting software' as their main reason?✓16.3%16.3%
What percentage of the trial admins invited completed the survey?✓28%28%
Among respondents whose main reason was no integration with their accounting software, what percentage said they would be likely (fairly or very) to pay if Tallyroom connected to it?✓61.3%61.3%

Traps in the data 7 of 7 handled

  • ✓Teams on 'Other' accounting software (base 17) mostly cite missing integration.
  • ✓Teams of 50-249 people are 22% of respondents but 10% of lapsed trials. They cite missing integration and use Sage far more, so unweighted figures overstate integration and Sage demand and understate setup time: unweighted, missing integration is the top reason; weighted, setup time is.
  • ✓Q2 is multi-select: percentages are of respondents and add up to more than 100%.
  • ✓Likelihood to pay is a hypothetical stated intention.
  • ✓The head of sales believes the trial is too long and teams forget about it.
  • ✓460 of 1,640 invited admins completed (about 28%), with a £10 voucher; those who respond may differ from those who do not.
  • ✓Only teams that did not pay were surveyed, so the survey cannot show what distinguishes them from teams that did; paying teams may also have struggled with setup.

The analysis

Recommendation

Put next quarter's effort into assisted setup first, a scoped Sage integration second, and a pricing-model test third — and do not shorten the trial. (1) Trial length: answer the head of sales with a clear no. 'Too long' is one respondent out of 460 (0.2%); 'forgot' is 17 (3.7%); meanwhile 86 (18.7%, ~21.6% weighted, ~355 teams) never finished setup, and 11 open answers say the clock ran out mid-setup. Cutting to 14 days would shorten the runway for the 55% of the base who are 1-9 employees, where setup failure is highest (27.4%). Fix the forgetting problem with lifecycle nudges and setup-progress-triggered extensions, not a shorter clock. (2) Product: ship assisted setup for the small-business core — pre-filled chart of accounts and staff import from Xero/QuickBooks, a default single-approver flow that works out of the box, and a booked 30-minute done-with-you session for trials that have not reached first approved expense by day 3. This addresses the largest weighted barrier (~355 teams), is the cheapest of the four options, and needs no new integration. (3) Sage: answer the head of product with 'demand is intense but narrow' — 85.5% of 117 Sage users need it, 3.3% of Xero and 0.9% of QuickBooks users do; weighted need is 18.1%, not the headline 23.7%, and Sage is 68 of the 93 integration-blocked respondents. Commit to a scoped build (start with Sage Business Cloud/Accounting, which most open answers name) but gate the start on two numbers we do not yet have: Sage's share of lapsed and new trials, and paid-pilot commitment from Sage design partners. Answer the finance director honestly: 61.3% stated likelihood among the integration-blocked is not a conversion rate — only 25.8% are 'very likely', a quarter remain unlikely even with the connector, and some have a second blocker (price, approvals). Illustratively, if Sage is ~20-25% of the 1,640 lapsed trials (~330-410 teams) and a disciplined 15-25% of the integration-blocked convert, that is roughly 50-85 paying teams; assisted setup at a 25-35% recovery on ~355 setup-blocked teams is roughly 90-125. Both are worth doing; setup is faster and cheaper, so it goes first. (4) Pricing: do not launch a blanket cheaper plan. The objection is per-seat shape, not level — 20.4% need per-company pricing (~21.0% weighted) against 15.9% citing expense, with ten open answers asking for a flat fee or objecting to paying for occasional claimants. Test a flat per-company tier or active-user pricing and measure conversion and ARPA together. (5) Flag to the decision-makers that the option list is incomplete: custom approval chains are needed by 22.0% (~20.5% weighted, more than Sage) and are the main reason for 12.6% overall and 21.8% of larger firms, with 15 open answers specifying exactly what is missing (batch approval, dual/threshold sign-off, external accountant approver, email approval, category routing). Cost this against Sage before locking the quarter; on current evidence it is at least as large and it serves the already-integrated majority where 51.6% of Xero users said 'none of these' would help. (6) Sales effort: stop spending acquisition effort on Sage-accounting businesses until the connector ships — instead run a design-partner waitlist and, where you must sell, be explicit about the manual export step. Concentrate outbound and trial-conversion calls on Xero/QuickBooks small trials with hands-on setup, and work the accountant/bookkeeper channel: seven open answers show the bookkeeper both vetoes the tool and chooses the replacement.

What would change it

Four pieces of evidence would flip the ordering. (a) Sage's true share of lapsed and new trials: if the database shows Sage at ~35-40%+ of trials or of inbound pipeline (rather than the sub-25% we suspect after correcting the size skew), Sage integration overtakes assisted setup, because its addressable pool would then exceed the setup pool and its intent is far stronger (61.3% vs 27.9% net likely). (b) Trial telemetry on setup: if most trials actually complete setup and lapse after activating, the assisted-setup case collapses and the weight shifts to approvals, pricing and integration — the survey's 'didn't have time' answers would then be post-hoc rationalisation. (c) A live randomised trial-length test: if 14 days matches or beats 30 days on trial-to-paid with faster time-to-first-value, the head of sales is right and our reading of the stated reasons is wrong; we would adopt it. (d) Behavioural pricing evidence: if a flat per-company tier lifts conversion materially without diluting ARPA beyond the agreed tolerance, pricing becomes the primary lever rather than a parallel test. Additionally, win/loss interviews or lost-deal coding showing approvals or a named competitor blocking more revenue than Sage would reprioritise the roadmap; and paid pilots or pre-orders from Sage-using lapsed trials would replace the intent discount with real evidence and could justify starting Sage immediately.

Findings

  1. Reject the 14-day trial. Trial length is essentially absent as a reason for lapsing, and shortening it would directly worsen the largest fixable barrier — unfinished setup — for the smallest businesses, which are 55% of the lapsed-trial population.

    High confidenceBase 460

    Q2 (base 460): 'trial too long' 0.2% (1 respondent), 'trial too short' 1.5% (7), 'forgot the trial was running' 3.7% (17) — 5.4% combined. Against that, 'didn't have time to set it up' 18.7% (86), rising to 27.4% of 1-9 employees (46 of 168). 24 of 109 open answers cite setup time, and 11 of those explicitly say the trial ended before setup finished (T022, T073, T131, T146, T153, T263, T331, T340, T371, T431, T460). Not one open answer asks for a shorter trial.

  2. The raw tables overstate Sage demand and understate setup. Weight to the real size mix before any sizing: Sage need falls from 23.7% to 18.1%, setup as main reason rises from 18.7% to 21.6%, and integration as main reason falls from 20.2% to 16.3%.

    High confidenceBase 460

    Respondents are 36.5% / 39.6% / 23.9% by size (1-9 / 10-49 / 50-249) versus 55% / 35% / 10% in the 1,640-trial population. Weighted Q3 Sage = 0.55x7.7 + 0.35x28.0 + 0.10x40.9 = 18.1%. Weighted Q2 setup = 0.55x27.4 + 0.35x15.9 + 0.10x10.0 = 21.6% (~355 teams). Weighted Q2 no-integration = 16.3% (~267 teams).

  3. Sage demand is a segment requirement, not market-wide demand. It is near-universal among Sage users and near-zero elsewhere, so the business case rests entirely on how many trials and how much pipeline are Sage-based — a number this survey cannot weight.

    High confidenceBase 117

    Q3 Sage integration: 85.5% of Sage users (100 of 117), 6.1% of FreeAgent users (2 of 33), 3.3% of Xero (6 of 182), 0.9% of QuickBooks (1 of 111). Sage accounts for 68 of the 93 respondents whose main reason was no integration (73%); FreeAgent 11, other software 14. Q4 Sage share is 25.4% unweighted, but no Q4-by-size table exists, so the population Sage share is unknown and is probably lower than 25.4% given the over-sampling of larger firms.

  4. Do not give the finance director 61.3% as a conversion forecast. It is hypothetical intent at a stated price from a self-selected group, and only a quarter of them are 'very likely'. Plan on a fraction of it.

    Medium confidenceBase 93

    Q5 among the integration-blocked (base 93): net likely 61.3%, but 'very likely' only 25.8% (24 respondents), 'fairly likely' 35.5% (33), and 25.8% still unlikely even with the integration, 12.9% neutral. Q5 was shown to everyone including the 63.7% already on Xero/QuickBooks, for whom 'if Tallyroom connected to your software' is counterfactual — so the 37.6% total is not a forecast of anything. Applying a standard stated-intent haircut (roughly 20-35% of 'very likely' plus a small share of 'fairly likely') suggests low tens of paying teams per 100 integration-blocked trials, not 61.

  5. The price objection is mostly a pricing-model objection, not a price-level objection. A cheaper per-user plan is the wrong instrument; a per-company or active-user tier is the hypothesis worth testing.

    Medium confidenceBase 460

    Q2 'too expensive' 15.9% (73 of 460), but Q3 'per-company pricing (not per user)' 20.4% (94 of 460), ~21.0% weighted, and remarkably flat across sizes (22.6% / 18.7% / 20.0%). Ten open answers object specifically to paying per seat for occasional claimants or ask for a flat fee: T023, T059, T166, T208, T307, T322, T343 ('We'd have paid a reasonable flat rate'), T355 (seasonal headcount), T402 ('A flat fee would've been fine'), T449.

  6. Approval-workflow rigidity is the biggest barrier that is not on the decision list, and on weighted demand it is comparable to Sage. It should be scoped and costed alongside the four options rather than ignored.

    High confidenceBase 460

    Q2 'approval workflow didn't fit' 12.6% (58 of 460), 21.8% of 50-249 (24 of 110). Q3 'custom approval chains' 22.0% (101 of 460), ~20.5% weighted versus 18.1% for Sage — and unlike Sage it is spread across every software segment (Xero 20.3%, QuickBooks 21.6%, Sage 28.2%). 15 open answers name specific missing patterns: batch/period approval (T161, T216, T300), dual or threshold approval (T392), external accountant approver (T247), email-based sign-off (T077), category-based routing (T180), single MD with no middle layer (T010), ordered two-step (T455).

  7. About a fifth of lapsed trials are not recoverable by any of the four options, and for the already-integrated majority the listed fixes miss their blocker entirely. Set a realistic ceiling on what next quarter can deliver.

    Medium confidenceBase 460

    Q2 'decided we didn't need it' 10.0% (46) plus 'chose another tool' 8.7% (40) = 18.7% (86 of 460); weighted no-need ~11.9%. Q3 'None of these' 36.5% overall (168) and 51.6% of Xero users (94 of 182), 42.3% of QuickBooks (47 of 111). 12 open answers describe genuinely low expense volume (T015, T080, T083, T087, T143, T148, T150, T163, T164, T204, T211, T214) and 8 name a competitor already adopted (T030 Dext, T271 Rydoo, T330 Soldo, T341 Expensify, T346 Webexpenses, T367 Pleo, T384 Emburse, T446 Pleo).

  8. FreeAgent is a small but cheap second integration that skews to the core small-business segment — its demand rises under weighting, unlike Sage's. Worth a scoped estimate, not a quarter.

    Low confidenceBase 33

    Q3 FreeAgent 7.2% overall (33 of 460) but 13.7% of 1-9 (23 of 168); weighted 0.55x13.7 + 0.35x3.8 + 0.10x2.7 = 9.1%. 72.7% of FreeAgent users need it, but that is only 24 of 33 respondents (95% MoE about +/-15pp). Three open answers (T174, T205, T314); T205 explicitly says the app itself was 'lovely to use'. A further 4 answers name long-tail packages (T097 KashFlow, T130 Clear Books, T269, T458) — 14 of the 93 integration-blocked are outside Sage/FreeAgent, hinting that a generic export/API path has value.

  9. Blockers co-occur, so single-fix conversion models will over-promise. Q2 forces one 'main reason' and hides this; the open text shows it plainly.

    Medium confidenceBase 109

    T322 (per-seat pricing AND no Sage link), T334 (Sage link AND price), T159 (setup AND receipt capture), T329 (setup AND uploads), T400 (setup AND Android capture). 5 of 109 open answers name two independent blockers; Q3 multi-select averages well over one need per respondent outside the 'none' group (e.g. Sage column sums to 155.6%).

  10. Price sensitivity is not statistically different by company size — do not build a size-specific pricing narrative on these tables.

    High confidenceBase 278

    Q2 'too expensive': 19.0% of 1-9 (32 of 168) vs 15.5% of 50-249 (17 of 110). Difference 3.5pp, 95% CI on the difference approximately +/-9.0pp (SE 4.6pp). Not significant. The apparent gradient in the table is noise.

  11. Receipt capture and mobile reliability are a real, size-skewed product defect that none of the four options addresses; it should be on the fix list regardless of the strategic choice.

    Medium confidenceBase 460

    Q2 'problems with the mobile app' 6.3% (29 of 460), 7.7% of 1-9 and 8.2% of 10-49 versus 0.9% of 50-249 (1 respondent). Q3 'better receipt scanning' 15.0% (69), ~17.1% weighted, 20.2% of 1-9. 10 open answers give concrete failure modes: cropped totals (T106), OCR misreads (T195, T378: a GBP48.60 ticket read as GBP4.60), 4G upload failure (T125, T309), freeze on submit (T317), Android capture (T400).

  12. The survey's precision is fine; its representativeness is the weak point, and the size variable itself is partly mismeasured. Treat all weighted figures as indicative, not as forecasts.

    Medium confidenceBase 460

    460 of 1,640 completed = 28.0%, incentivised with a GBP10 voucher. Sampling MoE on the headline Sage figure is +/-3.9pp (n=460), but non-response bias is unmeasured and plausibly larger. Around 11 of 109 open answers state a headcount that contradicts their size band (T010 'only nine people' coded 50-249; T087 'team of five' coded 50-249; T164 'four-person' coded 50-249; T059 '25 people', T322 '25 staff', T343 '30 staff' all coded 50-249; T023 '15 employees', T208 'team of 12', T307 'ten-person', T449 '18 of us' coded 1-9; T166 '8 staff' coded 10-49). Weighting on a noisy variable transfers some error into every weighted number.

Caveats

  • 28.0% completion (460 of 1,640) with a GBP10 voucher. Respondents are self-selected and likely more engaged or more aggrieved than the 1,180 who did not reply; direction of the resulting bias is unknown. Sampling error (+/-3.9pp on the Sage figure, +/-4.6pp worst case at n=460) is smaller than the plausible non-response bias, so do not treat the margins of error as the true uncertainty.
  • Weighting corrects company size only. It cannot fix non-response on software choice, engagement or reason for lapse. Weighted figures (Sage 18.1%, setup 21.6%, integration 16.3%) are indicative planning numbers, not population estimates.
  • The size variable is partly unreliable: roughly 11 of 109 open answers state a headcount that contradicts their Q6 band (e.g. T010, T059, T087, T164, T322, T343 coded 50-249 but describe 4-30 staff; T023, T208, T307, T449 coded 1-9 but describe 10-18 staff). This contaminates both the segment columns and the weights.
  • No Q4-by-size cross-tab was supplied, so the Sage share of all 1,640 lapsed trials cannot be weighted. The 25.4% unweighted Sage share is almost certainly an over-estimate because larger firms (where Sage need is highest) are over-represented. The entire Sage business case hinges on this missing number.
  • Q5 measures hypothetical intent at a stated price, not behaviour, and was shown to respondents already on Xero/QuickBooks for whom the premise is counterfactual. The 37.6% total net-likely is uninterpretable as a conversion rate; even the 61.3% among the integration-blocked should be heavily discounted (only 25.8% 'very likely').
  • Q3 is a stated requirement list, not demand. 85.5% of Sage users saying they need Sage is close to tautological given Q2; the informative quantity is how many Sage-using trials exist, which the survey cannot supply.
  • 'Sage' is not one integration. Open answers name Sage Business Cloud (T334, T339), Sage Accounting (T433), Sage 50 (T413) and Sage 200 (T241) — different products, different APIs, different build costs. A single 'Sage integration' line item will mis-scope the work.
  • Small bases must not be read as precise: FreeAgent n=33 (+/-15pp on its 72.7%), Other software n=17 (+/-24pp), 50-249 mobile-app figure 0.9% = 1 respondent, 'trial too long' 0.2% = 1 respondent. The near-zero trial-length figures are strong evidence of absence only because the open text corroborates them.
  • Q2 forces a single main reason while several respondents clearly had two or more blockers (T322, T334, T159, T329, T400). Fixing one blocker will not convert those teams, so per-fix conversion estimates are upper bounds.
  • Screener S1 is self-reported ('trial ended in the last 90 days without paying'), with 'Not sure' excluded; some lapses may be misclassified, and admins may not speak for the finance decision-maker who actually vetoed the purchase (T023, T115, T332).
  • Three open answers refer to a 'fortnight' or 'trial week' (T106, T216, T446) despite a 30-day trial — either recall error or variation in trial length in the fielded base. Check trial configuration in the database before concluding anything about trial duration.
  • Q4 and Q6 both return exactly 0.0% for one option ('no accounting software', '250+'). Plausible, but verify these were genuine responses rather than routing or data-processing artefacts.
  • This is a lapsed-trial survey only. It contains no evidence about teams that did convert, no behavioural data on what happened inside the trial, and no win/loss against the eight named competitors. It can rank stated barriers; it cannot establish causes of conversion.

Next steps

  • decisionKeep the 30-day trial; replace the 'shorten to 14 days' proposal with setup-triggered extensions (auto-extend 14 days for trials that have not approved a first expense) plus day-3/day-10/day-25 lifecycle nudges naming the unfinished setup step.
  • researchPull the database and product telemetry cross-tabs the survey cannot give: accounting software x company size x trial-to-paid conversion for all 1,640 lapsed trials and recent cohorts; Sage share of trials and of new inbound; setup-completion rate, time-to-first-approved-expense, and where in the funnel trials stall.
  • experimentRandomised assisted-setup pilot on new small-business trials: auto-import staff and expense categories from Xero/QuickBooks, a working single-approver default, and a offered 30-minute done-with-you session for trials with no approved expense by day 3.
  • researchRun 12-15 win/loss interviews with Sage-using lapsed trials (recruit from T006, T021, T055, T109, T127, T169, T226, T241, T339, T413, T419, T433, T459) plus 5 accountants/bookkeepers, and offer a paid design-partner pilot; confirm which Sage product each uses and test GBP6 per user against a flat per-company quote.
  • researchScope and cost custom approval chains against the Sage connector before locking the quarter: batch/period approval, ordered multi-step, dual approval above a threshold, external (accountant) approver, approve-by-email, and category-based routing.
  • experimentTest pricing shape, not price level: randomise new trials into current per-user pricing versus a flat per-company tier (and an active-user variant), with ARPA and expansion tracked alongside conversion.
  • decisionGate the Sage build on the telemetry and design-partner evidence, then ship the highest-volume Sage variant first with a co-marketed accountant/bookkeeper channel; meanwhile give sales an explicit Sage-account playbook (waitlist plus honest manual-export workaround).
  • monitorInstrument and fix receipt capture and mobile upload reliability (crop/total detection, OCR accuracy on UK receipts, 4G upload retry, Android submit freeze), prioritising the 1-9 and 10-49 segments.
  • monitorReplace the incentivised email survey as the ongoing tracker with a one-question in-product exit capture at trial expiry (main reason, plus optional free text), reported monthly by size and accounting software.

Open-text themes it coded

No time to complete setup / setup took too long 24Trial expired before setup was finished (trial clock vs setup latency) 11Missing Sage integration 15Missing FreeAgent integration 3Other / long-tail accounting software unsupported 4Approval workflow does not match how the business signs off expenses 15Per-user / per-seat pricing model is the wrong shape (occasional claimants, want flat fee) 10Price too high / could not justify the subscription cost 21Receipt capture / OCR / mobile app unreliable 10No genuine need — too few expenses to justify any tool 12Chose a competitor instead 8Accountant / bookkeeper is the gatekeeper for the decision 7

The survey it planned

4 screening questions and 10 questions, as the model wrote them.

  1. S1

    Which of these best describes your involvement with Tallyroom at your business?

    One answer
    • I was the main person who set up and managed Tallyroom (the admin)
    • I shared the admin role with a colleague
    • I used Tallyroom but was not an admin
    • I had no involvement with Tallyroom
    • I don't know / can't remember

    Continues if I was the main person who set up and managed Tallyroom (the admin); I shared the admin role with a colleague

  2. S2

    As far as you know, what happened with your team's Tallyroom free trial?

    One answer
    • The trial ran to its end and we did not start paying
    • We stopped using it / ended the trial early and did not start paying
    • The trial ended and we started paying for a subscription
    • The trial is still running
    • I don't know / can't remember

    Continues if The trial ran to its end and we did not start paying; We stopped using it / ended the trial early and did not start paying

  3. S3

    Is your business based in the UK?

    One answer
    • Yes
    • No
    • Don't know

    Continues if Yes

  4. S4

    Including yourself, how many employees does your business have in total? Your best estimate is fine.

    One answer
    • 1–9
    • 10–49
    • 50–249
    • 250 or more
    • Don't know

    Continues if 1–9; 10–49; 50–249

  5. Q1

    In your own words, what was the main reason your team did not start paying for Tallyroom? Please be as specific as you can — a sentence or two is plenty.

    Open text
  6. Q2

    Which of the following, if any, played a part in your team not starting to pay for Tallyroom? Select all that apply.

    Any that apply
    • We never finished setting it up
    • Setting it up took more time or effort than we expected
    • It did not connect to the accounting software we use
    • It was missing a feature we needed
    • Staff did not start using it
    • The approval process did not fit how we work
    • The price (£6 per user per month) was too high for us
    • We could not get internal budget approval
    • We chose a different product
    • We lost momentum and stopped using it before the trial ended
    • The trial ended before we had finished evaluating it
    • We were only trying it out, with no plan to buy
    • Concerns about data security or privacy
    • Something else (please specify)
    • None of these
    • Don't know / can't remember
  7. Q3

    Which ONE of those was the biggest reason your team did not start paying? If you only selected one, please select it again.

    One answer
    • We never finished setting it up
    • Setting it up took more time or effort than we expected
    • It did not connect to the accounting software we use
    • It was missing a feature we needed
    • Staff did not start using it
    • The approval process did not fit how we work
    • The price (£6 per user per month) was too high for us
    • We could not get internal budget approval
    • We chose a different product
    • We lost momentum and stopped using it before the trial ended
    • The trial ended before we had finished evaluating it
    • We were only trying it out, with no plan to buy
    • Concerns about data security or privacy
    • Something else

    Routing Shown only if Q2 is: We never finished setting it up; Setting it up took more time or effort than we expected; It did not connect to the accounting software we use; It was missing a feature we needed; Staff did not start using it; The approval process did not fit how we work; The price (£6 per user per month) was too high for us; We could not get internal budget approval; We chose a different product; We lost momentum and stopped using it before the trial ended; The trial ended before we had finished evaluating it; We were only trying it out, with no plan to buy; Concerns about data security or privacy; Something else (please specify)

  8. Q4

    Which accounting software does your business mainly use for its books?

    One answer
    • Xero
    • QuickBooks Online
    • QuickBooks Desktop
    • Sage (e.g. Sage Business Cloud, Sage 50)
    • FreeAgent
    • Clear Books
    • Zoho Books
    • Microsoft Dynamics / Business Central
    • NetSuite
    • Another ERP or industry-specific system
    • None — spreadsheets, or our accountant handles it
    • Other (please specify)
    • Don't know
  9. Q5

    Which of these, if any, did your team need to do but could not do in Tallyroom during the trial? Select all that apply.

    Any that apply
    • Connect to the accounting software we use
    • Handle VAT on receipts correctly for our returns
    • Claim mileage
    • Handle foreign currency receipts
    • Match company card or bank transactions to receipts
    • Reimburse staff through payroll
    • Use multi-step or rule-based approvals
    • Set spend limits or budgets by team
    • Get receipts in from email or web upload, not just photos
    • Sign in with our company single sign-on (SSO)
    • Nothing — it could do everything we needed
    • Other (please specify)
    • Don't know / can't remember
  10. Q6

    When your free trial ended, what was the furthest point your team had reached with Tallyroom?

    One answer
    • We never logged in after signing up
    • We logged in, but no receipts were uploaded
    • Only a few test receipts were uploaded
    • A few staff were using it for real expenses
    • Most or all staff were using it for real expenses
    • Don't know / can't remember
  11. Q7

    Imagine the free trial had been 14 days instead of 30, with everything else about the product and price the same. How would that have affected your team's decision about paying?

    One answer
    • We would have been much more likely to pay
    • We would have been slightly more likely to pay
    • It would have made no difference
    • We would have been slightly less likely to pay
    • We would have been much less likely to pay
    • Don't know
  12. Q8

    Thinking about the price of £6 per user per month, which statement best fits your team's view at the time?

    One answer
    • It was more than we would pay, and a lower price would not have changed our decision
    • It was more than we would pay, but a lower price could have worked for us
    • The price was acceptable — something else stopped us
    • The price was good value for what it did
    • Don't know / can't remember
  13. Q9

    What is the highest price per user per month your team would have been willing to pay for Tallyroom as it was during your trial? Please give your best estimate in pounds, as a number (enter 0 if you would not have paid anything).

    Number

    Routing Shown only if Q8 is: It was more than we would pay, but a lower price could have worked for us

  14. Q10

    If Tallyroom had offered ONE of the following during your trial, which would have made your team most likely to start paying? Please pick the single most persuasive one, even if none of them is perfect.

    One answer
    • A connection to the accounting software we use
    • A booked session with a Tallyroom specialist who sets everything up with us
    • A lower-priced plan
    • A 14-day trial instead of a 30-day trial
    • None of these — we would not have started paying
    • Don't know
Sample plan

FRAME: one invitation per team (the admin contact held in the product database) for every UK team whose 30-day Tallyroom trial ended in the last 90 days without a paid subscription. At ~1,900 trial starts per quarter and 18% converting, the steady-state 90-day pool is roughly 1,550 lapsed teams. Before fielding, de-duplicate to one admin per team, verify trial end date and non-payment against billing records, and drop internal/test accounts, teams still inside a trial, teams that later paid, and anyone surveyed in the last 6 months. S1–S4 corroborate the database facts (role, lapsed status, UK, size band); self-reports that contradict the database are excluded at cleaning, not silently kept. Note that S2 deliberately allows teams that quit early, so the 'never activated' segment is not screened out. FIELD: census invitation of the whole frame by email, £10 voucher on completion, neutral subject line and copy that mentions neither trial length nor price (to avoid priming the head of sales' hypothesis), two reminders (day 3, day 8), 12-day field, one response per team, GDPR notice covering the appended usage data. Expected completes at a realistic 15–25% for lapsed-trial B2B admins: ~230–390. 450 is therefore the optimistic ceiling for a 90-day window, not a planning assumption. PRECISION CONFLICT (must be resolved with Product before fielding): ±6pp at 95% confidence within a band requires n≈267 per band (p=0.5), i.e. ~800 completes — about double the brief's 450 and roughly 50% of the entire 90-day frame. Options, in order of recommendation: (a) keep 450 completes with equal quotas of 150 per band and report ±8.0pp per band (±4.6pp for the total); (b) if ±6pp per band is non-negotiable, extend the frame to trials ended in the last 180 days (~3,100 teams), census it, and target ~800 completes (267+ per band), accepting older-cohort recall decay and adding trial-end month as a control; (c) report 1–9 and 10–49 separately at ±6pp and pool 50–249 into a '10+' read. Feasibility warning: in a small-business base the 50–249 band is likely only ~5–10% of the frame (~80–160 teams in 90 days), so even 150 completes in-band is probably unreachable — census that band, extend its window to 180–270 days, and add telephone/assisted follow-up; if it still lands under n=100, publish it with its actual (wider) interval or merge it, and never label it ±6pp. QUOTAS: equal allocation targets of 150 per band (hard floor 100, stretch 267 under option (b)); soft interlock on trial-end recency (0–30 / 31–60 / 61–90 days) so recall effects are testable; monitor accounting-software mix and aim for at least 150 non-Xero/non-QuickBooks respondents for O2, without a hard quota that would distort the size bands. WEIGHTING: bands are deliberately over-sampled relative to the frame, so weight total-sample estimates back to frame size-band proportions by post-stratification; always show unweighted band bases and per-band confidence intervals. APPENDED DATA (no question cost): attach database/telemetry fields per team — seats invited, receipts uploaded, approval actions, whether Xero/QuickBooks was connected, first and last active day, count of active days — so O4 rests on behaviour, not recall. NON-RESPONSE: compare respondents with the full frame on those fields and on band and recency; expect teams that never activated to be under-represented, and report the direction and likely size of that bias rather than assuming it away. DATA QUALITY (in place of a slot-consuming attention item, given the £10 incentive and the 10-question cap): drop completions under ~100 seconds, gibberish or off-topic Q1 text, and internally contradictory patterns (e.g. Q6 'never logged in' plus feature-level complaints in Q5); replace dropped completes to hold quotas. FIELD SETTINGS: randomise option order in Q2, Q3, Q5 and the first four options of Q10, anchoring 'Something else', 'None of these' and 'Don't know' last and making 'None of these' / 'Nothing — it could do everything we needed' exclusive; rotate which end of Q7 is shown first across respondents to control acquiescence. Estimated respondent time: ~4 minutes (screener ~30s, Q1 ~35s, Q2 ~35s, Q3 ~15s, Q4 ~10s, Q5 ~25s, Q6 ~15s, Q7 ~20s, Q8 ~10s, Q9 ~15s for the ~40% who see it, Q10 ~20s).