Research cluster: SMB digital-difficulty self-assessment widget (six briefs, June 2026)

Status: Synthesised June 2026. This is the META-master for the six-brief research cluster supporting Candid Creative's SMB digital-difficulty self-assessment widget.

What this widget is

A relative-tier diagnostic that helps an SMB owner understand the difficulty of their digital competitive landscape — and the kind of work that would close their gap — WITHOUT making promises the evidence does not support.

The six briefs and what each contributes

1. Research brief: SMB widget capture layer — what owners can vs cannot self-report (June 2026) — Capture layer. What owners CAN reliably self-report (observation/counting) vs CANNOT (quality self-ratings, relative position). Drives question selection.
2. Research brief: SMB widget spend benchmarks — feasibility of a "digital-minus-ads" % of revenue (June 2026) — Spend benchmarks. Verdict: no defensible "digital-minus-ads" % of revenue exists; widget outputs tiers + dollar ranges, never a % point estimate.
3. Research brief: SMB widget difficulty-to-work mapping — three tiers of work for three sizes of gap (June 2026) — Difficulty-to-work mapping. Three cumulative work tiers (foundational web / findability+trust / interactive tools). Drives the widget's recommendation output.
4. Research brief: SMB widget presentation layer — tiered results without overclaiming (June 2026) — Presentation layer. Tier output with visual scale + earned drivers + efficacy path. Avoids false precision, vague hedging, and lead-gen-grader gimmicks.
5. Research brief: SMB widget market difficulty — six ranked factors (June 2026) — Market difficulty factors. Six ranked factors (incumbent authority, proximity+reviews, SERP features, content saturation, review moats, KD-as-triage). Drives factor weights.
6. Research brief: SMB widget vertical difficulty — two-axis tiering by industry (June 2026) — Vertical difficulty. Two-axis tier model (commercial × local) per vertical. Drives the per-vertical adjustment of the diagnostic.

The widget's design through-line (synthesis across the six)

• INPUTS: observation/counting tasks only (Brief 1) — no self-ratings, no agree/disagree, no abstract competitor comparisons.
• WEIGHTS: factor weights derived from the six ranked market-difficulty factors (Brief 5) and adjusted per vertical (Brief 6).
• OUTPUT: a 4-tier difficulty diagnosis (Brief 4) — earned, visual, anchored, paired with a proximal feasible first step (Brief 4 efficacy pairing).
• RECOMMENDATION: mapped to a work tier (Brief 3) — foundational / findability / differentiation — with honest timelines and base rates.
• DELIBERATELY OMITTED: any precise % of revenue or dollar point estimate (Brief 2) — the data does not support it.

The cross-brief rules (the widget's commandments)

From Brief 1 (capture):

• R1 — Convert every judgment into an observation or counting task.
• R2 — Never use agree/disagree statements.
• R3 — Use concrete NAMED competitors (not abstract comparison).
• R4 — Recent distinctive events, not vague aggregates.
• R5 — Output tiers, not precise scores.
• R6 — In easy-looking markets, override owner optimism with observed counts.
• R7 — Read-off-screen counts beat recalled ones.

From Brief 2 (spend):

• Output tiers, NOT a hard percentage-of-revenue.
• Show dollar ranges as estimates, never point values.

From Brief 3 (work mapping):

• Tiers 1-3 are cumulative + ordered (never recommend higher on failing lower).
• Separate "necessary" (parity) from "differentiating" in output.
• Attach honest timelines + base rates.
• Flag uncontrollable constraints (proximity, time, vertical).

From Brief 4 (presentation):

• 4 tiers max, plain labels, visual scale, drivers shown.
• Replace vague hedging with confident conditional framing.
• Every "hard" tier ships with a feasible proximal first step.
• Frame competitor gaps as closeable and staged.
• Show result BEFORE any email gate.

From Brief 6 (vertical):

• Output two-axis tier (commercial × local), never collapse to one number.
• Show vendor-derived figures as ranges with vendor-source label.
• Relax Canadian vertical tier where adoption is shallow.
• Calibrate thresholds to local reality (metro vs rural ~1.5-2x).
• Allow sub-segmentation in hyper-competitive verticals.

The three pieces of causal evidence the whole cluster rests on

Most of the field is correlational. Three findings rise to genuinely causal:

1. Yelp 1-star → 5-9% revenue (Luca HBS regression-discontinuity) — see Luca HBS WP 12-016 — one-star Yelp = 5-9% revenue (regression-discontinuity, the causal anchor) in Brief 5.
2. Page speed → conversion (Google/SOASTA + web.dev A/B partners) — see Google/SOASTA 2017 — P(bounce) rises 32% at 1s→3s, 123% at 1s→10s in Brief 3.
3. GBP mechanics → local-pack presence (Sterling Sky / Whitespark controlled tests) — see Sterling Sky controlled tests — GBP posts, geotagged photos, description keywords DON'T move ranking in Brief 3.

Plus the antitrust-confirmed existence of NavBoost (NavBoost — Google's 13-month click-based re-ranking, confirmed under oath in DOJ v. Google 2023 in Brief 5).

Everything else is correlational, vendor-claimed, or mechanism-based. The widget's honesty depends on respecting this distinction.

Source-incentive meta-finding across the cluster

Most quantitative SMB digital-marketing data is vendor-originated. SEO/marketing SaaS and agencies have direct financial incentives to inflate the impact of the exact work they sell — and sometimes to inflate difficulty (sells more tools) and sometimes to deflate it (encourages entry). The non-vendor anchors used throughout: Gartner, Duke CMO Survey, US SBA, BDC, Statistics Canada, CFIB, Pew Research Center, Harvard Business School (Luca), Northwestern Spiegel Research Center, US DOJ v. Google trial testimony, Google's own Search Central documentation, peer-reviewed journals.