Research & Analysis Checklist for Agency & Consulting

Write a single decision question, the decision owner, the time horizon, and the metric that will change if your work is correct. This prevents scope creep and aligns the research with measurable impact.

Classify the client by ICP, ACV band, and vertical so you can reuse the right template set and benchmarks. This keeps research depth consistent across similar accounts and improves margin.

List 5-10 hypotheses tied to specific metrics like CAC, LTV, or organic share and define disconfirming evidence. This forces crisp methodology and makes insight reviews faster.

Prepare 8-10 pointed questions for product, sales, and customer success to capture internal realities and constraints. Schedule 30-minute sessions upfront to avoid mid-stream pivots.

Document prohibited data types, geos with consent requirements, and robots.txt respect for scraping. Reference GDPR, CCPA, and platform TOS to reduce legal risk.

Agree on a 1-page executive brief, a feature matrix, a TAM model, and an artifact archive with weekly check-ins. Upfront clarity minimizes rewrites and accelerates approval.

List free and paid sources such as SEMrush, Ahrefs, Similarweb, Google Trends, Crunchbase, Statista, and G2 with owner, credentials, and export methods. This reduces hunt time and enables consistent repeat runs.

Secure read-only exports from GA4, Search Console, Google Ads, Meta Ads, Shopify or Stripe, and CRM like HubSpot or Salesforce. Consolidating first-party and public signals is essential for defensible insights.

Use Screaming Frog for IA and metadata, and BuiltWith or Wappalyzer for technology detection across competitors. This reveals product maturity, CMS, analytics usage, and experimentation footprint.

Configure Brandwatch or Meltwater for brand and category keywords, with BuzzSumo or native platform APIs as fallback. Keep a query library and track changes to maintain consistent panels.

Document field names, units, time granularity, and primary keys for joining across sources. A clear schema reduces modeling errors and speeds team onboarding.

Run null rate checks, outlier detection, and join coverage tests in SQL or pandas, and consider Great Expectations or Deequ. Set stop conditions when data fails to meet accuracy thresholds.

Use a structured folder or warehouse schema for raw, processed, and curated layers, with read-only sharing for clients. Consistent naming and permissions prevent version drift.

Classify core, adjacent, and analog competitors using Similarweb traffic, funding or revenue proxies from Crunchbase, and client references. Tiering aligns research depth with business threat.

Scrape docs and pricing pages and log features in Airtable or Notion, then verify with product screenshots or trials. A clean matrix prevents subjective debates during client readouts.

Capture plan names, usage caps, add-ons, and discount heuristics from quotes and public pages. Normalize to price-per-unit comparisons to reveal effective positioning gaps.

Use Google Ads auction insights, Meta Ads Library, SpyFu, and SEMrush share data to infer paid mix, with organic share from SEMrush or Ahrefs. This informs probable CAC and competitive aggression.

Classify homepages, landing pages, and ads into jobs-to-be-done and pain themes using a coding framework in Dovetail or Notion. This crystallizes where your client can differentiate.

Build a bottom-up model from query volumes, CPCs, conversion rates, and ACVs, then triage with Statista or Gartner estimates. State assumptions and provide a confidence range.

Inventory marketplaces, integration directories, and partner programs from platforms your ICP uses. Identify VAR or white-label openings to shortcut acquisition channels.

Sample reviews from G2 and Capterra, Reddit threads, and support tickets to extract pains and desired outcomes. Tag sentiment and importance to prioritize messages and features.

Export keywords from SEMrush or Ahrefs, then cluster using regex rules or k-means by modifiers like compare, best, near me, and map to TOFU, MOFU, BOFU. This guides content and landing page strategy.

Crawl competitor blogs and resources with Screaming Frog and map to formats like calculators, templates, and case studies. Identify high-intent formats your client lacks.

Pull HubSpot or Salesforce call logs and notes, then code recurring objections and deal-breakers. Feed top themes into messaging tests and sales enablement assets.

Use Meta Ad Library, TikTok Creative Center, and LinkedIn ads to capture creative patterns, CTAs, and offers. Pair with landing page performance where visible to infer conversion hypotheses.

Compare SERPs, CPCs, and messaging norms across key regions and account for regulation like cookie consent. Avoid porting a single-market insight globally without validation.

Build curated tables in BigQuery or Snowflake with dbt models and snapshot dates for reproducibility. This ensures every chart can be regenerated with the same inputs later.

Calculate MoM, YoY, moving averages, and seasonality adjustments before forecasting. Clean baselines reduce noise and make trends trustworthy.

Model CAC, LTV, conversion rate, and ASP sensitivities with tornado charts to show which assumptions move the outcome most. Share low, base, and high cases for stakeholder buy-in.

Combine weighted indices for SEO, paid search, social engagement, and PR mentions into a unified score. Weightings should be justified by channel contribution to pipeline for the ICP.

Check model outputs against third-party panels like Similarweb panel estimates or YouGov surveys. Record confidence intervals and document where estimates diverge.

List coverage gaps, survivorship bias, and noisy proxies with mitigation steps like alternative sources or wider error bars. Surfacing caveats increases client trust.

Lead with TLDR, three insights, and three actions tied to revenue or efficiency. Keep it skimmable to secure stakeholder attention before deep dives.

Use Looker Studio, Tableau, or Power BI with clear filters and source notes. Document refresh frequency and owner so clients know when to expect updates.

Provide Jupyter or Observable notebooks with parameter blocks and environment files, plus a static PDF. This enables peer review and handoff to in-house teams.

Facilitate a 60-minute session in Miro to score actions using RICE or ICE and assign owners. Convert top items into backlog tickets during the meeting.

Generate Jira, Asana, or ClickUp tickets with acceptance criteria, KPIs, and data dependencies for each recommended action. Frictionless handoff accelerates implementation.

Archive raw data, processed datasets, code, screenshots, and consent artifacts with a readme and index file. This reduces audit time and enables future refreshes.

For any third-party tool or data feed, record SOC 2 or ISO 27001 status, DPAs, data residency, and TOS. Keep a matrix of acceptable uses to protect clients from compliance drift.