AI Workflows Intermediate 20 minutes

Research → Competitor Analysis → Marketing Plan

Research your market with cited sources, build a real competitor comparison and SWOT, then turn it into a positioning statement and a 90-day marketing plan — in about 20 minutes with Perplexity, ChatGPT and Notion.

Difficulty
Intermediate
Completion time
20 minutes
Time saved
1–2 days of desk research

Workflow overview

Competitor analysis usually fails in one of two directions. Either it never happens, or it becomes a forty-page document that nobody converts into a decision.

This workflow keeps the research honest and the output short. Perplexity does the gathering because every claim comes with a source you can click; ChatGPT does the structuring, which is where it is strongest; Notion keeps it alive as a document you revisit each quarter.

The rule that makes it work: research and analysis are separate steps. Never let a chat model invent competitor pricing. Gather facts first, then reason over them.

Goal

Understand exactly who you are competing with, where the gap is, and what to do about it over the next 90 days.

Best for: Founders validating a market, small businesses entering a new niche, and students writing strategy coursework.

Required AI tools

Perplexity

Cited, up-to-date research you can verify source by source.

ChatGPT

The reasoning and writing engine at the centre of every workflow.

Notion AI

Turns raw notes into structured pages, tasks and databases.

The workflow at a glance

1
Cited market research
2
Competitor matrix
3
SWOT
4
Positioning statement
5
90-day plan

Step-by-step guide

Work through the steps in order. Every prompt is copy-and-paste ready — swap the bracketed parts for your own details.

1

Map the market with cited research

Perplexity 5 min

Use Perplexity, not a plain chat model, for this step — every statement comes with a source link you can verify. Ask for a specific number of competitors across tiers, because 'who are my competitors' returns only the biggest names, and the biggest names are rarely who you actually lose customers to.

Prompt

I run [BUSINESS] serving [AUDIENCE] in [LOCATION/MARKET]. Identify competitors in three tiers: 1. Three direct competitors offering nearly the same thing 2. Three indirect competitors solving the same problem differently 3. Two substitutes — what customers do today instead of buying anything For each: what they sell, published pricing, positioning line from their own site, target customer, and one recent visible change (launch, pivot, campaign). Cite every claim. If pricing is not public, say so rather than estimating.

What you should get back

Eight competitor profiles with linked sources, honest gaps where information is not public, and the substitutes most analyses forget entirely.

2

Gather the customer language

Perplexity 4 min

Positioning is won with the customer's words, not yours. Ask Perplexity to find what people actually complain about in reviews and forums. Recurring complaints about competitors are the cheapest product roadmap you will ever get.

Prompt

For the competitors above, find what real customers say in public reviews, Reddit threads, forums and app store reviews. Return: - The five most frequent complaints, with a representative quote and source for each - The three things customers consistently praise - The exact words and phrases customers use to describe the problem - Any unmet need mentioned more than once Quote directly. Do not paraphrase into marketing language.

What you should get back

A list of verbatim customer complaints and phrases — the raw material for both your positioning and your ad copy.

3

Build the comparison table and SWOT

ChatGPT 5 min

Now switch to ChatGPT and paste everything you gathered. This step is pure structuring: no new facts, only organisation and interpretation. Ask for the SWOT to be specific and evidence-linked — a SWOT full of words like 'agile' and 'market conditions' is decoration, not analysis.

Prompt

Here is my research on [MARKET]: [PASTE COMPETITOR PROFILES AND CUSTOMER QUOTES] My business: [DESCRIPTION, PRICING, SIZE, CAPABILITIES] Produce: 1. A comparison table: Competitor | Price | Target customer | Key strength | Key weakness | Their positioning 2. A SWOT for MY business. Every entry must be specific and tied to evidence above — no generic words like agile, innovative or market conditions. 3. The three gaps in this market that a small player could credibly own. 4. The one competitor I should worry about most, and why. Use only the information provided. Mark any inference clearly as an assumption.

What you should get back

A clean comparison table, an evidence-linked SWOT, three defensible gaps, and a blunt answer on the real threat.

4

Write the positioning statement

ChatGPT 3 min

One sentence, and it should make some customers walk away. Positioning that appeals to everyone differentiates against nobody. Ask for three options at different levels of aggression, then pressure-test the one you like.

Prompt

Using the gaps identified, write three positioning statements in this format: For [specific customer] who [specific problem], [BUSINESS] is the [category] that [single differentiated promise], unlike [main competitor] which [their limitation]. Make them progressively bolder: safe, focused, and deliberately narrow. Then for each: who this actively repels, what we must be able to prove, and how easily the main competitor could copy it.

What you should get back

Three positioning options with an honest assessment of who each one loses, plus the proof each requires and how defensible it is.

5

Turn it into a 90-day plan in Notion

Notion AI 3 min

Analysis that does not become a calendar is entertainment. Ask for a 90-day plan constrained by your real budget and hours, then paste it into Notion as a database with owners and dates so it behaves like work rather than a document.

Prompt

Based on the chosen positioning, write a 90-day marketing plan. Constraints: budget [AMOUNT]/month, [HOURS] hours per week available, team of [NUMBER]. Structure: - Days 1–30: foundation — what must exist before spending anything - Days 31–60: activation — channels to test, with a specific hypothesis each - Days 61–90: scale or cut — decision criteria stated as numbers For each action: owner, effort in hours, cost, and the metric that tells us it worked. End with the three metrics to review weekly and the two things we will deliberately not do.

What you should get back

A phased plan with costed actions, numeric decision criteria, and an explicit not-doing list — ready to paste into a Notion database.

Copy-and-paste prompt library

Every prompt used in this workflow, in one place. Tap copy and paste straight into your tool of choice.

Prompt 1 — Tiered competitor map

Identify three direct competitors, three indirect competitors and two substitutes for [BUSINESS] in [MARKET]. For each: offering, published pricing, positioning line, target customer, recent visible change. Cite every claim; say so if pricing is not public.

Prompt 2 — Customer language mining

Find what real customers say about these competitors in reviews, Reddit and forums. Return the five most frequent complaints with quotes and sources, three consistent praises, the exact phrases customers use, and any repeated unmet need. Quote directly.

Prompt 3 — Comparison table and SWOT

Build a comparison table (Competitor | Price | Target | Strength | Weakness | Positioning) and a SWOT for my business. Every entry must be specific and evidence-linked — no generic words. Then name three ownable gaps and the competitor I should worry about most.

Prompt 4 — Positioning statements

Write three positioning statements: For [customer] who [problem], [BUSINESS] is the [category] that [promise], unlike [competitor] which [limitation]. Make them progressively bolder, and state who each repels and what proof it needs.

Prompt 5 — 90-day plan

Write a 90-day marketing plan under [BUDGET] and [HOURS]/week: days 1–30 foundation, 31–60 activation with a hypothesis per channel, 61–90 scale-or-cut with numeric criteria. Include owner, hours, cost and success metric per action, plus two things we will not do.

Prompt 6 — Quarterly refresh

Here is last quarter's competitor analysis and what actually happened: [RESULTS]. What has changed, which assumptions were wrong, and what should we stop, start or continue? Be specific about the evidence.

Expected output examples

Comparison table row

| Northline Studio | £2,400/project | Series-A SaaS | Strong brand work | 6-week lead time | "Design that scales with you" |

SWOT weakness (specific)

Weakness: We have no published pricing, and four of the five customer complaints about competitors mention pricing opacity — meaning our silence puts us in the same bucket as the thing buyers dislike most.

Positioning statement

For independent clinics who need a booking site live this month, Meridian is the web studio that ships in ten working days at a fixed fee, unlike agencies that quote six-week timelines and hourly billing.

90-day plan extract

Days 31–60 · Test: Google Search on "clinic website design" · Hypothesis: high-intent local search converts above 4% · Owner: Ana · 6 hrs · £400 · Success: 8 enquiries at under £50 each

Real-world example

Bookkeeping practice entering a new vertical

A three-person bookkeeping practice wanted to specialise in hospitality clients but had no idea who already owned that niche locally.

The tiered research surfaced something they had missed entirely: their real competitor was not another practice but restaurant owners doing it themselves in spreadsheets, plus one national franchise with published pricing 30% below theirs.

Customer language mining found the same complaint repeatedly — accountants who did not understand tips, split shifts and weekly payroll cycles. That single insight became the positioning, and the 90-day plan replaced generic local ads with two hospitality-specific guides and a partnership with a POS reseller. They signed four restaurants in the first quarter at full rate.

If you're a student

This maps almost exactly onto a business strategy assignment. Keep the Perplexity citations — being able to show sourced evidence behind every SWOT entry is what separates a top grade from a generic one.

If you run a business

Re-run the refresh prompt every quarter with your actual results attached. Competitor analysis is only useful as a repeated habit; a one-off document is stale within two months.

Best practices

  • Use a cited research tool for facts and a chat model for reasoning — never mix the two roles.
  • Always include substitutes. 'Doing nothing' is usually the market leader.
  • Ban generic SWOT vocabulary in the prompt; demand evidence links.
  • Choose positioning that repels someone. If nobody is excluded, nothing is differentiated.
  • Constrain the plan by real budget and hours or it will recommend a team you do not have.
  • Spot-check at least five citations manually before acting on the analysis.

Common mistakes

  • Asking a chat model for competitor pricing

    It will produce confident, outdated or invented numbers. Gather pricing from cited sources or record it as unknown.

  • Only listing the market leaders

    You rarely lose deals to the biggest name. The competitors that matter are the ones your customers actually shortlist.

  • A SWOT full of adjectives

    "Agile team" and "changing market conditions" are not analysis. Every entry should reference something specific you found.

  • Stopping at the analysis

    Without the 90-day plan and numeric decision criteria, nothing changes. The plan is the deliverable, not the research.

Variations

Job-hunting version

Swap competitors for target employers: what they build, what their engineers say publicly, and where your skills fill a stated gap. The positioning statement becomes your pitch.

Content gap analysis

Narrow the research to competitor content and search coverage, then build the 90-day plan entirely around topics nobody in your niche has covered well.

Pricing review

Focus the research on published pricing and packaging only, then use the comparison table to test whether your price sits where your positioning claims it does.

Frequently asked questions

Why Perplexity instead of ChatGPT with browsing?

Both can browse, but Perplexity's answer format ties each claim to a visible source by default, which makes verification fast. For strategy work where a wrong number is expensive, that matters more than writing quality.

How accurate is AI-gathered competitor data?

Good for public facts — pricing pages, positioning lines, launches. Weak on anything private like margins or churn. Verify anything you intend to act on and mark inferences as assumptions.

How often should I redo this?

A full pass twice a year, and the lighter refresh prompt quarterly. More often than that and you are researching instead of executing.

Can I use this for a business that does not exist yet?

Yes — it is arguably more valuable pre-launch. The substitutes tier will tell you quickly whether people are already solving the problem well enough for free.

Do I need paid tiers?

The free tiers of Perplexity, ChatGPT and Notion are enough for one full pass. Paid Perplexity mainly helps with deeper multi-source research runs.

Get the next workflow in your inbox

One practical AI workflow every week — no fluff, no hype, just the steps and prompts.

Put this workflow to work today

Compare the tools used here, then go deeper with our written guides.