Twelve decisions behind a
TikTok content strategy.

"A list of content ideas with nothing underneath it is just a set of guesses wearing confidence — so I treated a content-ideas task as a strategy task."

On a York University co-op placement at Phoenix Agency — a placed student, not a direct employee of the agency — I was asked to come up with TikTok content ideas for one of its clients, Environmental Factor. The task was narrow: ideas, not briefs, not finished posts. I went wider, because the only way to produce ideas worth keeping is to build the small strategy they are meant to sit inside.

TikTok punishes you for skipping that. It is a top-of-funnel platform — people show up for genuinely helpful, authentic content, not for ads — so the content had to start with what customers do around the products, and it had to reach the 90%-plus still working out what they need. One honest note: the original co-op notes are long gone, so this is deliberately the method as I would run it today, not a transcript of the 2023 work.

"At the awareness stage, difference is the thing you are actually marketing — and you cannot market a difference you cannot explain."

The website told me what Environmental Factor sells: three product lines — pest control, soil amendments, lawn care — across retail and commercial. The About Us page told me what makes it different: biological, neonic-free pesticides. But "neonics" was a word I did not actually understand.

So I ran a quick Google search rather than waving past it. Neonicotinoids turn out to be systemic pesticides that persist and kill beneficial insects and pollinators — a single coated seed can hold enough to kill 250,000 bees. That detour is what turned a label claim into a usable USP: with EF products you are not destroying the ecosystem just to grow your garden.

"Useful content earns the follow, and the follow earns the later permission to sell — that is the whole pathway."

Environmental Factor sells into both retail and commercial sectors, but for a first TikTok test I chose B2C, because that is where the platform's organic reach actually is. The audience pathway is someone browsing gardening content, following creators like Creative Explained and Epic Gardening — people far more likely to follow a brand than to respond to a sales ad.

This was not hypothetical. The Instagram posts I had already produced for EF, and the brand's YouTube content, were both awareness-first — benefits of having plants, how to get rid of raccoons. The TikTok strategy extends a pattern the brand already had rather than inventing one.

"Nobody who has not heard of the brand searches for 'Genus MC8.' They search for healthier soil, or for why their lawn keeps dying."

I went through the full product range — grub, slug and ant control, beneficial nematodes, water retention crystals, diatomaceous earth, Cropsil — and deliberately recorded each item by what it does rather than what it is. Water retention crystals became "reduce how often you water." Genus MC8 became "get healthier soil."

A product name is a dead end for awareness content; a "does" phrase is a candidate topic. The one exception I kept under its own name was pot poppers — a specific, slightly unusual product people might be curious about by name. With a list this long, the real question is which topics people are already searching for.

"TikTok has no real equivalent of SEO software for seeing topic demand — so a serious plan has to build the demand signal from places never designed to be combined."

I set three content clusters for the non-brand-aware searches that matter at the awareness stage: pest control, soil amendments, lawn care — each mapping directly onto what the company sells.

Before collecting any data I built the spreadsheet that would hold it — divided by cluster from the start, with the per-platform traffic columns pre-added. Schema changes mid-project orphan the older rows, so I would say the first job is deciding the shape, not gathering data. The shape is the part you cannot fix cheaply later.

"Trends is the gate, not the verdict — its job is to confirm each cluster has a pulse before I commit time to expansion."

Google Trends is not a TikTok tool, which is exactly why it goes first: free, fast, and able to answer one cheap question per cluster — is there real interest here at all? Pest control, soil amendments and lawn care all showed reasonable, sustained demand.

The recurring judgement across all three was intent-shaped. A lot of strong terms — "pest control near me," "lawn care services" — are people trying to hire a company, not do the thing themselves. Environmental Factor sells products, not a local service, so I promoted only the consumer-intent terms and dropped the rest.

"It would be a mistake to ask the Creative Center to be a demand tool. It is not one — so I used it for exactly the two things it can answer."

TikTok's Creative Center is built to show data on specific ads, not topic demand. So I used it narrowly: to find which of TikTok's own categories the brand belongs in, and to spot trending hashtags. The industry filters surfaced "your home," "cockroach killer" and "cockroach exterminator"; the trends view confirmed #garden, #plants and #patio as live.

This is the discipline that keeps a multi-tool method honest. Trends validates clusters, the Creative Center finds categories and tags, Answerthepublic builds the demand web. No single tool gets asked to do a job it was not built for.

"The mechanical work is what you hand to Claude. The judgement about what the data means is exactly what you do not."

The spider web builds the demand signal TikTok won't give you: a seed term goes into Answerthepublic, every related query with traffic becomes the next seed, and so on. I ran the first term — "pest control," 246K volume — fully by hand to show the method, then handed the expansion to a Claude project so it could scale.

The seam between automation and judgement was a fixed rule sheet: add a term only if it has traffic, no competitors, no zero-traffic terms, nothing tied to a non-Canadian location. The rules shaped the data fast and left every actual decision with me.

"The script did exactly what it was told — sort by traffic, take the top ten — and produced a precise-looking but almost worthless list."

A several-hundred-row dataset is too large to audit whole, so I wrote a PowerShell script to pull a top-ten by traffic. The first run was useless: "garden of the gods" (a landmark), "plants vs zombies" (a video game), "pest control near me" — high traffic, no relevance.

The fix was not a cleverer sort. It was tightening what the script could consider — scoping it to pest control, lawn care and fertilizer. When a ranking looks wrong, the problem is almost always upstream, in the inputs, not in the ranking logic.

"A 418-row spreadsheet does not get used. It gets opened once, found intimidating, and closed."

The manual audit produced 418 post observations across 129 queries, with 314M of total engagement. Rather than leave that in a spreadsheet, I turned it into a small searchable tool — the TikTok Audit Explorer — that anyone can open, search and sort in a couple of clicks.

One column is honestly empty: zero posts carried usable view data, because TikTok's search cards don't reliably expose views, so the audit left it blank rather than guessing. The Explorer also made one thing obvious — demand sits around the products, not on their features, which is ideal for awareness content.

"Just featuring the product is not enough. People want to be inspired about how the product is used in a real task."

Working query by query through the Explorer, the patterns held up across the board. Compost and organic amendments rewarded do-it-yourself method content. Garden soil queries gave a clean opening to show a USP inside a proven before-and-after format. Home pest control, termite and the related-soil queries all rewarded a strong shock hook that names a problem fast.

The repeatable structure is the same each time: educate first, let the product appear naturally inside the task, and show the USP as the close rather than the opening pitch — the way grocery stores feature their own products inside recipes.

"The brief asked for ideas — so the method has to land on some. But every idea traces back to an audited pattern rather than a hunch."

The chapter ends on ten concrete video ideas, from "how to do organic composting" to "your chemical fertiliser is ruining your new garden" — each one drawn directly from a pattern in the audit: method content, shock hooks, surprising facts, USP-at-the-close.

It also says what the work does not prove. It is a present-day reconstruction, not a record of the original 2023 co-op work. The work it covers is topic and format discovery rather than a content calendar or an executed campaign. It is deliberately B2C-first. And the audit itself is a snapshot of a platform that changes fast. The value is a method the next person can re-run and trust — not any single number inside it.