The problem: the old playbook stopped working in 2021
Paid social advertising used to reward audience engineering. You built lookalikes, stacked interests, split ad sets by a dozen segments and let the pixel do the rest. Apple’s App Tracking Transparency, browser cookie restrictions and platform consolidation of targeting options changed that. Interest targeting is coarser, attribution windows are shorter, and the platforms now do most of the audience work themselves.
What remains under your control is creative, offer, landing experience and the quality of the conversion data you send back. Accounts that still spend their effort on audience segmentation, and report on in-platform ROAS without questioning it, tend to plateau and then quietly decline. The work now is a creative production and testing system, plus a measurement setup you can trust.
What paid social advertising covers
Paid social is any advertising bought inside social platforms: feed, stories, reels and shorts placements, in-app shops, lead forms and messaging ads. We run it on the platforms that fit your buyer rather than on all of them by default.
Meta (Facebook and Instagram)
The default for direct-to-consumer e-commerce, apps, local services and many B2B awareness plays. Advantage+ shopping and sales campaigns, catalogue ads, lead ads and Reels placements, with the Conversions API as a baseline requirement.
Expensive per click but unmatched for firmographic targeting. Right for SaaS, technology and professional services with a high deal value, using document ads, thought-leader ads, lead-gen forms and matched-audience retargeting.
TikTok
For products with a visual demonstration and a younger or trend-driven audience. Native-looking creator content matters more than polish; we plan production accordingly and use Spark Ads where creators are involved.
Useful for home, fashion, food, wedding, travel and gift categories where users plan purchases weeks ahead. Shopping and collection ads tied to your catalogue, with longer consideration windows.
Search and Shopping campaigns are managed separately under PPC management, and we frequently run the two together so that social demand generation is measured through the search lift it produces.
Who this service is for
- Direct-to-consumer brands with a proven product whose Meta account has plateaued and whose creative pipeline is the bottleneck.
- B2B and SaaS companies that need LinkedIn to generate qualified pipeline, not just whitepaper downloads that never convert.
- Businesses that have lost confidence in their reported ROAS since privacy changes and want measurement rebuilt properly.
- Companies launching a new product or market who need demand created rather than captured, working alongside lead generation or e-commerce goals.
What Axoria does
How we approach creative, audiences and structure
Creative is the primary lever
On Meta and TikTok in particular, the platform’s delivery system decides who sees an ad largely by who engages with it. Creative therefore determines the audience. We build concepts around a specific customer problem, objection or use case, not around brand adjectives. For a skincare brand, that might mean one ad addressing “does it work on sensitive skin”, another comparing it to a dermatologist visit, and a third built from a customer review, each tested as a hook rather than a finished campaign.
A testing cadence you can sustain
We plan creative in cycles: a fixed number of new concepts and iterations per fortnight or month, sized to your production capacity. Each cycle tests new angles against the current control, iterates on winners (new hook, same body; same hook, new format), and retires ads showing frequency and CTR fatigue. Iteration usually outperforms invention, so most of the cadence is spent building on what already works.
Broad targeting, strong exclusions
With enough conversion signal, broad or Advantage+ audiences generally outperform hand-built interest stacks, and they avoid audience overlap that inflates costs. The exceptions are LinkedIn, where firmographic targeting is the point, and low-volume accounts that need a narrower start. In every case we exclude existing customers from prospecting, cap retargeting frequency and use first-party lists as seeds rather than as walls.
Full-funnel, with honest budget splits
Retargeting always looks efficient because it reaches people who were already going to buy. We size retargeting to the actual pool of eligible users and put the majority of budget into prospecting, then judge the account on blended cost per acquisition and new-customer growth rather than on the retargeting campaign’s ROAS in isolation.
Why the Conversions API is not optional
Browser-side pixels now miss a meaningful share of conversions because of consent choices, ad blockers and Safari’s tracking limits. Sending events server-side through the Conversions API, deduplicated against the pixel and enriched with hashed customer data, restores much of that signal. Better signal means better delivery and more trustworthy reporting. We treat CAPI, event match quality and consent handling as the first job in any Meta account.
Our paid social process
Audit and customer research
Account, tracking and creative review, plus interviews, review mining and support-ticket analysis to identify the angles worth testing.
Measurement setup
Conversions API, event deduplication, consent handling, UTM conventions and GA4 configuration so that platform and blended numbers can be compared.
Structure and launch
Prospecting, retargeting and retention campaigns built with clear budgets and exclusions. Initial creative set launched in a testing campaign with defined success criteria.
Creative cycles
New concepts and iterations on a fixed cadence, promoted into scaling campaigns by rule rather than by feel. Fatigued ads retired before they drag performance.
Scale and prove lift
Budget increased in steps, geo or holdout tests run when spend justifies it, and quarterly reviews of platform mix and creative direction.
Tools we typically work with
Meta Ads Manager and Events Manager, LinkedIn Campaign Manager, TikTok Ads Manager and Pinterest Ads. Google Tag Manager, including server-side containers, for Conversions API delivery; Shopify and other native integrations where they are the cleaner route. GA4 and Looker Studio for blended reporting. Motion or similar for creative analytics where volume justifies it, Figma and Canva for static production, and creator platforms for UGC sourcing. Northbeam, Triple Whale or a similar tool where a client already uses one, with the understanding that these are models, not ground truth.
How results are measured after privacy changes
Platform-reported conversions are modelled, use platform-specific attribution windows and cannot be added across channels without double counting. We report them, but never alone.
| Layer | What we look at | What it tells you |
|---|---|---|
| Platform metrics | CPM, CTR, cost per result, ROAS by campaign and creative, frequency, event match quality | Which creative and structure decisions are working inside the platform |
| Blended metrics | Total spend against total new customers or qualified leads, blended CAC, marketing efficiency ratio | Whether the whole programme is paying for itself |
| Incrementality | Geo holdouts, conversion-lift studies, brand search volume changes | How many sales the ads caused rather than merely touched |
| Downstream quality | CRM lead status, repeat purchase rate, LTV by acquisition channel | Whether the customers acquired are the ones you want |
We set attribution windows deliberately (7-day click, 1-day view is our usual Meta default, with view-through reported separately) and keep them consistent so trends are comparable. The tracking foundation is built through our analytics and tracking service, and the broader measurement logic is explained in our performance marketing guide.
What to expect
Creative production is a real cost
Paid social without a creative pipeline stalls within weeks. We can brief your team, work with our production partners or source creators, but the budget for production needs to exist alongside media spend.
Platform ROAS will look different from blended ROAS
Once tracking is honest and retargeting is right-sized, in-platform numbers often look less impressive while business results improve. We explain the difference before it shows up in a report.
Learning phases are real
New campaigns and large budget changes trigger platform learning. We scale in steps of roughly 20% and avoid frequent structural edits during learning.
No guaranteed outcomes
Results depend on product, offer, creative quality and market conditions. We commit to a testing system, transparent measurement and clear recommendations, not to a ROAS number.