Programmatic advertising solves a problem most small businesses don't have. Here's when the complexity is worth it, when it isn't, and the three buys that work without a DSP at all.
Programmatic advertising solves a problem most small businesses don't have. Here's when the complexity is worth it, when it isn't, and the three buys that work without a DSP at all.
Programmatic advertising is the most over-prescribed solution in paid media. Agencies pitch it as "scale, efficiency, targeting." For businesses under £20M revenue running ecommerce, programmatic usually adds complexity without proportional results — three simpler buys cover most of what teams actually need.
This piece is when programmatic earns its place, when it doesn't, and what to do instead.
Programmatic = automated buying of ad inventory through real-time auctions, run via a demand-side platform (DSP) like DV360, The Trade Desk, or StackAdapt. The "automation" buys across thousands of publishers, exchanges, and ad formats simultaneously based on a bid model trained on your audience signals.
It earns its complexity when:
If none of those four are true, you're paying programmatic infrastructure cost (DSP fees, ad-tech overhead, viewability/brand-safety stack) for buying that could be done more simply.
Four structural problems below the £20M revenue mark:
If a customer is searching for your category, you should be on the page. Search ads capture explicit intent. Performance Max extends across YouTube, Display, and Shopping within Google's owned inventory — better-controlled than open programmatic, still automated bid optimisation, and 75%+ working media.
For ecommerce under £20M, Meta is usually the largest paid channel and produces the cleanest signal. Advantage+ Shopping Campaigns (ASC) automate the bid + audience + placement decisions within Meta's owned inventory. Working media percentage is high. See the Facebook ads management piece for the operational depth.
If there are 3–8 publications your target customer reads, a direct deal (sponsored content, newsletter sponsorship, native takeover) beats programmatic display by 5–10× on engagement. The trade-off is operational overhead — you negotiate, brief, and traffic each placement manually. Worth it under £20M because the publishers reachable directly are the ones that matter most.
| Buy type | Best for | Working media | When to start |
|---|---|---|---|
| Google Ads (Search + PMax) | Demand capture, brand defense. | ~80% | £3k/mo minimum. |
| Meta (Facebook + Instagram) | Demand creation, retargeting. | ~80% | £5k/mo minimum. |
| Publisher-direct | Niche audience reach, brand association. | ~85% | £10k/mo minimum, per placement. |
| Programmatic (DSP) | £100k+/mo spend with first-party data. | ~40% | £100k+ monthly minimum. |
Three specific situations where the complexity is justified:
Outside those three, programmatic for a sub-£20M business is paying for plumbing that doesn't add value.
No. Display is the format (banner ads). Programmatic is the buying method. You can buy display directly from a publisher (no DSP involved) or programmatically through a DSP. Most display today is bought programmatically but they're separate concepts.
~£100k/month is the bar most agencies set for full open-exchange DSP campaigns. Below that the platform fees + minimum service hours dominate the cost. CTV-only DSPs go lower — £15k–£30k/month can work for connected TV alone.
For very large ecommerce (£100M+) with rich first-party data and broad customer journeys, yes. For sub-£20M DTC, Meta + Google + email cover 80% of the workload more efficiently. The complexity-to-result ratio doesn't flip until you're spending more on paid than Meta + Google can absorb.
DSP (demand-side platform) buys ad inventory on the advertiser's behalf — DV360, The Trade Desk, StackAdapt, MediaMath. SSP (supply-side platform) sells inventory on the publisher's behalf — Google Ad Manager, Magnite, PubMatic. The two transact via real-time bidding through ad exchanges. As an advertiser you interact with the DSP.
Run a geo-holdout test. Pause programmatic in one region for 30 days, keep it running elsewhere. Compare incremental revenue in the test vs control. If the lift in the holdout-paused region is less than 5% of programmatic spend, the campaign is largely buying credit for conversions that would have happened anyway.
Qwrki runs paid media as part of one operating layer — the team that runs your Meta + Google also writes the cohort attribution that tells you whether each pound is working. For most clients under £20M we recommend skipping programmatic entirely and reallocating that budget to better creative + better lifecycle. Book a call — first hour we audit the existing paid stack honestly.