Creative for software companies, built to be tested, not admired.

More than feels dignified, and less than an agency selling volume will quote. The benchmark that settles the argument is the hit rate: only around 5% of creatives become winners, with a good hook rate near 30% and a good hold rate near 25%. For most software companies spending moderately, that means a handful of genuinely distinct angles each month plus iterations of whatever is working — not forty variations of one layout. We size the cadence against your spend so the account is never starved and never producing work nobody has time to read.

Screenshots are not the problem; unreadable ones are. A product shot that needs pinch-zoom on a phone is asking for effort at the exact moment attention is cheapest, and B2B and SaaS advertisers already average only a 0.6-1.2% link click-through rate on Meta. What works is a cropped detail of the one screen that proves the claim, a cursor doing the job in three seconds, or a recorded moment of the workflow — product truth at a size a thumb can read.

By testing arguments rather than executions, and by changing one variable at a time. Five layouts of the same claim produce noise; five distinct claims produce a decision. We run each angle with enough budget to reach a readable result, keep audience and placement constant while the message moves, and write the hypothesis down before launch so nobody re-interprets the outcome afterwards. Losers are retired without ceremony, winners are iterated in the direction the data suggests rather than redesigned.

Yes, with different economics. LinkedIn buys precision at a much higher cost per click, so the creative can carry more text and a heavier ask; Meta buys cheap reach and rewards a hook in the first second; YouTube sits between them and rewards demonstration. We produce each angle in the formats a placement actually wants rather than resizing one asset five ways, and our LinkedIn ads and YouTube ads teams run the media from the same plan.

Ask what they will measure. An agency that answers with awards, brand lift or impressions is selling craft; one that answers with cost per qualified opportunity and a testing cadence is selling a system. Then ask who does the work and whether you get the source files. The best test in a first meeting is to ask what creative they would kill on your current account and why — an honest answer arrives with reasoning, and an evasive one arrives with a case study.

That is how most of these engagements should end. We build editable templates, type and colour rules, a naming convention and a documented brief format, then run the cadence with your team until the loop is theirs. Plenty of software companies keep us for angle development and heavy motion work while in-house handles iteration, which is usually the cheapest split. Nothing is delivered as a flattened export you cannot edit.

A brand book covers how you look; positioning covers what you claim, and advertising fails on the second far more often than the first. We work inside your existing visual system and spend the effort on the argument: the alternative you replace, the objection that stalls deals, and the proof that answers it for each role in the group. That matters more as committees grow — Forrester now sees 14-member buying groups on genAI-feature purchases against seven without — because one message rarely satisfies both the user and the signer.

Then the creative should stop explaining and start demonstrating. Complicated software sells on a moment of recognition — the ugly spreadsheet, the 2am alert, the four tools duct-taped together — rather than on an architecture diagram. We usually open with the problem in the customer's own words, show ten seconds of the product doing the job, and leave the technical depth to the landing page and the sales call. If the team cannot agree on the one sentence, that is a positioning problem surfacing early, which is a good outcome for a creative project.

Hook and hold read within days, cost per qualified opportunity within four to six weeks, and durable conclusions after two or three cycles of the queue. We resist calling a winner early because early leaders often revert once the audience broadens. What you get in the meantime is a weekly note on what shipped, what is being cut and what the data suggests next — and a monthly review that states plainly whether the work earned its budget.

Meta works for SaaS when the creative does the qualifying that LinkedIn targeting normally does. On Meta you are buying cheap attention from a broad audience, so the persona has to be named inside the first frame — the role, the tool being replaced, the moment of pain — and the offer has to suit somebody at an earlier stage of awareness than a demo request. Meta rewards volume and iteration; LinkedIn rewards precision and patience. Most software companies we work with run both: Meta for reach, testing and sign ups on self-serve products, LinkedIn for the accounts sales actually wants. The same angle queue feeds both, resized and rewritten per platform rather than exported twice.

Creative is where the strategy becomes visible, so the two are best planned together. A demand generation plan decides which buyer personas matter, which stage of awareness the campaigns are addressing, and what the messaging framework claims; creative then turns each of those decisions into something a person will actually watch. When agencies split the work — one running paid media, another making assets — the usual result is beautiful content aimed at the wrong stage. We keep positioning, the content calendar and the ads under one plan so a change in strategy reaches the creative in days rather than in the next quarterly campaign.

Four things, in this order. A messaging framework you can read and recognise as your own product, not a template. A visible testing plan with a cadence, so the best ideas are found rather than guessed. Evidence that the agency understands demand generation economics — what a sign up is worth at your price point, and which stage of awareness the budget is buying. And plain ownership terms: source files, templates and ad accounts in your name. Ask each shortlisted agency what differentiation they would claim for you and why; the best answer usually comes with a caveat attached, which is how you know somebody has thought about the market rather than the pitch.

A fixed monthly fee scoped to the cadence you need, quoted separately from media spend. After a first review you get a plan tied to a testing cadence and a target, and we will tell you where in-house production would be cheaper than ours rather than absorbing the whole scope. Where the useful engagement is a positioning sprint and one production batch instead of a retainer, we scope it that way and say so on the first call. Source files and ad accounts stay in your name throughout.